From: bayko@borealis.cs.uregina.ca (John Bayko) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 12 Jan 1998 18:30:23 GMT Organization: University of Regina, Dept. of Computer Science Lines: 60 Message-ID: <69dnfv$re4$1@sue.cc.uregina.ca> References: <348fa205.3834263@news.compuserve.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> NNTP-Posting-Host: borealis.cs.uregina.ca Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.sprintlink.net!news.sprintlink.net!Sprint!newsfeed.internetmci.com!206.172.150.11!news1.bellglobal.com!dragon.sk.sympatico.ca!mongol.sasknet.sk.ca!news.uregina.ca!not-for-mail In article <69d5qt$dqb@examiner.concentric.net>, Steve Quest wrote: [...] > I was working for Atari at the time out in sunnyvale, and the >story we heard was that IBM chose Intel because Intel was in chapter 11 >bankruptcy and would do anything to save themselves, EVEN LICENSING THEIR >PRODUCT TO BIG BLUE for a bargain price! I think "it was cheaper" is the >real story, otherwise IBM was going to go with their first choice, Zilog! >Zilog could have right now been the largest CPU maker. Come to think of >it, I think IBM got to use Intel product royalty-free for either one or >two years as IBM experimented with the "personal computer" idea.......sq Unlikely. The Zilog Z-8000 was too late. The 68000 was available, but didn't have support chips available at the time. However there was a licensing agreement apparently involved... Here's what the much ridiculed but persistant Great Microprocessors List has to say - a note to fans, it has just been upgraded to version 10.0.0, now including the Hitachi 6301, MIL-STD-1750, and microJava CPUs, plus a revamped RISC vs. CISC section. It's available from my home page, or other sites such as the CPU Info Center at: http://infopad.eecs.berkeley.edu/CIC/ Great Microprocessors of the Past and Present (V 10.0.0) Section Three: The Great Dark Cloud Falls: IBM's Choice. Part VII: Intel 8086, IBM's choice (1978) [...] So why did IBM chose the 8086 series for the IBM 5150 PC (1981) when most of the alternatives were so much better? Apparently IBM's own engineers wanted to use the 68000, and it was used later in the forgotten IBM Instruments 9000 Laboratory Computer, but IBM already had rights to manufacture the 8086, in exchange for giving Intel the rights to its bubble memory designs. Apparently IBM was using 8086s in the IBM Displaywriter word processor. Other factors were the 8-bit 8088 (1979) version, which could use existing low cost 8085-type components, and allowed the computer to be based on a modified 8085 design. 68000 components were not widely available, though it could use 6800 components to an extent. After the failure and expense of the IBM 5100 (1974/5/6? - their first attempt at a peronal computer - discrete random logic CPU with no bus, built in BASIC and APL as the OS), cost was a large factor in the design of the PC. The availability of CP/M-86 is also likely a factor, since CP/M was the operating system standard for the computer industry at the time. However Digital Research founder Gary Kildall was unhappy with the legal demands of IBM, so Microsoft, a programming language company, was hired instead to provide the operating system (initially known at varying times as QDOS, SCP-DOS, and finally 86-DOS, it was purchased by Microsoft from Seattle Computer Products and renamed MS-DOS). Digital Research did eventually produce CP/M 68K for the 68000 series, making the operating system choice less relevant than other factors. Intel bubble memory was on the market for a while, but faded away as better and cheaper memory technologies arrived. -- John Bayko (Tau). bayko@cs.uregina.ca http://www.cs.uregina.ca/~bayko ###### From: 100612.1245@compuserve.com (Mike Davies) Subject: Re: Why IBM chose Intel Date: Tue, 13 Jan 1998 12:13:57 GMT Message-ID: <34bb5a83.1875937@news.compuserve.com> References: <348fa205.3834263@news.compuserve.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> X-Newsreader: Forte Free Agent 1.1/32.230 Newsgroups: comp.arch.embedded,alt.folklore.computers Lines: 10 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!204.186.0.13.MISMATCH!ptdnetP!newsgate.ptd.net!newsfeed.sover.net!arl-news-svc-7.compuserve.com!news-ntawwabp.compuserve.com >9000 Laboratory Computer, but IBM already had rights to manufacture the >8086, in exchange for giving Intel the rights to its bubble memory designs. So IBM didn't always lose at the negotiating table ? ;-) Mike ###### From: albaugh@agames.com (Mike Albaugh) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded,alt.folklore.computers Date: 13 Jan 1998 21:01:42 GMT Organization: Atari Games Corporation Lines: 14 Message-ID: <69gknm$2tf$1@void.agames.com> References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> NNTP-Posting-Host: java.agames.com X-Newsreader: TIN [version 1.2 PL2] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.he.net!Supernews60!supernews.com!newshub1.home.com!news.home.com!newshub1-work.home.com!void.agames.com!albaugh John Bayko (bayko@borealis.cs.uregina.ca) wrote: : A lot of people say 'personal' computer means designed to be used : by a single person, and point to the LINC or PDP-8 as the first : example. I'd nominate the Bendix G-15, designed (mainly?) by Harry Huskey. Tubes (valves) and semiconductor diodes. Better MTBF (or, at least MTB-ctrl-alt-del :-) than the average Wintel machine today... Some of the earliest graphics programming I ever saw, too (on a plotter, of course, and one did have to be patient :-) Mike | albaugh@agames.com, speaking only for myself ###### Path: ccw.ch!usenet From: Neil.Franklin.nospam@ccw.ch (remove .nospam) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 13 Jan 1998 23:59:59 +0100 Organization: My own Private Self Lines: 36 Message-ID: References: <348fa205.3834263@news.compuserve.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> X-Newsreader: Gnus v5.3/Emacs 19.34 Steve Quest wrote: > I was working for Atari at the time out in sunnyvale, and the >story we heard was that IBM chose Intel because Intel was in chapter 11 >bankruptcy and would do anything to save themselves, EVEN LICENSING THEIR >PRODUCT TO BIG BLUE for a bargain price! bayko@borealis.cs.uregina.ca (John Bayko) answered: >Apparently IBM's own engineers wanted >to use the 68000, and it was used later in the forgotten IBM Instruments >9000 Laboratory Computer, but IBM already had rights to manufacture the >8086, in exchange for giving Intel the rights to its bubble memory designs. >Apparently IBM was using 8086s in the IBM Displaywriter word processor. > >Other factors were the 8-bit 8088 (1979) version, which could use existing >low cost 8085-type components, and allowed the computer to be based on a >modified 8085 design. According to Byte magazine, September 1990, The Creation of the IBM PC, by David J Bradley (one of its designers), page 416 [shortened by me]: 1. The 64K-Byte address limit had to be overcome 2. The processor and its peripherals had to be available immediately 3. We couldn't afford a long learning period, had to use technology we were familiar with 4. There had to be both an OS and applications available for the processor We narrowed our decision down to the Intel. The Boca Raton engineers were familiar with these processors. A bonus we got from chosing Intel was the numeric coprocessor. Bradley himself came from the System/23 (Datamaster), 8085+Basic based. -- Neil.Franklin.nospam@ccw.ch (remove .nospam), http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: ivie@cc.usu.edu (Roger Ivie) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Message-ID: <2+yiEGJmmRZE@cc.usu.edu> Date: 14 Jan 98 01:01:44 MDT References: <348fa205.3834263@news.compuserve.com> <69h99r$fk0@nnrp1.farm.idt.net> Organization: Utah State University Lines: 12 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!cs.utexas.edu!news.cs.utah.edu!cc.usu.edu!ivie In article <69h99r$fk0@nnrp1.farm.idt.net>, Eric Chomko writes: > But CP/M was first consideration, correct? Why didn't it become what > MS-DOS DID become? But it did! CP/M-86 -> Concurrent CP/M-86, where it picked up DOS compatability and became Concurrent DOS -> DR DOS -> Novell DOS -> Caldera OpenDOS. CP/M-86 became an MS-DOS clone. -- -------------------------+------------------------------------------------ Roger Ivie | Impeach Bob Palmer! ivie@cc.usu.edu | http://cc.usu.edu/~ivie/ | ###### From: Eric Chomko Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 14 Jan 1998 02:52:43 GMT Organization: IDT Lines: 39 Message-ID: <69h99r$fk0@nnrp1.farm.idt.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve NNTP-Posting-Host: u2.farm.idt.net X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!atl-news-feed1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.gip.net!news.gsl.net!gip.net!news.idt.net!nntp.farm.idt.net!u2.farm.idt.net!not-for-mail Organization: IDT Internet Services Distribution: In alt.folklore.computers John Bayko wrote: : In article <69fve3$25q@nnrp1.farm.idt.net>, [...] : Not meant as a home computer, but neither was the IBM PC - the : only computer for home IBM made back then was the PCjr, which was : exactly the type of thing you'd get from a business computer company : making a machine for the home. Yet many people bought the PC. And the clone market definitely made the PC-system more attractive. : A lot of people say 'personal' computer means designed to be used : by a single person, and point to the LINC or PDP-8 as the first : example. Yet, what one individual could afford to buy these systems or their following generations? The PC world (not my favorite, BTW) is still alive and is dominated by individual buyers. : > : >The irony here is that IBM got the same "legal" issue response from Gates : >as they got from Kildall, but rather than go to a third party, they : >argreed with Gates' terms. It was rumored that they actually tried to go : >back to Kildall and offer him the deal that they finally gave Gates, but : >he (Kildall) was out flying his plane. : I don't believe this is the case. Microsoft agreed to IBMs terms : in delivering the OS for the IBM PC - PC-DOS - but they did negotiatd : a technical loophole that allowed them to produce and sell MS-DOS on : its own terms, which was probably the difference. But CP/M was first consideration, correct? Why didn't it become what MS-DOS DID become? Eric ###### From: "David Thompson" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 14 Jan 1998 09:48:00 -0800 Organization: WRQ Inc. Seattle, WA Lines: 18 Message-ID: <69itol$agj$1@wrqnews.wrq.com> References: <348fa205.3834263@news.compuserve.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> NNTP-Posting-Host: 150.215.90.112 X-Newsreader: Microsoft Outlook Express 4.72.2106.4 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.2106.4 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.ecn.uoknor.edu!solace!xinit!fci-se!fci!nntprelay.mathworks.com!newsfeed.internetmci.com!192.220.251.22!netnews.nwnet.net!wrq.com!not-for-mail I remember reading that Intel created the 8088 from the 8086 specifically for IBM so IBM could lower the cost of the first PCs. The 8088 had an eight bit external bus versus 16 for the 8086. I am not sure if I have the terminology correct, but Intel's willingness to cut its chip off at the knees may have been a factor. >>PRODUCT TO BIG BLUE for a bargain price! I think "it was cheaper" is the >>real story, otherwise IBM was going to go with their first choice, Zilog! >>Zilog could have right now been the largest CPU maker. Come to think of >>it, I think IBM got to use Intel product royalty-free for either one or >>two years as IBM experimented with the "personal computer" idea.......sq > > Unlikely. The Zilog Z-8000 was too late. The 68000 was available, >but didn't have support chips available at the time. However there was >a licensing agreement apparently involved... ###### From: "Bill B." Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 14 Jan 1998 11:04:13 -0800 Organization: Trying, but not quite Lines: 28 Message-ID: <34BD0C2D.2781@US.ORACLE.COM> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69h9 <69iojn$1nd@nnrp1.farm.idt.net> NNTP-Posting-Host: upsizeme.us.oracle.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01 (X11; I; AIX 2) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.onenet.net!news.oru.edu!logbridge.uoregon.edu!su-news-hub1.bbnplanet.com!su-news-feed1.bbnplanet.com!news.bbnplanet.com!inet16.us.oracle.com!not-for-mail Eric Chomko wrote: > > Bill B. wrote: > : Eric Chomko wrote: > : > > : > But CP/M was first consideration, correct? Why didn't it become what > : > MS-DOS DID become? > > : Because, according to the PBS show "Triumph of the Nerds," Kildall > : refused to allow the IBM representatives to enter his house in > : Pacific Grove, Cal. (On the Western tip of Monterey penninsula) > > Believe me I know exactly where Pacific Grove, Pebble Beach, Carmel and > Monterey are located. My dad lives in Salinas. I was out there over > Thanksgiving. Hey Eric, great to hear your geography is up to snuff... However, I suspect that most people would have no clue as to the whereabouts of P.G. (Unless they're one of mega-masses to have visited the Acquarium!) > Anyway, is it true that Gary Kildall fell off a barstool in that same area > and died? That is the story I heard from the locals while hiking with some > of them in the Grand Canyon about 5 years ago. The Kildall incident would > have been in 90 or 91. Is any of this true? It sounds just too bizarre. Don't know about the barstool incident, but I do know that Gary has passed on..... ###### From: Eric Chomko Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 14 Jan 1998 15:46:24 GMT Organization: IDT Internet Services Lines: 15 Message-ID: <69imkg$1nd@nnrp1.farm.idt.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69h9 NNTP-Posting-Host: u3.farm.idt.net X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.ecn.uoknor.edu!solace!xinit!fci-se!fci!nntprelay.mathworks.com!news.idt.net!nntp.farm.idt.net!u3.farm.idt.net!not-for-mail In alt.folklore.computers Roger Ivie wrote: : In article <69h99r$fk0@nnrp1.farm.idt.net>, Eric Chomko writes: : > But CP/M was first consideration, correct? Why didn't it become what : > MS-DOS DID become? : But it did! CP/M-86 -> Concurrent CP/M-86, where it picked up DOS compatability : and became Concurrent DOS -> DR DOS -> Novell DOS -> Caldera OpenDOS. CP/M-86 : became an MS-DOS clone. Somehow I think you missed my point.... Where is Kidall's gazillions? I heard that he's no longer with us in fact. Sad tale, if it's to be believed. Eric ###### From: Eric Chomko Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 14 Jan 1998 16:20:07 GMT Organization: IDT Internet Services Lines: 20 Message-ID: <69iojn$1nd@nnrp1.farm.idt.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69h9 NNTP-Posting-Host: u3.farm.idt.net X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!news-peer.gip.net!news.gsl.net!gip.net!news.idt.net!nntp.farm.idt.net!u3.farm.idt.net!not-for-mail Bill B. wrote: : Eric Chomko wrote: : > : > But CP/M was first consideration, correct? Why didn't it become what : > MS-DOS DID become? : Because, according to the PBS show "Triumph of the Nerds," Kildall : refused to allow the IBM representatives to enter his house in : Pacific Grove, Cal. (On the Western tip of Monterey penninsula) Believe me I know exactly where Pacific Grove, Pebble Beach, Carmel and Monterey are located. My dad lives in Salinas. I was out there over Thanksgiving. Anyway, is it true that Gary Kildall fell off a barstool in that same area and died? That is the story I heard from the locals while hiking with some of them in the Grand Canyon about 5 years ago. The Kildall incident would have been in 90 or 91. Is any of this true? It sounds just too bizarre. Eric ###### Newsgroups: comp.arch.embedded,alt.folklore.computers From: mzenier@netcom.com (Mark Zenier) Subject: Re: Why IBM chose Intel Message-ID: Organization: Netcom On-Line Services References: <348fa205.3834263@news.compuserve.com> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve <69h99r$fk0@nnrp1.farm.idt.net> Date: Wed, 14 Jan 1998 19:18:18 GMT Lines: 28 Sender: mzenier@netcom6.netcom.com Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!streamer1.cleveland.iagnet.net!qual.net!iagnet.net!newsfeed.internetmci.com!194.162.162.196!newsfeed.nacamar.de!ix.netcom.com!mzenier In article <69h99r$fk0@nnrp1.farm.idt.net>, Eric Chomko wrote: >: I don't believe this is the case. Microsoft agreed to IBMs terms >: in delivering the OS for the IBM PC - PC-DOS - but they did negotiatd >: a technical loophole that allowed them to produce and sell MS-DOS on >: its own terms, which was probably the difference. > >But CP/M was first consideration, correct? Why didn't it become what >MS-DOS DID become? Because Digital Research decided to set the price for the IBM PC version at the same as all the other computers that it was used in. By the time the machines really got out there to the public, both were available. The first time I saw a IBM in a store, there on the software shelf where THREE different operating systems that you could buy for it. The UCSD (or whatever the commercial firm with the rights to it) P-System, CPM-86, and PC-DOS. PC-DOS was in the $50 range and CPM-86 was about $200. Functionality was about the same. Guess which sold... As for the P-System, it was interpeted pseudocode. Since the IBM-PC was two or three times slower that the typical high end small system of the time, it wouldn't have been slow, it would have been glacial. (Typical high end machines of the period would have been 8 MHz 8086 S-100 systems running CPM-86). The $300-$500 price didn't help. Mark Zenier mzenier@eskimo.com mzenier@netcom.com ###### From: peb@transcontech.co.uk ("Paul E. Bennett") Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 14 Jan 98 20:35:46 GMT Organization: Transport Control Technology Ltd. Message-ID: <884810146snz@transcontech.co.uk> References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> Reply-To: peb@transcontech.co.uk X-Mail2News-User: peb@transcontech.co.uk X-Mail2News-Path: tcontec.demon.co.uk X-Trace: mail2news.demon.co.uk 884891078 23856 peb transcontech.co.uk X-Complaints-To: abuse@demon.net X-Newsreader: Demon Internet Simple News v1.30 Lines: 21 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!vixen.cso.uiuc.edu!ais.net!news.idt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!transcontech.co.uk!peb In article <01bd208d$848db940$1bc4accf@mfc> mjcoyle@erols.com "Michael F. Coyle" writes: > BTW, the Motorola 6809 was a major contender for the PC CPU. Supposedly, Bill > Gates convinced IBM that a 16-bitter was the way to go. The 8088 was a > singularly poor choice, though. I would have rated the 6809 as the better choice, then being a Forthist I would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 has 16 bit ALU operations and a nicer feel to it. I know that the data bus was 8 bits but that is not always the performance bottleneck. Since BG persuaded IBM to go that way we have all had to live with a crazy architecture. -- Paul E. Bennett ................... Transport Control Technology Ltd. +44 (0)117-9499861 Going Forth Safely ###### From: "David Hunt" Subject: Re: Why IBM chose Intel Newsgroups: comp.arch.embedded,alt.folklore.computers References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve <69h99r$fk0@nnrp1.farm.idt.net> Message-ID: <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net> X-Newsreader: Microsoft Internet News 4.70.1161 Lines: 66 Date: Wed, 14 Jan 1998 21:31:49 GMT NNTP-Posting-Host: usr18-dialup7.mix1.willowsprings.mci.net NNTP-Posting-Date: Wed, 14 Jan 1998 16:31:49 EST Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.ecn.uoknor.edu!solace!xinit!fci-se!fci!news.idt.net!newsfeed.internetmci.com!204.71.1.51!spamkiller.internetmci.com!news.internetMCI.com!not-for-mail I must take issue with the statement that the Original IBM PC was not intended for home use. There was a game port and that game port was definitely for home use. The game port was designed into the original IBM PC motherboard. The PC was designed to be a direct competitor in both the home and educational markets with the Apple II. Regarding CP/M vs. MS-DOS. MS-DOS is CP/M rebundled. CP/M was ripped (rewritten for other processors) from Kildall by Seattle Computer and sold to DR and MS. Personal opinion - if I had been in IBM's shoes I would have gone with Microsoft since the only real issue was support. GK was a nice guy, but BG was a hard worker (and probably still is.) Therefore, I contend that CP/M DID become what MS-DOS now is There was a PCjr and it was for the home market. Eric Chomko wrote in article <69h99r$fk0@nnrp1.farm.idt.net>... > Organization: IDT Internet Services > Distribution: > > In alt.folklore.computers John Bayko wrote: > : In article <69fve3$25q@nnrp1.farm.idt.net>, > [...] > : Not meant as a home computer, but neither was the IBM PC - the > : only computer for home IBM made back then was the PCjr, which was > : exactly the type of thing you'd get from a business computer company > : making a machine for the home. > > Yet many people bought the PC. And the clone market definitely made the > PC-system more attractive. > > : A lot of people say 'personal' computer means designed to be used > : by a single person, and point to the LINC or PDP-8 as the first > : example. > > Yet, what one individual could afford to buy these systems or their > following generations? The PC world (not my favorite, BTW) is still alive > and is dominated by individual buyers. > > > : > > : >The irony here is that IBM got the same "legal" issue response from Gates > : >as they got from Kildall, but rather than go to a third party, they > : >argreed with Gates' terms. It was rumored that they actually tried to go > : >back to Kildall and offer him the deal that they finally gave Gates, but > : >he (Kildall) was out flying his plane. > > : I don't believe this is the case. Microsoft agreed to IBMs terms > : in delivering the OS for the IBM PC - PC-DOS - but they did negotiatd > : a technical loophole that allowed them to produce and sell MS-DOS on > : its own terms, which was probably the difference. > > But CP/M was first consideration, correct? Why didn't it become what > MS-DOS DID become? > > Eric > ###### From: bayko@pollux.cs.uregina.ca (John Bayko) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 14 Jan 1998 23:46:47 GMT Organization: University of Regina, Dept. of Computer Science Lines: 40 Message-ID: <69jip7$eam$1@sue.cc.uregina.ca> References: <348fa205.3834263@news.compuserve.com> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> NNTP-Posting-Host: pollux.cs.uregina.ca Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.onenet.net!news.oru.edu!logbridge.uoregon.edu!scanner.worldgate.com!rover.ucs.ualberta.ca!tribune.usask.ca!news.uregina.ca!not-for-mail In article <01bd208d$848db940$1bc4accf@mfc>, Michael F. Coyle wrote: > >BTW, the Motorola 6809 was a major contender for the PC CPU. Supposedly, Bill >Gates convinced IBM that a 16-bitter was the way to go. The 8088 was a >singularly poor choice, though. Unlikely - the CPU was chosen before the OS was being selected. There was also no reason why IBM would consider the 6809, mainly because it had it's own unique bus/clock interface. The 6809 was considered for the Macintosh, as the same Great Microprocessors list recounts: Great Microprocessors of the Past and Present (V 10.0.0) Section Three: The Great Dark Cloud Falls: IBM's Choice. Part IV: Motorola 68000, a refined 16/32 bit CPU (1979) . . . . . . . . . [...] Few people wonder why Apple chose the Motorola 68000 for the Macintosh, while IBM's decision to use Intel's 8088 for the IBM PC has baffled many. It wasn't a straightforward decision though. The Apple Lisa was the predecessor to the Macintosh, and also used a 68000 (eventually - 8086 and slower bitslice CPUs (which Steve Wozniak thought were neat) were initially considered before the 68000 was available). It also included a fully multitasking, GUI based operating system, highly integrated software, high capacity (but incompatible) 'twiggy' 5 1/4" disk drives, and a large workstation-like monitor. It was better than the Macintosh in almost every way, but was correspondingly more expensive. The Macintosh was to include the best features of the Lisa, but at an affordable price - in fact the original Macintosh came with only 128K of RAM and no expansion slots. Cost was such a factor that the 8 bit Motorola 6809 was the original design choice, and some prototypes were built, but they quickly realised that it didn't have the power for a GUI based OS, and they used the Lisa's 68000, borrowing some of the Lisa low level functions (such as graphics toolkit routines) for the Macintosh. -- John Bayko (Tau). bayko@cs.uregina.ca http://www.cs.uregina.ca/~bayko ###### From: albaugh@agames.com (Mike Albaugh) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded Date: 15 Jan 1998 00:18:10 GMT Organization: Atari Games Corporation Lines: 44 Message-ID: <69jkk2$7ad$1@void.agames.com> References: <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net> NNTP-Posting-Host: java.agames.com X-Newsreader: TIN [version 1.2 PL2] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.onenet.net!news.oru.edu!logbridge.uoregon.edu!news-peer.gip.net!news.gsl.net!gip.net!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!newshub2.home.com!news.home.com!newshub1-work.home.com!void.agames.com!albaugh David Hunt (dhunt@mci2000.net) wrote: : I must take issue with the statement that the Original IBM PC was not : intended for home use. There was a game port and that game port was : definitely for home use. The game port was designed into the original : IBM PC motherboard. The PC was designed to be a direct competitor : in both the home and educational markets with the Apple II. : Regarding CP/M vs. MS-DOS. MS-DOS is CP/M rebundled. CP/M was : ripped (rewritten for other processors) from Kildall by Seattle : Computer and sold to DR and MS. And I have to take issue with that statement. CP/M really was an OS, albeit a small one. It had a fair implementation of device-independant I/O and a file structure that, while primitive, actually worked pretty well. QDOS (Quick and Dirty Operating System, which was Seattle Computer's original name) used the "file structure" (which we have all come to know and hate, aka FAT) ripped from MicroSoft Disk Basic. This was a hack from the get-go, but originally only had to compete with paper tape and audio cassettes, so neither speed nor robustness was an issue. As for device independance, yeah, right. The clone market was born from the total lack of discipline among app writers (admittedly faced with the least functional BIOS ever to disgrace a system). The prevalence of "going around the BIOS" was such as to tie IBM's hands, and present a sitting target for the cloners. As a side effect (hmmm, conspiracy theory fans want to get in on this? :-) It made Seattle's right to sell MSDOS pretty much moot, as the clones ran PCDOS, available only from guess who. Now, you can argue that the total stagnation of hardware development fueled the drastic price drops that we now expect, but it also spawned such monstrosities as SVGA and all sorts of "secret handshakes" with I/O devices, and bloatware. YMMV whether the net effect was positive or negative. From where I sit, the total man-hours spend dinking with config.sys and registry wizards per year is probably about equal to the alleged productivity gains from PCs :-) : Therefore, I contend that CP/M DID become what MS-DOS now is CP/M became a multi-tasking OS with a device independant GUI by the mid 80's. I'm still waiting for MSDOS to pull that off... Mike | albaugh@agames.com, speaking only for myself ###### From: Eric Chomko Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 15 Jan 1998 14:45:33 GMT Organization: IDT Lines: 33 Message-ID: <69l7ed$1c1@nnrp4.farm.idt.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69io NNTP-Posting-Host: u2.farm.idt.net X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!ais.net!news.idt.net!nntp.farm.idt.net!u2.farm.idt.net!not-for-mail Organization: IDT Internet Services Distribution: Bill B. wrote: Eric Chomko: : > : > Believe me I know exactly where Pacific Grove, Pebble Beach, Carmel and : > Monterey are located. My dad lives in Salinas. I was out there over : > Thanksgiving. : Hey Eric, great to hear your geography is up to snuff... However, : I suspect that most people would have no clue as to the whereabouts : of P.G. (Unless they're one of mega-masses to have visited the : Acquarium!) Sorry, didn't want to sound so pendantic! Yes, the aquarium is a nice place to visit. I really like the Outer Bay exhibit. : > Anyway, is it true that Gary Kildall fell off a barstool in that same area : > and died? That is the story I heard from the locals while hiking with some : > of them in the Grand Canyon about 5 years ago. The Kildall incident would : > have been in 90 or 91. Is any of this true? It sounds just too bizarre. : Don't know about the barstool incident, but I do know that Gary : has passed on..... Yeah, that's the story I heard. Eric ###### From: "Don Barr" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 15 Jan 1998 14:58:20 GMT Organization: OnRamp, http://www.onramp.net/ Lines: 28 Message-ID: <01bd21c5$c6fec060$083e32ce@louisville.keil.com> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve <69h99r$fk0@nnrp1.farm.idt.net> <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net> NNTP-Posting-Host: locutus.keil.com X-Newsreader: Microsoft Internet News 4.70.1155 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newspeer.monmouth.com!news-peer.gip.net!news.gsl.net!gip.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!newsfeed.direct.ca!news.he.net!news-feeder.onramp.net!news.onramp.net!not-for-mail What game port? IIRC, the original IBM PC had a connector for the keyboard and one for a tape (read "cassette") drive (removed on the XT). There was _NO_ game port on the motherboard, you could get a card for it later on though. The PC was a competitor for the Apple and others, but business and education were the primary goals. The home market for computers in general was still pretty much the realm of the "geeks" and "nerds". As far as using the 8-bit data path, this was simply a cost cutting decision. There were plenty of 8-bit peripherial chips around, using two each (for a 16-bit bus) would have added complexity and cost to the system that IBM felt wasn't necessary (besides, who was even going to buy these things, a real market would likely never develop (IBM strategy of the day)). -- Later DBarr donbarr@keil.com These opinions are my own. David Hunt wrote in article <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net>... > > I must take issue with the statement that the Original IBM PC was not > intended for home use. There was a game port and that game port was > definitely for home use. The game port was designed into the original > IBM PC motherboard. The PC was designed to be a direct competitor > in both the home and educational markets with the Apple II. > SNIP ###### From: Simon Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Thu, 15 Jan 1998 16:08:17 +0000 Organization: Imperial College Lines: 50 Message-ID: <34BE3471.40B3@ic.ac.uk> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve <69h99r$fk0@nnrp1.farm.idt.net> <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net> <01bd21c5$c6fec060$083e32ce@louisville.keil.com> Reply-To: s.j.harris@ic.ac.uk NNTP-Posting-Host: linpc.me.ic.ac.uk Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.0 (Win95; I) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!nntp.news.xara.net!xara.net!server5.netnews.ja.net!server3.netnews.ja.net!server4.netnews.ja.net!news.cc.ic.ac.uk!not-for-mail Don Barr wrote: > > .... As far as using the 8-bit data path, this was simply a cost > cutting decision. There were plenty of 8-bit peripherial chips around, > using two each (for a 16-bit bus) would have added complexity and cost to > the system that IBM felt wasn't necessary (besides, who was even going to > buy these things, a real market would likely never develop (IBM strategy of > the day)). > -- I'd suspect that it was the overall 8-bit data architecture that sold the 8088. There is not really a problem connecting 8-bit peripherals to a 16 bit bus (except remembering that all data is only 8 bits wide and all addressed on even (or odd) addresses). I believe this is similar to how the 68000 handles 8 bit peripherals with the movep instruction. What the 8 bit architecture does give you is: simpler board design - easier to route 8 than 16 lines of data. I seem to remember seeing a photograph in Byte of the prototype PC with wirewrapped main-board. Remember also that the data/address busses are multiplexed on the 8086/8088, so using the 8088 simplifies the design around the address latches (LSB 8 bits only needed latching). use of single cheap EPROMs for the BIOS. I believe Intel may just have brought out 16 bit wide EPROMs when the PC appeared, but the 8 bit ones were considerably cheaper. An EPROM pair is a possibility, but increases cost and board complexity. cheaper RAM configurations. RAM can be installed in multiples of 9 (including parity) chips instead of 18. Back in '79 RAM was expensive. simpler backplane socket arrangements (remember in early PCs there had to be a lot of these as the only things on the motherboard then were: CPU base memory EPROMs keyboard controller timers / speaker control interrupt controller) It would have been interesting to see whether if IBM has gone the Zilog way, whether we would now have 300 MHz, Z80000000s (as Zilog seem to add an extra 0 with each generation) which still had to be backward compatible with the original Z80 :) Simon. ###### Path: ccw.ch!usenet From: Neil.Franklin.nospam@ccw.ch (remove .nospam) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 15 Jan 1998 16:57:52 +0100 Organization: My own Private Self Lines: 49 Message-ID: References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69imkg$1nd@nnrp1.farm.idt.net> X-Newsreader: Gnus v5.3/Emacs 19.34 Eric Chomko wrote: > Where is Kidall's gazillions? Killdall was simply not interested in making them. He was an engineer by heart, not a business man like Gates. Not everyone regard a mound of money as the pinacle of human development. He only wrote CP/M for his own use and gave it out to help others. Actually his wife did all the business side of things. ------- repost of an old message of mine From: Neil.Franklin.nospam@ccw.ch (remove .nospam) Newsgroups: comp.unix.admin,comp.os.linux.advocacy,comp.unix.advocacy Subject: Re: Linux/Unix versus NT Date: 02 Jan 1998 04:24:35 +0100 Message-ID: felinoid@news.ccnet.com (Jeffery McLean) wrote: >Digial Research and IBM was not seeing eye to eye for some reason... >MsDos would not have existed if IBM got the CP/M 86 contract. According to an TV Interview of the IBM Manager responsible for getting an OS for the PC with Robert X Cringley the following happened: IBM (he) appeared at DRI and wanted a CP/M license. They presented the person at DRI doing the deal (Gary Killdals wife) their normal NDA contract. She was scared by it and had an befriended lawyer read it before she would sign. He regarded some of the stuff in it as dubious. So she didn't sign. While IBM was going through internal processes to resolve the impasse Bill Gates heard of the episode. Knowing that Tim Paterson of Seattle Computers had written an CP/M clone called QDOS he quickley made an offer to IBM, then bought QDOS. Rumour has it for $50'000. The rest, as they say, is history. Then we all called IBM the evil, now Microsoft. ------- end repost > I heard that he's no longer with us in fact. > Sad tale, if it's to be believed. This is true. Accoring to an Computerworld I read (ca 1992/93) he was killed in an armed robbery when leaving a restaurant. R.I.P. -- Neil.Franklin.nospam@ccw.ch (remove .nospam), http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: "Michael F. Coyle" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 15 Jan 1998 20:54:35 GMT Organization: Erol's Internet Services Lines: 19 Message-ID: <01bd21f7$848e5100$8cc3accf@mfc> References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> NNTP-Posting-Host: smt-as1s13.erols.com X-Trace: winter.news.erols.com 884897675 10393 207.172.195.140 (15 Jan 1998 20:54:35 GMT) X-Complaints-To: abuse@erols.com X-Newsreader: Microsoft Internet News 4.70.1161 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.onenet.net!news.oru.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.gip.net!news.gsl.net!gip.net!howland.erols.net!winter.news.erols.com!not-for-mail "Paul E. Bennett" wrote in article <884810146snz@transcontech.co.uk>... > I would have rated the 6809 as the better choice, then being a Forthist I > would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 > has 16 bit ALU operations and a nicer feel to it. I know that the data bus > was 8 bits but that is not always the performance bottleneck. Can't agree with you here. Although the 6809 was, arguably, the most elegant 8-bit micro around, it was *still* an 8-bitter with a 64K address space. Nifty instructions and addressing modes aren't much help if you're trying to address more memory than the processor can handle. In an interview in Byte, Don Estridge said that IBM was looking for a ten-year life for the PC architecture. No way could it have been done with an 8-bit CPU. - Michael ###### From: Ben Hutchings Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 15 Jan 1998 23:59:33 -0000 Organization: Not organised Lines: 34 Sender: womble@max93.public.ox.ac.uk Message-ID: <69m7t5$3el$1@max93.public.ox.ac.uk> References: <348fa205.3834263@news.compuserve.com> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> NNTP-Posting-Host: max93.public.ox.ac.uk Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!newspeer.monmouth.com!news-peer.gip.net!news-lond.gip.net!news.gsl.net!gip.net!baron.netcom.net.uk!netcom.net.uk!server3.netnews.ja.net!news.ox.ac.uk!not-for-mail In article <01bd21f7$848e5100$8cc3accf@mfc>, Michael F. Coyle wrote: >"Paul E. Bennett" wrote in article ><884810146snz@transcontech.co.uk>... > >> I would have rated the 6809 as the better choice, then being a Forthist I >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus >> was 8 bits but that is not always the performance bottleneck. > >Can't agree with you here. Although the 6809 was, arguably, the most >elegant 8-bit micro around, it was *still* an 8-bitter with a 64K >address space. Nifty instructions and addressing modes aren't much >help if you're trying to address more memory than the processor can >handle. Nevertheless, 8-bit machines have been made with significantly more than 64K RAM, and I don't think Intel's segmented memory model was significantly less ugly than their paging systems. >In an interview in Byte, Don Estridge said that IBM was looking for a ten-year >life for the PC architecture. No way could it have been done with an 8-bit >CPU. That's interesting to hear, given that they managed to re-do most of the architecture (and get it wrong again) with the AT after somewhat less than 10 years, and that Intel had to introduce 2 completely new memory models for the x86 in that time. -- Ben Hutchings, M&CS student | Jay Miner Society website: http://www.jms.org/ email/finger m95bwh@ecs.ox.ac.uk | homepage http://users.ox.ac.uk/~worc0223/ Asking whether machines can think is like asking whether submarines can swim. ###### From: dski@cameonet.cameo.com.twx (Dan Strychalski) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 16 Jan 1998 04:41:47 GMT Organization: Cameo Communications, Inc. Lines: 23 Message-ID: <69moeb$cgi@news.seed.net.tw> NNTP-Posting-Host: 192.72.104.4 Originator: dski@ Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.gip.net!news.gsl.net!gip.net!spring.edu.tw!feeder.seed.net.tw!news.seed.net.tw!!dski Neil Franklin wrote in news:soqppw5b.fsf@chonsp.franklin.lugs.ch -- > Killdall was simply not interested in making them. He was an engineer > by heart, not a business man like Gates. Not everyone regard a mound > of money as the pinacle of human development. He only wrote CP/M for > his own use and gave it out to help others. Actually his wife did all > the business side of things. [snip -- DS] >> I heard that he's no longer with us in fact. >> Sad tale, if it's to be believed. > > This is true. Accoring to an Computerworld I read (ca 1992/93) he was > killed in an armed robbery when leaving a restaurant. R.I.P. This last account does not match what I've read. For remembrances of Gary Kildall, see -- http://www.forbes.com/forbes/97/0707/6001336a.htm and either http://biz2.idbsu.edu/is/is102/kildall.htm or http://www.intac.com/acgnj/newslett/newsx011.html Dan Strychalski dski at cameonet, cameo, com, tw ###### From: kenl@ausonics.mpx.com.au (Ken Lee) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 16 Jan 1998 07:24:41 GMT Organization: Microplex Pty Ltd Lines: 27 Message-ID: <34bf0790.28532447@news.mpx.com.au> References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> NNTP-Posting-Host: 203.15.64.15 X-Newsreader: Forte Free Agent 1.11/32.235 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsm.ibm.net!ibm.net!uunet!in1.uu.net!ozemail!news.mel.aone.net.au!newsfeed-in.aone.net.au!inferno.mpx.com.au!news On Wed, 14 Jan 98 20:35:46 GMT, peb@transcontech.co.uk ("Paul E. Bennett") wrote: >In article <01bd208d$848db940$1bc4accf@mfc> > mjcoyle@erols.com "Michael F. Coyle" writes: > >> BTW, the Motorola 6809 was a major contender for the PC CPU. Supposedly, Bill >> Gates convinced IBM that a 16-bitter was the way to go. The 8088 was a >> singularly poor choice, though. > >I would have rated the 6809 as the better choice, then being a Forthist I >would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 >has 16 bit ALU operations and a nicer feel to it. I know that the data bus >was 8 bits but that is not always the performance bottleneck. I could be wrong but doesn't the 6809 have a memory space of 64K. I guess it could be paged, but have you ever used a paged system, I have and it's murder. Probably the 8088 wasn't the best choice, but it's got to be better than the 6809. I think the Colour Tandy used the 6809 (please correct me if I'm wrong), but it wasn't as successful as the TRS-80. >Since BG persuaded IBM to go that way we have all had to live with a crazy >architecture. No one these days worries about the 640K barrier except for those still running MSDOS and these people must be on the extinction list. Ken. ###### From: Eric Chomko Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 16 Jan 1998 19:00:25 GMT Organization: IDT Lines: 23 Message-ID: <69oao9$9f8@nnrp1.farm.idt.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69im NNTP-Posting-Host: u1.farm.idt.net X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!news-peer.gip.net!news.gsl.net!gip.net!news.idt.net!nntp.farm.idt.net!u1.farm.idt.net!not-for-mail Organization: IDT Internet Services Distribution: In alt.folklore.computers remove .nospam wrote: : Eric Chomko wrote: : > Where is Kidall's gazillions? : Killdall was simply not interested in making them. He was an engineer : by heart, not a business man like Gates. Not everyone regard a mound : of money as the pinacle of human development. He only wrote CP/M for : his own use and gave it out to help others. Actually his wife did all : the business side of things. I ain't buying it. Sure, he may not been as driven as Gates and certainly becoming a billionare may not have been his driving passion. But theres' no way that once passed up by IBM and after seeing Gates strike it rich, that Kildall didn't feel green for not having more green. Eric P.S. I bet YOU are an engineer, too, right? ###### From: J. Chris Hausler Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 16 Jan 98 19:55:19 -0500 Organization: Delphi (info@delphi.com email, 800-695-4005 voice) Lines: 20 Message-ID: References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> <884810 <01bd21f7$848e5100$8cc3accf@mfc> NNTP-Posting-Host: 199.93.4.4 X-To: "Michael F. Coyle" Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!news-peer.gip.net!news.gsl.net!gip.net!news-peer.sprintlink.net!news-peer-east.sprintlink.net!news.sprintlink.net!Sprint!news.idt.net!cam-news-hub1.bbnplanet.com!cam-news-feed1.bbnplanet.com!news.bbnplanet.com!news.delphi.com!news "Michael F. Coyle" writes: >> I would have rated the 6809 as the better choice, then being a Forthist I >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus >> was 8 bits but that is not always the performance bottleneck. > >Can't agree with you here. Although the 6809 was, arguably, the most elegant >8-bit micro around, it was *still* an 8-bitter with a 64K address space. Nifty >instructions and addressing modes aren't much help if you're trying to address >more memory than the processor can handle. Although the 09 is a favorite of mine, still have several runable SS-50 systems, from a marketing point of view the 09 was a little too little, a little too late. Some 6800/6809-likes, the HC11 et all have been quite successful in the controller market. Of course the RS Color Computer (COCO) was an 09 as well but it was never intended as a "serious" computer (despite what some of its owners did with it with os9). Chris ###### From: TT Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 16 Jan 1998 23:11:36 -0500 Organization: self Lines: 43 Message-ID: <34C02F78.215F268C@mediaone.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69h9 <69imkg$1nd@nnrp1.farm.idt.net> Reply-To: tt0001@usa.net NNTP-Posting-Host: sloopy.ne.mediaone.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.04 [en] (Win95; U) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!arclight.uoregon.edu!newsfeed.internetmci.com!24.128.1.125!chnws03.mediaone.net!24.128.1.107!chnws04.ne.mediaone.net!not-for-mail If a TV program I saw is to believed, Kidall was only worth about $5M when it was recorded (about 1990?). No Bill Gates, but still a comfortable living... As for your question, in the end, THERE CAN BE ONLY ONE! (highlander:) Eric Chomko wrote: > > In alt.folklore.computers Roger Ivie wrote: > : In article <69h99r$fk0@nnrp1.farm.idt.net>, Eric Chomko writes: > : > But CP/M was first consideration, correct? Why didn't it become what > : > MS-DOS DID become? > > : But it did! CP/M-86 -> Concurrent CP/M-86, where it picked up DOS compatability > : and became Concurrent DOS -> DR DOS -> Novell DOS -> Caldera OpenDOS. CP/M-86 > : became an MS-DOS clone. > > Somehow I think you missed my point.... > > Where is Kidall's gazillions? I heard that he's no longer with us in fact. > Sad tale, if it's to be believed. > > Eric -- ============================================================= Tim Tait No SPAM! Note that reply-to address is temporary and will expire eventually. Real Email: ttait at above domain ============================================================= And for you automated email spammers out there, here's the email addresses of the current board of the Federal Communications Commission: Chairman Reed Hundt: rhundt@fcc.gov Commissioner James Quello: jquello@fcc.gov Commissioner Susan Ness: sness@fcc.gov Commissioner Rachelle Chong: rchong@fcc.gov ###### From: TT Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 16 Jan 1998 23:14:09 -0500 Organization: self Lines: 37 Message-ID: <34C03011.25CFAF21@mediaone.net> References: <348fa205.3834263@news.compuserve.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69itol$agj$1@wrqnews.wrq.com> Reply-To: tt0001@usa.net NNTP-Posting-Host: sloopy.ne.mediaone.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.04 [en] (Win95; U) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.he.net!news-peer.gip.net!news.gsl.net!gip.net!newsfeed.internetmci.com!24.128.1.125!chnws03.mediaone.net!24.128.1.107!chnws04.ne.mediaone.net!not-for-mail The Z80 was around then. According to EDN magazine a few years back in their microprocessor directory, the Z80 and it's variants have more design wins than any other processor. David Thompson wrote: > > I remember reading that Intel created the 8088 from the 8086 specifically > for IBM so IBM could lower the cost of the first PCs. The 8088 had an eight > bit external bus versus 16 for the 8086. I am not sure if I have the > terminology correct, but Intel's willingness to cut its chip off at the > knees may have been a factor. > > >>PRODUCT TO BIG BLUE for a bargain price! I think "it was cheaper" is the > >>real story, otherwise IBM was going to go with their first choice, Zilog! > >>Zilog could have right now been the largest CPU maker. Come to think of > >>it, I think IBM got to use Intel product royalty-free for either one or > >>two years as IBM experimented with the "personal computer" idea.......sq > > > > Unlikely. The Zilog Z-8000 was too late. The 68000 was available, > >but didn't have support chips available at the time. However there was > >a licensing agreement apparently involved... -- ============================================================= Tim Tait No SPAM! Note that reply-to address is temporary and will expire eventually. Real Email: ttait at above domain ============================================================= And for you automated email spammers out there, here's the email addresses of the current board of the Federal Communications Commission: Chairman Reed Hundt: rhundt@fcc.gov Commissioner James Quello: jquello@fcc.gov Commissioner Susan Ness: sness@fcc.gov Commissioner Rachelle Chong: rchong@fcc.gov ###### Newsgroups: comp.arch.embedded,alt.folklore.computers From: john@cognac.localnet (John Pearson) Subject: Re: Why IBM chose Intel Sender: news@huiac.apana.org.au (news) References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> X-Newsreader: slrn (0.9.3.2 UNIX) Organization: John & Terry Pearson at home Message-ID: X-Nntp-Posting-Host: dewiac.localnet Date: Sat, 17 Jan 1998 01:38:51 GMT Lines: 47 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!streamer1.cleveland.iagnet.net!qual.net!iagnet.net!news.idt.net!ais.net!uunet!in3.uu.net!munnari.OZ.AU!news.mel.connect.com.au!news.ade.connect.com.au!news.mtx.net.au!chuckie.apana.org.au!huiac!john In article <01bd208d$848db940$1bc4accf@mfc>, Michael F. Coyle wrote: >Frank van der Hulst wrote in article ><34BBE9CE.646126A1@pec.co.nz>... >> John Bayko wrote: >> > >The IBM 5100 was not for home use, unless you had some $$$. At $20K a pop >> > >they were business machines. (I worked on both the 5110 and the 5100 - in >> > >> > Not meant as a home computer, but neither was the IBM PC - the >> >> I disagree... one of the early add-ons for a PC was a 'game >> controller' card (the cheesy R-C network 4-channel A/D and 4 digital >> inputs). I recall this was included in IBM's Technical Reference >> Manual. Now, of course *I* don't play games at work, and maybe you do, >> but I'd say this is a clear indication that the 'home' market was >> included in the target. > >Plus, the original PC had a cassette interface, for those home users who >couldn't afford floppies. IBM apparently wasn't sure what configurations and >price points would work, so they covered all the bases. (Everybody bought the >machine with disks, though). > >BTW, the Motorola 6809 was a major contender for the PC CPU. Supposedly, Bill >Gates convinced IBM that a 16-bitter was the way to go. The 8088 was a >singularly poor choice, though. > With the benefit of hindsight, perhaps. Even given the choice of the 8088, they certainly should have been able to come up with a more future-proof hardware design ("well, I mean: edge-triggered interrupts, memory mapping hardware at the top of a 20-bit address space, no beyond- the-box expansion, expansion bus speed fixed at 4.77MHz; what were they thinking?" - hindsight). It was my understanding that one of the main reasons for choosing the 8088 was the availabilitiy of support chips: the 8088 was a 16-bit CPU (derived from the 8086) which used a 8-bit data bus (sort of like the right half of an 80386SX :-); it had the advantage that it could use many of the support chips designed for the popular 8085 whereas most 16-bit CPUs were quite new, and didn't yet have a well-proven support family (ever wonder why there are only 8 interrupt levels on the PC, of which they used one as a cascade on the AT to bring it up to 15?). John P. -- john@huiac.apana.org.au ###### From: peb@transcontech.co.uk ("Paul E. Bennett") Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sat, 17 Jan 98 12:42:24 GMT Organization: Transport Control Technology Ltd. Message-ID: <885040944snz@transcontech.co.uk> References: <348fa205.3834263@news.compuserve.com> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> Reply-To: peb@transcontech.co.uk X-Mail2News-User: peb@transcontech.co.uk X-Mail2News-Path: tcontec.demon.co.uk X-Trace: mail2news.demon.co.uk 885126873 12654 peb transcontech.co.uk X-Complaints-To: abuse@demon.net X-Newsreader: Demon Internet Simple News v1.30 Lines: 46 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!nntprelay.mathworks.com!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!transcontech.co.uk!peb In article <69m7t5$3el$1@max93.public.ox.ac.uk> worc0223@sable.ox.ac.uk "Ben Hutchings" writes: > In article <01bd21f7$848e5100$8cc3accf@mfc>, > Michael F. Coyle wrote: > >"Paul E. Bennett" wrote in article > ><884810146snz@transcontech.co.uk>... > > > >> I would have rated the 6809 as the better choice, then being a Forthist I > >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 > >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus > >> was 8 bits but that is not always the performance bottleneck. > > > >Can't agree with you here. Although the 6809 was, arguably, the most > >elegant 8-bit micro around, it was *still* an 8-bitter with a 64K > >address space. Nifty instructions and addressing modes aren't much > >help if you're trying to address more memory than the processor can > >handle. If you need the extra RAM Motorola had the MMU to cope with significantly larger amounts of memory. That too was elegant. Whilst 6809's databus was 8 bit there were several registers and the ALU that were 16 bit and could perform 16 bit operations. > >In an interview in Byte, Don Estridge said that IBM was looking for a ten- > >year > >life for the PC architecture. No way could it have been done with an 8-bit > >CPU. > > That's interesting to hear, given that they managed to re-do most of > the architecture (and get it wrong again) with the AT after somewhat > less than 10 years, and that Intel had to introduce 2 completely new > memory models for the x86 in that time. To get long life in electronic systems these days you have to have a sound structure to begin building with and one that will not cripple you in the long run. The embedded industrial control market seems to have been very happy with the progress from 6809/6309 to 68008 to 68000 .... 68040 etc. in much the same time period. -- Paul E. Bennett ................... Transport Control Technology Ltd. +44 (0)117-9499861 Going Forth Safely ###### From: root@localhost Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sun, 18 Jan 1998 00:25:34 GMT Message-ID: <885083134.5471.10.nnrp-06.9e98ee68@news.demon.co.uk> References: <01bd21f7$848e5100$8cc3accf@mfc> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 32 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!arclight.uoregon.edu!newsfeed.internetmci.com!4.1.16.34!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-15 mjcoyle@erols.com said: -Can't agree with you here. Although the 6809 was, arguably, the -most elegant 8-bit micro around, it was *still* an 8-bitter with a -64K address space. Nifty instructions and addressing modes aren't -much help if you're trying to address more memory than the -processor can handle. perhaps not, but if you're using the 8088 that's precisely the situation you're faced with. it's a 16 bit processor, period. the segmentation stuff was a kludge; a better way to approach it would have been to have a separate MMU and page in memory. the 6809 had one of those. the 8088 didn't need one, but was a godawful mess. our complaint against the 8088, though, is more fundamental. 16 cycles for a jump instruction? come off it!!! -In an interview in Byte, Don Estridge said that IBM was looking for -a ten-year life for the PC architecture. No way could it have been -done with an 8-bit CPU. cp/m nearly made it through to the end of the decade, you know... and by 1991, who the hell was using 8088 systems any more? (except the penniless, ie. moi) even then, such things as the 640k barrier &c. were presenting a real problem. -- Communa (Communa@lisardrock.demon.net) Don't just blindly reply. You'll talk to your sysop. Net-Tamer V 1.09.2 - Test Drive ###### From: root@localhost Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sun, 18 Jan 1998 00:25:36 GMT Message-ID: <885083136.5471.11.nnrp-06.9e98ee68@news.demon.co.uk> References: <34BE3471.40B3@ic.ac.uk> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 23 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-15 s.j.harris@ic.ac.uk said: -It would have been interesting to see whether if IBM has gone -the Zilog way, whether we would now have 300 MHz, Z80000000s -(as Zilog seem to add an extra 0 with each generation) which -still had to be backward compatible with the original Z80 :) naah. but they may well have had to be compatible with the original z8000 - but then, given that this design wasn't microcoded and used fewer transistors than the 8088 anyway, and also that the z80000 had many things that intel only came up with several years later after scratching their head and going "what the hell?" for extended periods, doing that probably wouldn't have been all that difficult. mmm, z8000... does anyone out there (uk-ish, preferably) still have an olivetti m20? what's it like? still need it...? :> -- Communa (Communa@lisardrock.demon.net) Don't just blindly reply. You'll talk to your sysop. Net-Tamer V 1.09.2 - Test Drive ###### From: "Michael F. Coyle" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 18 Jan 1998 01:36:51 GMT Organization: Erol's Internet Services Lines: 19 Message-ID: <01bd23b1$40fc44e0$a6c3accf@mfc> References: <348fa205.3834263@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> NNTP-Posting-Host: smt-as1s39.erols.com X-Trace: winter.news.erols.com 885087411 2662 207.172.195.166 (18 Jan 1998 01:36:51 GMT) X-Complaints-To: abuse@erols.com X-Newsreader: Microsoft Internet News 4.70.1161 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!207.114.4.11.MISMATCH!nntp.abs.net!feed2.news.erols.com!winter.news.erols.com!not-for-mail John Pearson wrote in article ... > It was my understanding that one of the main reasons for choosing > the 8088 was the availabilitiy of support chips: the 8088 was a > 16-bit CPU (derived from the 8086) which used a 8-bit data bus > (sort of like the right half of an 80386SX :-); it had the advantage > that it could use many of the support chips designed for the popular > 8085 whereas most 16-bit CPUs were quite new, and didn't yet have > a well-proven support family (ever wonder why there are only 8 > interrupt levels on the PC, of which they used one as a cascade on > the AT to bring it up to 15?). Yes, but the 68000 had a mode where it could use 6800/6500 peripheral chips. There were at least as many of these as there were 82xx peripherals. So what's the hangup? - Michael ###### From: "Michael F. Coyle" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 18 Jan 1998 02:13:44 GMT Organization: Erol's Internet Services Lines: 50 Message-ID: <01bd23b5$70471000$a6c3accf@mfc> References: <348fa205.3834263@news.compuserve.com> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> NNTP-Posting-Host: smt-as1s39.erols.com X-Trace: winter.news.erols.com 885089624 25596 207.172.195.166 (18 Jan 1998 02:13:44 GMT) X-Complaints-To: abuse@erols.com X-Newsreader: Microsoft Internet News 4.70.1161 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.nodak.edu!netnews3.nwnet.net!netnews.nwnet.net!news.u.washington.edu!newsfeed.direct.ca!howland.erols.net!feed1.news.erols.com!winter.news.erols.com!not-for-mail Ben Hutchings wrote in article <69m7t5$3el$1@max93.public.ox.ac.uk>... > In article <01bd21f7$848e5100$8cc3accf@mfc>, > Michael F. Coyle wrote: > >"Paul E. Bennett" wrote in article > ><884810146snz@transcontech.co.uk>... > > > >> I would have rated the 6809 as the better choice, then being a Forthist I > >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 > >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus > >> was 8 bits but that is not always the performance bottleneck. > > > >Can't agree with you here. Although the 6809 was, arguably, the most > >elegant 8-bit micro around, it was *still* an 8-bitter with a 64K > >address space. Nifty instructions and addressing modes aren't much > >help if you're trying to address more memory than the processor can > >handle. > > Nevertheless, 8-bit machines have been made with significantly more > than 64K RAM, and I don't think Intel's segmented memory model was > significantly less ugly than their paging systems. True. But note: 8088 memory management is not really segmentation -- it is a kind of mapping barely more sophisticated than bank switching. The 286 had genuine segmentation, but with a 64K limit. The 386 removed this limitation. Segmentation per se is not a bad thing -- but segments with a small maximum size creates headaches for the poor programmer. > >In an interview in Byte, Don Estridge said that IBM was looking for a ten-year > >life for the PC architecture. No way could it have been done with an 8-bit > >CPU. > > That's interesting to hear, given that they managed to re-do most of > the architecture (and get it wrong again) with the AT after somewhat > less than 10 years, and that Intel had to introduce 2 completely new > memory models for the x86 in that time. I don't think he meant they would be selling 8088 machines for 10 years -- rather, that it would be possible to develop follow-on machines with more speed, more memory, more peripherals, and the ability to run existing software. In this sense, the PC-AT and later 386-based PCs, though not elegant, show that it was possible. I don't think that any 8-bit CPU would have allowed for this kind of evolution over the 17+ years that the PC architecture has been around. (Which is not to say that some other 16- or 32-bit CPU couldn't have done a better job.) - Michael ###### From: ttoews@telusplanet.net (Tony Toews) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sun, 18 Jan 1998 02:29:31 GMT Organization: TELUS Communications Inc. Lines: 31 Message-ID: <34c14ce6.71890229@news.calgary.telusplanet.net> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> NNTP-Posting-Host: vrmlpx01-port-6.agt.net X-Newsreader: Forte Agent .99e/32.227 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!nntprelay.mathworks.com!newsfeed.internetmci.com!206.172.150.11!news1.bellglobal.com!news.agtac.net!news.telusplanet.net!news Eric Chomko wrote: >: Other factors were the 8-bit 8088 (1979) version, which could use existing >: low cost 8085-type components, and allowed the computer to be based on a >: modified 8085 design. 68000 components were not widely available, though it >: could use 6800 components to an extent. After the failure and expense of the >: IBM 5100 (1974/5/6? - their first attempt at a peronal computer - discrete >The IBM 5100 was not for home use, unless you had some $$$. At $20K a pop >they were business machines. (I worked on both the 5110 and the 5100 - in >that order some time ago. There was even an APL interpreter for the 5110 I >believe). Correct. I worked on those as well. We sold quite a few general purpose acounting systems on the 5110. Basic and 32Kb of memory was standard. APL and an additional 32 kb were options. The IBM 5100 was a tape only machine. The 5110 and repackaged 5120 had two 8" floppies. I worked on these in 1979 and 1980 and saw the 5120's in use until the late 80's. Up here in Canada they went for about $25k a pop. However I was a junior programmer and thus not involved in any of the sales side of things so I'm not sure of the exact price. Tony ---- Message posted to newsgroup and emailed. Tony Toews, Independent Computer Consultant The Year 2000 crisis: Will my parents or your grand parents still be receiving their pension in January, 2000? See www.granite.ab.ca/year2000 for more info. Microsoft Access Hints, Tips & Accounting Systems at www.granite.ab.ca/accsmstr.htm ###### From: Neil.Franklin.remove.this@ccw.ch Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 18 Jan 1998 04:31:49 +0100 Organization: My own Private Self Lines: 47 Message-ID: References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69oao9$9f8@nnrp1.farm.idt.net> X-Newsreader: Gnus v5.3/Emacs 19.34 Eric Chomko wrote: > Where is Kidall's gazillions? wrote: : Killdall was simply not interested in making them. He was an engineer : by heart, not a business man like Gates. Not everyone regard a mound : of money as the pinacle of human development. He only wrote CP/M for : his own use and gave it out to help others. Actually his wife did all : the business side of things. Eric Chomko replied: >I ain't buying it. Sure, he may not been as driven as Gates and certainly >becoming a billionare may not have been his driving passion. That line of yours (Sure...) is all I claimed: He was not driven by the desire to make riches, like Gates is. So actually you have bought it. >But theres' >no way that once passed up by IBM and after seeing Gates strike it rich, >that Kildall didn't feel green for not having more green. Oh sure. I also assume that after-the-fact he got angry. Missing such riches being served to you is surely frustrating. But bar having a time machine he had no way to go back and change his (or rather Dorothy's, his wifes) decision made in 1980/81. Most likely he never imagined that CP/M would let him get rich. After all 1970s PCs were not big business. That there was a gold mine there (just open the tap) may never have dawned on him. >P.S. I bet YOU are an engineer, too, right? Actually system/net/web-admin, but that goes in the same box. As you are bringing my person into the discussion: given the choice to exchange with BG (both the billions and the job situation), say in form of a time machine, I would pass it up. I have yet to find anything I would want to buy and can't afford that would make having his job (time use, competition climate) acceptable. Actually with a time machine I would take the entire Linux system source and fly back and jump-start the free software movement. -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: Ben Hutchings Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 18 Jan 1998 06:34:51 -0000 Organization: Not organised Lines: 59 Sender: womble@max89.public.ox.ac.uk Message-ID: <69s7qb$df2$1@max89.public.ox.ac.uk> References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> NNTP-Posting-Host: max89.public.ox.ac.uk Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!newsxfer3.itd.umich.edu!europa.clark.net!152.158.16.55!newsfeed2.uk.ibm.net!ibm.net!news-lond.gip.net!news.gsl.net!gip.net!baron.netcom.net.uk!netcom.net.uk!server3.netnews.ja.net!news.ox.ac.uk!not-for-mail In article <01bd23b5$70471000$a6c3accf@mfc>, Michael F. Coyle wrote: >Ben Hutchings wrote in article ><69m7t5$3el$1@max93.public.ox.ac.uk>... >> Nevertheless, 8-bit machines have been made with significantly more >> than 64K RAM, and I don't think Intel's segmented memory model was >> significantly less ugly than their paging systems. > >True. But note: 8088 memory management is not really segmentation -- >it is a kind of mapping barely more sophisticated than bank >switching. The 286 had genuine segmentation, but with a 64K limit. >The 386 removed this limitation. > >Segmentation per se is not a bad thing -- but segments with a small >maximum size creates headaches for the poor programmer. Well, I was referring to 8088 "segmentation" here of course. >> >In an interview in Byte, Don Estridge said that IBM was looking >> >for a ten-year life for the PC architecture. No way could it have >> >been done with an 8-bit CPU. >> >> That's interesting to hear, given that they managed to re-do most of >> the architecture (and get it wrong again) with the AT after somewhat >> less than 10 years, and that Intel had to introduce 2 completely new >> memory models for the x86 in that time. > >I don't think he meant they would be selling 8088 machines for 10 >years -- rather, that it would be possible to develop follow-on >machines with more speed, more memory, more peripherals, and the >ability to run existing software. In this sense, the PC-AT and later >386-based PCs, though not elegant, show that it was possible. I >don't think that any 8-bit CPU would have allowed for this kind of >evolution over the 17+ years that the PC architecture has been >around. (Which is not to say that some other 16- or 32-bit CPU >couldn't have done a better job.) But the 286 and 386 introduced entirely new operating modes and memory models, with the 8088 model highly deprecated (though DOS kept it in use for a long time). I don't see the 8088 as an architecture that has lasted 20 years, but one that has been supported for 20 years purely for backwards compatibility. The preferred operating modes of the i386 architecture, which was introduced in 1985, seem to have almost nothing to do with the original 8088 architecture. I don't suppose starting off with an 8-bit architecture would have made the transitions to a 32-bit architecture significantly more painful. The 680x0 series kept the same user register model and the same memory model from 1979 right up to the 68060 in 1994(?) at which point it stopped. Motorola didn't see the point in piling more money into it when their main high-end user (Apple) was happy to switch to PowerPC. If they had the money Intel has to spend on processors they could have kept on extending and accelerating it in the same way Intel has. -- Ben Hutchings, M&CS student | Jay Miner Society website: http://www.jms.org/ email/finger m95bwh@ecs.ox.ac.uk | homepage http://users.ox.ac.uk/~worc0223/ I haven't lost my mind; it's backed up on tape somewhere. ###### From: dpeschel@u.washington.edu (D. Peschel) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 18 Jan 1998 07:28:25 GMT Organization: University of Washington, Seattle Lines: 10 Message-ID: <69saup$1398$1@nntp6.u.washington.edu> References: <348fa205.3834263@news.compuserve.com> <69dnfv$re4$1@sue.cc.uregina.ca> <69oao9$9f8@nnrp1.farm.idt.net> NNTP-Posting-Host: saul5.u.washington.edu X-Trace: nntp6.u.washington.edu 885108505 36136 (None) 140.142.64.7 X-Complaints-To: help@cac.washington.edu NNTP-Posting-User: dpeschel Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!atl-news-feed1.bbnplanet.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!news-peer.sprintlink.net!news.sprintlink.net!Sprint!worldnet.att.net!news.u.washington.edu!dpeschel In article , wrote: >Actually with a time machine I would take the entire Linux system >source and fly back and jump-start the free software movement. You mean the entire Neilux system source? I don't think it would be called Linux under those circumstances. -- Derek ###### From: No@Spam@vcd.hp.com (No@Spam) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded,alt.folklore.computers Date: 19 Jan 1998 02:00:22 GMT Organization: Hewlett-Packard Lines: 13 Message-ID: <69uc3m$117@wwproxy.vcd.hp.com> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve <69h99r$fk0@nnrp1.farm.idt.net> <01bd2133$f3bde8c0$472737a6@dhunt.iquest.net> <01bd21c5$c6fec060$083e32ce@louisville.keil.com> <34BE3471.40B3@ic.ac.uk> NNTP-Posting-Host: hpvcljan.vcd.hp.com X-Newsreader: TIN [version 1.2 PL2.9] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!fci-se!fci!nntprelay.mathworks.com!newsfeed.internetmci.com!128.174.5.49!vixen.cso.uiuc.edu!sdd.hp.com!hp-corv.cv.hp.com!wwproxy.vcd.hp.com!No@Spam Simon (s.j.harris@ic.ac.uk) wrote: : It would have been interesting to see whether if IBM has gone : the Zilog way, whether we would now have 300 MHz, Z80000000s : (as Zilog seem to add an extra 0 with each generation) which : still had to be backward compatible with the original Z80 :) I think their 16 bit chip (Z8+some 0's) was already incompatible with the Z80. --- #include about this post being a personal opinion #define being bogus to avoid unwanted junk EMail ###### From: "Michael F. Coyle" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 19 Jan 1998 02:24:06 GMT Organization: Erol's Internet Services Lines: 75 Message-ID: <01bd247c$b5930de0$bcc3accf@mfc> References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> NNTP-Posting-Host: smt-as1s61.erols.com X-Trace: winter.news.erols.com 885176646 1308 207.172.195.188 (19 Jan 1998 02:24:06 GMT) X-Complaints-To: abuse@erols.com X-Newsreader: Microsoft Internet News 4.70.1161 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!204.186.110.126.MISMATCH!ptdnetP!newsgate.ptd.net!fastnet!howland.erols.net!winter.news.erols.com!not-for-mail Ben Hutchings wrote in article <69s7qb$df2$1@max89.public.ox.ac.uk>... > In article <01bd23b5$70471000$a6c3accf@mfc>, > Michael F. Coyle wrote: > >Ben Hutchings wrote in article > ><69m7t5$3el$1@max93.public.ox.ac.uk>... > > >> Nevertheless, 8-bit machines have been made with significantly more > >> than 64K RAM, and I don't think Intel's segmented memory model was > >> significantly less ugly than their paging systems. > > > >True. But note: 8088 memory management is not really segmentation -- > >it is a kind of mapping barely more sophisticated than bank > >switching. The 286 had genuine segmentation, but with a 64K limit. > >The 386 removed this limitation. > > > >Segmentation per se is not a bad thing -- but segments with a small > >maximum size creates headaches for the poor programmer. > > Well, I was referring to 8088 "segmentation" here of course. > > >> >In an interview in Byte, Don Estridge said that IBM was looking > >> >for a ten-year life for the PC architecture. No way could it have > >> >been done with an 8-bit CPU. > >> > >> That's interesting to hear, given that they managed to re-do most of > >> the architecture (and get it wrong again) with the AT after somewhat > >> less than 10 years, and that Intel had to introduce 2 completely new > >> memory models for the x86 in that time. > > > >I don't think he meant they would be selling 8088 machines for 10 > >years -- rather, that it would be possible to develop follow-on > >machines with more speed, more memory, more peripherals, and the > >ability to run existing software. In this sense, the PC-AT and later > >386-based PCs, though not elegant, show that it was possible. I > >don't think that any 8-bit CPU would have allowed for this kind of > >evolution over the 17+ years that the PC architecture has been > >around. (Which is not to say that some other 16- or 32-bit CPU > >couldn't have done a better job.) > > But the 286 and 386 introduced entirely new operating modes and memory > models, with the 8088 model highly deprecated (though DOS kept it in > use for a long time). I don't see the 8088 as an architecture that > has lasted 20 years, but one that has been supported for 20 years > purely for backwards compatibility. True. Backwards compatibility is important to customers. Any scheme, either Intel/Microsoft's hardware compatibility or Apple's software emulation (on the Power PC) that gets the job done is ok. > The preferred operating modes of > the i386 architecture, which was introduced in 1985, seem to have > almost nothing to do with the original 8088 architecture. I don't > suppose starting off with an 8-bit architecture would have made the > transitions to a 32-bit architecture significantly more painful. I don't think so either. Sigh. > The 680x0 series kept the same user register model and the same memory > model from 1979 right up to the 68060 in 1994(?) at which point it > stopped. Motorola didn't see the point in piling more money into it > when their main high-end user (Apple) was happy to switch to PowerPC. > If they had the money Intel has to spend on processors they could have > kept on extending and accelerating it in the same way Intel has. True, but the original 68000 didn't have any kind of memory management -- this was added later by an external chip (68450?). The '010 added virtualization, the '020 added memory management on board. So like the 8086, the 68K has been extended throughout its lifetime, though the 68K evolution has been *much* less traumatic for its customers. That just about covers it. Did we leave anything out? - Michael ###### From: "Michael F. Coyle" Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 19 Jan 1998 02:24:08 GMT Organization: Erol's Internet Services Lines: 60 Message-ID: <01bd2480$e6d3dca0$bcc3accf@mfc> References: <348fa205.3834263@news.compuserve.com> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> NNTP-Posting-Host: smt-as1s61.erols.com X-Trace: winter.news.erols.com 885176648 1308 207.172.195.188 (19 Jan 1998 02:24:08 GMT) X-Complaints-To: abuse@erols.com X-Newsreader: Microsoft Internet News 4.70.1161 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!howland.erols.net!winter.news.erols.com!not-for-mail "Paul E. Bennett" wrote in article <885040944snz@transcontech.co.uk>... > In article <69m7t5$3el$1@max93.public.ox.ac.uk> > worc0223@sable.ox.ac.uk "Ben Hutchings" writes: > > > In article <01bd21f7$848e5100$8cc3accf@mfc>, > > Michael F. Coyle wrote: > > >"Paul E. Bennett" wrote in article > > ><884810146snz@transcontech.co.uk>... > > > > > >> I would have rated the 6809 as the better choice, then being a Forthist I > > >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 > > >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus > > >> was 8 bits but that is not always the performance bottleneck. > > > > > >Can't agree with you here. Although the 6809 was, arguably, the most > > >elegant 8-bit micro around, it was *still* an 8-bitter with a 64K > > >address space. Nifty instructions and addressing modes aren't much > > >help if you're trying to address more memory than the processor can > > >handle. > > If you need the extra RAM Motorola had the MMU to cope with significantly > larger amounts of memory. That too was elegant. Whilst 6809's databus was > 8 bit there were several registers and the ALU that were 16 bit and could > perform 16 bit operations. You're missing my point. *Any* kludge to extend a 64KB address space is still a kludge -- a large linear address space is much easier to use when apps want to keep (say) a couple of megabytes of data in RAM. (BTW, the 6809 MMU was the 6829; it could address up to 2MB of RAM and divided it into 2KB pages. I doubt that having a user program manipulate 2KB pages is any easier than manipulating the 8088 "segment" registers.) > > >In an interview in Byte, Don Estridge said that IBM was looking for a ten- > > >year > > >life for the PC architecture. No way could it have been done with an 8-bit > > >CPU. > > > > That's interesting to hear, given that they managed to re-do most of > > the architecture (and get it wrong again) with the AT after somewhat > > less than 10 years, and that Intel had to introduce 2 completely new > > memory models for the x86 in that time. > > To get long life in electronic systems these days you have to have a sound > structure to begin building with and one that will not cripple you in the > long run. The embedded industrial control market seems to have been very > happy with the progress from 6809/6309 to 68008 to 68000 .... 68040 etc. > in much the same time period. But the 6800 ---> 68000 migration forgoes binary (object) code compatibility, as well as assembly source compatibility. This may be OK in the embedded world, but not in the mass-market PC world where millions of customers would have to throw away their existing software. - Michael ###### From: gmiller@inca.co.nz (Gaven Miller) Subject: Re: Why IBM chose Intel Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Followup-To: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> X-Newsreader: TIN [version 1.2 PL2] NNTP-Posting-Host: inca.co.nz Message-ID: <34c2d897.0@comnet.co.nz> Date: 19 Jan 98 04:37:43 GMT Lines: 28 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!news-feed.inet.tele.dk!bofh.vszbr.cz!meganews.telstra.net!news.telstra.net.nz!comnet.co.nz!inca01!gmiller Michael F. Coyle (mjcoyle@erols.com) wrote in alt.folklore.computers: [snip] > True, but the original 68000 didn't have any kind of memory management > -- this was added later by an external chip (68450?). The MMU was the 68451, but you needed an '010 - the 68k would do an F-line exception if one tried to access a coprocessor, because the 68k lacked "talk to coprocessor" brains. > The '010 added virtualization, the '020 added memory management on > board. So like the 8086, the 68K has been extended throughout its > lifetime, though the 68K evolution has been *much* less traumatic for > its customers. The 68020 lacked an on-board MMU. The external 68851 MMU chip made up for this. The 68030 was essentially the merging of the 68020 and 68851. > That just about covers it. Did we leave anything out? -- All email sent to my inca address will fail, however I can now be contacted via an intermediary : gem at tos pl net. I would like to apologise to the genuine respondents that this may inconvenience. ###### From: Robert Billing Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 19 Jan 98 07:31:30 GMT Message-ID: <885195090snz@tnglwood.demon.co.uk> References: <348fa205.3834263@news.compuserve.com> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> <01bd2480$e6d3dca0$bcc3accf@mfc> <69um37$8tc@access4.digex.net> Reply-To: unclebob@tnglwood.demon.co.uk X-Mail2News-User: unclebob@tnglwood.demon.co.uk X-Mail2News-Path: post-20.mail.demon.net!post.mail.demon.net!tnglwood.demon.co.uk X-Trace: mail2news.demon.co.uk 885230978 10000 unclebob tnglwood.demon.co.uk X-Complaints-To: abuse@demon.net X-Newsreader: Demon Internet Simple News v1.29 Lines: 22 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!bullseye.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!tnglwood.demon.co.uk!unclebob In article <69um37$8tc@access4.digex.net> jdc@access4.digex.net "John Cochran" writes: > So? Are you implying that the migration from 8080 -> 8086 was object code > compatable? The 68000 has kept the same user model for all versions. From The one thing that nobody has commented on in this thread is that some 8080 family processors had block move and compare instructions which, in the late 70's, were touted by Intel as being desirable or vital for business computing, something which IIRC no member of the 68k family has. I remember going to a presentation by Intel in about 79 or 80 at which this was extensively described. -- I am Robert Billing, Christian, inventor, traveller, cook and animal lover, I live near 0:46W 51:22N. http://www.tnglwood.demon.co.uk/ "Bother," said Pooh, "Eeyore, ready two photon torpedoes and lock phasers on the Heffalump, Piglet, meet me in transporter room three" ###### From: Tom Stepleton Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 19 Jan 1998 17:57:40 -0600 Organization: Washington University in St. Louis Lines: 26 Message-ID: <34C3E874.7DA0C3DA@rototiller.wustl.edu> References: <348fa205.3834263@news.compuserve.com> <69dnfv$re4$1@sue.cc.uregina.ca> <69oao9$9f8@nnrp1.farm.idt.net> <69saup$1398$1@nntp6.u.washington.edu> <67ngtm49.fsf@chonsp.franklin.lugs.ch> NNTP-Posting-Host: as1-isdn-110.wustl.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01 (X11; I; Linux 2.0.33 i586) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!arclight.uoregon.edu!newsfeed.internetmci.com!206.229.87.25!news-peer.sprintlink.net!news-backup-west.sprintlink.net!news-in-west.sprintlink.net!news.sprintlink.net!Sprint!199.125.85.9!news.mv.net!newspump.wustl.edu!newsreader.wustl.edu!not-for-mail Neil.Franklin.remove.this@ccw.ch wrote: > >Actually with a time machine I would take the entire Linux system > >source and fly back and jump-start the free software movement. (Later) > OK, I would be doing the adapting to whatever hardware I could get > in 1980. But Linux porters don't rename it to Porterux :-) ^^^^ It would be interesting to see you ungzip/untar that 6 MB source code archive on someone's TRS-80. Clearly you would have to look into places with rarely used minicomputers. (Maybe that PDP-7 at Bellcore is empty again... :) Seriously, what kinds of computers would our intrepid time-traveller want to use? Do we have any of those M68k UNIX workstations around yet? Perhaps something from Xerox? Isn't there already an effort to make a VAX Linux port (or was that just uVAX)? APPLE ][ LILO UNCOMPRESSING LINUX... Tom -- s/rototiller/artsci in my address to e-mail me ###### From: dhansen@btree.com (Dave Hansen) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 19 Jan 1998 18:15:53 GMT Organization: B-Tree Systems, Inc. Lines: 44 Message-ID: <34c38fc7.9133484@192.168.2.34> References: <348fa205.3834263@news.compuserve.com> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> <01bd2480$e6d3dca0$bcc3accf@mfc> <69um37$8tc@access4.digex.net> NNTP-Posting-Host: steggy.minn.net X-Newsreader: Forte Free Agent 1.1/32.230 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newspeer.monmouth.com!logbridge.uoregon.edu!news1.chicago.iagnet.net!iagnet.net!chippy.visi.com!news-out.visi.com!news.minn.net!steggy.minn.net!fire.btree.com Posted and emailed. On 18 Jan 1998 23:50:47 -0500, jdc@access4.digex.net (John Cochran) wrote: >In article <01bd2480$e6d3dca0$bcc3accf@mfc>, >Michael F. Coyle wrote: > >[snip...] > >>But the 6800 ---> 68000 migration forgoes binary (object) code compatibility, >>as well as assembly source compatibility. This may be OK in the embedded >>world, but not in the mass-market PC world where millions of customers would >>have to throw away their existing software. >> >>- Michael > >So? Are you implying that the migration from 8080 -> 8086 was object code >compatable? The 68000 has kept the same user model for all versions. From Earlier posts in this thread were claiming that the 6809 would have been a better choice than the 8088. I think he was trying to argue against that choice, not the 68000. If IBM _had_ chosen the the 6809, Moto might have been motivated to provide a 16-bit (and then 32-bit) upgrade path from that chip that _was_ object-code compatible. And then we'd be cursing IBM, Moto, and Microsoft for the original design and the bonehead kludges made to get _that_ working. But I expect that such a product would probably been approximately as successful as the 80x86 is today. Never underestimate the power of an installed base. A technically superior product will never win against a well-established one, unless there is an obvious, overriding advantage to the consumer. "Crashing less often" is not such an advantage. Neither is "less nonsensical architecture." "Runs all my current applications faster" isn't even, unless the consumer was going to upgrade in the first place, and the new solution actually costs _less_. Regards, -=Dave dhansen@btree.com Just my (10 - 010) cents worth. I can barely speak for myself, so I certainly can't speak for B-Tree. ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 19 Jan 1998 18:22:14 +0100 Organization: My own Private Self Lines: 22 Message-ID: <67ngtm49.fsf@chonsp.franklin.lugs.ch> References: <348fa205.3834263@news.compuserve.com> <69dnfv$re4$1@sue.cc.uregina.ca> <69oao9$9f8@nnrp1.farm.idt.net> <69saup$1398$1@nntp6.u.washington.edu> X-Newsreader: Gnus v5.3/Emacs 19.34 In article , wrote: >Actually with a time machine I would take the entire Linux system >source and fly back and jump-start the free software movement. dpeschel@u.washington.edu (D. Peschel) joked: >You mean the entire Neilux system source? I don't think it would be called >Linux under those circumstances. LOL. Great Joke. No. I would call it Linux, in honour of its writer. I would after all only be the transport vector for an existing software written in the 1990s, not the author in 1980. OK, I would be doing the adapting to whatever hardware I could get in 1980. But Linux porters don't rename it to Porterux :-) -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: albaugh@agames.com (Mike Albaugh) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded,alt.folklore.computers Date: 19 Jan 1998 21:36:05 GMT Organization: Atari Games Corporation Lines: 36 Message-ID: <6a0h05$mrk$1@void.agames.com> References: <348fa205.3834263@news.compuserve.com> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> <01bd2480$e6d3dca0$bcc3accf@mfc> <69um37$8tc@access4.digex.net> <885195090snz@tnglwood.demon.co.uk> NNTP-Posting-Host: java.agames.com X-Newsreader: TIN [version 1.2 PL2] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!news-peer.gip.net!news.gsl.net!gip.net!newshub1.home.com!news.home.com!newshub1-work.home.com!void.agames.com!albaugh Robert Billing (unclebob@tnglwood.demon.co.uk) wrote: : In article <69um37$8tc@access4.digex.net> : jdc@access4.digex.net "John Cochran" writes: : > So? Are you implying that the migration from 8080 -> 8086 was object code : > compatable? The 68000 has kept the same user model for all versions. From : The one thing that nobody has commented on in this thread is that some : 8080 family processors had block move and compare instructions which, : in the late 70's, were touted by Intel as being desirable or vital for : business computing, something which IIRC no member of the 68k family : has. I suppose it depends on your interpretation. I dunno about the x86s, but many of those "block" instructions on the Z80 were slower than the equivalent software approach without them (_especially_ the nybble-oriented ones). The 68010 introduced "loop mode" for certain two-instructions sequences, e.g. the guts of strcpy() :-) When in loop-mode, although you wrote two instructions (a MOVE and a DBCC), the effect, viewed from the outside, was essentially that of a block-move. Note that, again, a well-crafted set of MOVEM's would beat that "instruction" for real (say, page-size) block moves. I suspect that it would also beat the x86 equivalent. And then came the 68020, with a (tiny, but big enough for some important parts of Dhrystone) Icache, and the whole issue started sliding toward "moot" :-) : I remember going to a presentation by Intel in about 79 or 80 at which : this was extensively described. Depending on a vendor's slides for technical enlightenment is pretty much always a bad idea.... Mike | albaugh@agames.com, speaking only for myself ###### From: spam@lisardrock.demon.co.uk Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 19 Jan 1998 23:59:51 GMT Message-ID: <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> References: <01bd2480$e6d3dca0$bcc3accf@mfc> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 38 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!bullseye.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-19 mjcoyle@erols.com said: -You're missing my point. *Any* kludge to extend a 64KB address -space is still a kludge -- a large linear address space is much -easier to use when apps want to keep (say) a couple of megabytes of -data in RAM. true enough - but in the days when these processors were designed, nobody did things that way. the pdp11 was the order of the day, and it did things using paging too (eight 8k pages iwrc). nobody wrote programs that thought of themselves as larger than 64k. -(BTW, the 6809 MMU was the 6829; it could address up to 2MB of RAM -and divided it into 2KB pages. I doubt that having a user program -manipulate 2KB pages is any easier than manipulating the 8088 -"segment" registers.) no, but having your os do it is probably a hell of a lot easier, because then it can be done transparently(-ish). also you could remap pages into any order, virtualise them, etc. - things you just can't do with segmentation (or at least, the perverted form the 8086 used). -But the 6800 ---> 68000 migration forgoes binary (object) code -compatibility, as well as assembly source compatibility. This may -be OK in the embedded world, but not in the mass-market PC world -where millions of customers would have to throw away their existing -software. funny, the 68000 --> powerPC migration forgoes the same thing, and it doesn't seem to have bothered either mac owners or amiga fans overly. and it isn't as if the powerPC was designed for the purpose... -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: spam@lisardrock.demon.co.uk Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 19 Jan 1998 23:59:54 GMT Message-ID: <885254394.5398.6.nnrp-01.9e98ee68@news.demon.co.uk> References: <885195090snz@tnglwood.demon.co.uk> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 48 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!nntprelay.mathworks.com!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-19 unclebob@tnglwood.demon.co.uk said: -The one thing that nobody has commented on in this thread is that -some 8080 family processors had block move and compare instructions -which, in the late 70's, were touted by Intel as being desirable or -vital for business computing, something which IIRC no member of the -68k family has. indeed it did; unfortunately, they weren't general enough to be much use outside the confines of block moving and searching. the z80 suffers from this too. also the 8086 block moves were limited to 64k at a time. the 68000, on the other hand, took the pdp11 approach of "we've given you the bits, now it's up to you" - it had auto-inc addressing modes (as did the 6809) which were a more general solution to the same problem. the only problem was that the prefetch queue got in the way, and the 68010 fixed that (unfortunately, too late for all the 68k code which had been written as: start: move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ move.w @As+, @Ad+ bra start because the 68010 would only pick up on start: move.w @As+, @Ad+ bra start *sigh*) nowadays, the MOVS instructions are regarded as more of a hindrance than a help, especially on the 486+ where the simpler instructions do it faster... -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: Ben Hutchings Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: 20 Jan 1998 00:09:04 -0000 Organization: Not organised Lines: 37 Sender: womble@max77.public.ox.ac.uk Message-ID: <6a0pv0$19a$1@max77.public.ox.ac.uk> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> NNTP-Posting-Host: max77.public.ox.ac.uk Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!nntprelay.mathworks.com!newsfeed.eerie.fr!news-raspail.gip.net!news-lond.gip.net!news.gsl.net!gip.net!baron.netcom.net.uk!netcom.net.uk!server3.netnews.ja.net!news.ox.ac.uk!not-for-mail In article <34c2d897.0@comnet.co.nz>, Gaven Miller wrote: >Michael F. Coyle (mjcoyle@erols.com) wrote in alt.folklore.computers: >[snip] >> True, but the original 68000 didn't have any kind of memory management >> -- this was added later by an external chip (68450?). > >The MMU was the 68451, but you needed an '010 - the 68k would do an F-line >exception if one tried to access a coprocessor, because the 68k lacked >"talk to coprocessor" brains. I never heard of there being a co-processor protocol for the 68010. However, this was not the reason why virtualisation was impossible on the 68000. The real problem was that instructions that failed with exception 2 (access error) on the 68000 could neither be restarted or continued, so any external MMU could protect memory but could not support swapping. (The fact that user programs could read the real status register was an additional smaller problem.) >> The '010 added virtualization, the '020 added memory management on >> board. So like the 8086, the 68K has been extended throughout its >> lifetime, though the 68K evolution has been *much* less traumatic for >> its customers. > >The 68020 lacked an on-board MMU. > >The external 68851 MMU chip made up for this. > >The 68030 was essentially the merging of the 68020 and 68851. It also added a 256-byte data cache. -- Ben Hutchings, M&CS student | Jay Miner Society website: http://www.jms.org/ email/finger m95bwh@ecs.ox.ac.uk | homepage http://users.ox.ac.uk/~worc0223/ friends: People who know you well, but like you anyway. ###### From: TT Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 1998 00:31:59 -0500 Organization: self Lines: 28 Message-ID: <34C436CF.ED091C40@mediaone.net> References: <348fa205.3834263@news.compuserve.com> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> <01bd2480$e6d3dca0$bcc3accf@mfc> <69um37$8tc@access4.digex.net> Reply-To: tt0001@usa.net NNTP-Posting-Host: sloopy.ne.mediaone.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.04 [en] (Win95; U) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!nntprelay.mathworks.com!newsfeed.internetmci.com!24.128.1.125!chnws03.mediaone.net!24.128.1.107!chnws04.ne.mediaone.net!not-for-mail From what I can recall of programming my Father's Radioshak CoCo :) in assy, and my 68000 lab course, the only backward compatibility from 68xx -> 68xxx was you got to keep some s*** slow peripherals like a PIT and UART and gluelessly attach them to your 68k in a special mode. The architecture of the 68k was entirely different (better). John Cochran wrote: > > In article <01bd2480$e6d3dca0$bcc3accf@mfc>, > Michael F. Coyle wrote: > > [snip...] > > >But the 6800 ---> 68000 migration forgoes binary (object) code compatibility, > >as well as assembly source compatibility. This may be OK in the embedded > >world, but not in the mass-market PC world where millions of customers would > >have to throw away their existing software. > > > >- Michael > > So? Are you implying that the migration from 8080 -> 8086 was object code > compatable? The 68000 has kept the same user model for all versions. From ============================================================= Tim Tait No SPAM! Note that reply-to address is temporary and will expire eventually. Real Email: ttait at above domain ============================================================= ###### From: z80@ds.com (Peter) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 1998 09:04:58 GMT Organization: x Lines: 17 Message-ID: <34c66210.231590990@news.netcomuk.co.uk> References: <348fa205.3834263@news.compuserve.com> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> <01bd2480$e6d3dca0$bcc3accf@mfc> <69um37$8tc@access4.digex.net> <885195090snz@tnglwood.demon.co.uk> NNTP-Posting-Host: dialup-20-60.netcomuk.co.uk Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: taliesin.netcom.net.uk 885287041 4214 194.42.233.60 (20 Jan 1998 09:04:01 GMT) X-Complaints-To: abuse@corp.netcom.net.uk X-Newsreader: Forte Agent 1.5/32.451 X-No-Archive: yes Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!newsxfer3.itd.umich.edu!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!baron.netcom.net.uk!netcom.net.uk!knife.netcom.net.uk!netcom.net.uk!not-for-mail > The one thing that nobody has commented on in this thread is that some >8080 family processors had block move and compare instructions which, >in the late 70's, were touted by Intel as being desirable or vital for >business computing, something which IIRC no member of the 68k family >has. Not the 8080. The Z80 has the LDIR, CPIR etc. These can be extremely useful in optimising certain algorithms. Peter. Return address is invalid to help stop junk mail. E-mail replies to z80@digiXYZserve.com but remove the XYZ. ###### From: z80@ds.com (Peter) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 1998 09:04:59 GMT Organization: x Lines: 24 Message-ID: <34c76295.231723831@news.netcomuk.co.uk> References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> NNTP-Posting-Host: dialup-20-60.netcomuk.co.uk Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: taliesin.netcom.net.uk 885287042 4214 194.42.233.60 (20 Jan 1998 09:04:02 GMT) X-Complaints-To: abuse@corp.netcom.net.uk X-Newsreader: Forte Agent 1.5/32.451 X-No-Archive: yes Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!newsxfer3.itd.umich.edu!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!baron.netcom.net.uk!netcom.net.uk!knife.netcom.net.uk!netcom.net.uk!not-for-mail >funny, the 68000 --> powerPC migration forgoes the same thing, and it >doesn't seem to have bothered either mac owners or amiga fans overly. Probably because there aren't very many of them :) Also, most Mac users tend to be running a very small range of apps (compared with most PC users) and provided that *those* apps become available in PPC binary versions, these people won't be bothered. I don't want to start a Mac/PC flame war, please. I just deal with Mac (UK) users on a daily basis, and this is self-evident. Regarding the IBM PC, despite the crap put out in the PC magazines, many people are sticking to old trusted applications, mostly win16 but very often DOS. This is why Intel's hopes that 32-bit will take over have been repeatedly squashed. Peter. Return address is invalid to help stop junk mail. E-mail replies to z80@digiXYZserve.com but remove the XYZ. ###### From: mojaveg@ridgecrest.ca.us (Everett M. Greene) Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 98 15:49:24 PST Organization: none that you'd notice Lines: 33 Message-ID: <19980120.799BD40.E33E@mojaveg.ridgecrest.ca.us> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> NNTP-Posting-Host: annex178.ridgecrest.ca.us Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-NewsSoftware: GRn 2.1 Feb 19, 1994 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!news-peer.gip.net!news.gsl.net!gip.net!newsfeed.internetmci.com!199.120.150.5!news.ridgecrest.ca.us!not-for-mail In article <6a0pv0$19a$1@max77.public.ox.ac.uk> Ben Hutchings writes: > In article <34c2d897.0@comnet.co.nz>, Gaven Miller wrote: > >Michael F. Coyle (mjcoyle@erols.com) wrote in alt.folklore.computers: > >[snip] > >> True, but the original 68000 didn't have any kind of memory management > >> -- this was added later by an external chip (68450?). > > > >The MMU was the 68451, but you needed an '010 - the 68k would do an F-line > >exception if one tried to access a coprocessor, because the 68k lacked > >"talk to coprocessor" brains. > > I never heard of there being a co-processor protocol for the 68010. > However, this was not the reason why virtualisation was impossible on > the 68000. The real problem was that instructions that failed with > exception 2 (access error) on the 68000 could neither be restarted or > continued, so any external MMU could protect memory but could not > support swapping. (The fact that user programs could read the real > status register was an additional smaller problem.) > > >> The '010 added virtualization, the '020 added memory management on > >> board. So like the 8086, the 68K has been extended throughout its > >> lifetime, though the 68K evolution has been *much* less traumatic for > >> its customers. > > > >The 68020 lacked an on-board MMU. > > > >The external 68851 MMU chip made up for this. > > > >The 68030 was essentially the merging of the 68020 and 68851. > > It also added a 256-byte data cache. There's both instruction and data cache (1K each?). ###### From: tph@longhorn.uucp (Tom Harrington) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded,alt.folklore.computers Date: 20 Jan 1998 16:47:19 GMT Organization: Mechanist Industries Lines: 32 Message-ID: <6a2keo$der3@eccws1.dearborn.ford.com> References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> <34c76295.231723831@news.netcomuk.co.uk> Reply-To: tph@rmi.net NNTP-Posting-Host: cs0053.eld.ford.com X-Newsreader: TIN [version 1.2 PL2] Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!newsxfer3.itd.umich.edu!jobone!dailyplanet.srl.ford.com!eccws1.dearborn.ford.com!longhorn!tph Peter (z80@ds.com) wrote: : >funny, the 68000 --> powerPC migration forgoes the same thing, and it : >doesn't seem to have bothered either mac owners or amiga fans overly. : Probably because there aren't very many of them :) : Also, most Mac users tend to be running a very small range of apps : (compared with most PC users) and provided that *those* apps become : available in PPC binary versions, these people won't be bothered. That's not the real reason that that binary non-compatibility between the 68k and the PPC didn't bother Mac owners. The real reason is that when Apple switched to the PPC, they added full 68020 emulation to the the Mac ROMs. They had to; significant portions of the operating system were (and are) written in 68k assembly. The effect of this is that almost every Mac 68k binary will run on a PPC Mac without recompiling, so the end user gets the illusion that the chips _are_ binary compatible. Not everyone, of course; the fastest of the 68040 Macs could beat the slower PPC systems when running 68k code. And there was no FPU emulation (although a software FPU emulator was available from a third-party developer). But in the time since the change, most apps have migrated to either pure PPC or "fat" binary files with both executables. I'm not sure how this affected Amiga users. -- Tom Harrington --------- tph@rmii.com -------- http://rainbow.rmii.com/~tph WARNING: Poster is not responsible for damage to fragile world-views or beliefs. Proceed at your own risk. Cookie's Revenge: ftp://ftp.rmi.net/pub2/tph/cookie/cookies-revenge.sit.hqx ###### From: spam@lisardrock.demon.co.uk Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 1998 19:30:05 GMT Message-ID: <885324605.20565.0.nnrp-04.9e98ee68@news.demon.co.uk> References: <885254394.5398.6.nnrp-01.9e98ee68@news.demon.co.uk> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 16 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail -because the 68010 would only pick up on -start: -move.w @As+, @Ad+ -bra start we said. oops... :%s/bra /dbcc / -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: Ben Hutchings Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 20 Jan 1998 19:57:47 -0000 Organization: Not organised Lines: 23 Sender: womble@max85.public.ox.ac.uk Message-ID: <6a2vjr$3hv$1@max85.public.ox.ac.uk> References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> <34c76295.231723831@news.netcomuk.co.uk> <6a2keo$der3@eccws1.dearborn.ford.com> NNTP-Posting-Host: max85.public.ox.ac.uk Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!feeder.qis.net!newsxfer.visi.net!nntp.news.xara.net!xara.net!cableinet-uk!news.cableinet.co.uk!baron.netcom.net.uk!netcom.net.uk!server3.netnews.ja.net!news.ox.ac.uk!not-for-mail In article <6a2keo$der3@eccws1.dearborn.ford.com>, Tom Harrington wrote: >Peter (z80@ds.com) wrote: > >: >funny, the 68000 --> powerPC migration forgoes the same thing, and it >: >doesn't seem to have bothered either mac owners or amiga fans overly. > >: Probably because there aren't very many of them :) >I'm not sure how this affected Amiga users. Don't ask. It's rather complicated at the moment. The German company Phase 5 made some Amiga accelerator boards that take a 680x0 and a PowerPC processor, plus some software to schedule tasks and set up the MMU on the PPC, to synchronise the two processors, and to effect calls from one to the other. There is no sign of a port of the AmigaOS yet, though; nor is it yet sure that any port would be to the PowerPC. -- Ben Hutchings, M&CS student | Jay Miner Society website: http://www.jms.org/ email/finger m95bwh@ecs.ox.ac.uk | homepage http://users.ox.ac.uk/~worc0223/ Design a system any fool can use, and only a fool will want to use it. ###### From: spam@lisardrock.demon.co.uk Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 20 Jan 1998 22:51:39 GMT Message-ID: <885336699.6530.8.nnrp-01.9e98ee68@news.demon.co.uk> References: <34c76295.231723831@news.netcomuk.co.uk> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 41 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!news-xfer.siscom.net!streamer1.cleveland.iagnet.net!qual.net!iagnet.net!nntp.news.xara.net!xara.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-20 z80@ds.com(Peter) said: ->funny, the 68000 --> powerPC migration forgoes the same thing, and ->it doesn't seem to have bothered either mac owners or amiga fans ->overly. -Probably because there aren't very many of them :) :> well all right, but granted. (besides, when you consider the sheer size of the market these days, 10% of it is still no small share in absolute terms.) -Also, most Mac users tend to be running a very small range of apps -(compared with most PC users) and provided that *those* apps become -available in PPC binary versions, these people won't be bothered. that may have more to do with it - though the amiga might be a very different matter. however... considering that some versions of the *os* that people run on their powerpcs is still emulated 68k code, it's not really that big an issue. ok, so translator technology and code caching has improved a helluva lot since 1981... but still, jumping binary architecture boundaries isn't that big a deal these days. btw. congrats to tao on his new baby. :> -Regarding the IBM PC, despite the crap put out in the PC magazines, -many people are sticking to old trusted applications, mostly win16 -but very often DOS. This is why Intel's hopes that 32-bit will take -over have been repeatedly squashed. we know. we're one of them. gimme dos any day (until we recompile tripos and start using that instead :> ) -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 21 Jan 1998 04:25:03 +0100 Organization: My own Private Self Lines: 70 Message-ID: References: <348fa205.3834263@news.compuserve.com> <69dnfv$re4$1@sue.cc.uregina.ca> <69oao9$9f8@nnrp1.farm.idt.net> <69saup$1398$1@nntp6.u.washington.edu> <67ngtm49.fsf@chonsp.franklin.lugs.ch> <34C3E874.7DA0C3DA@rototiller.wustl.edu> X-Newsreader: Gnus v5.3/Emacs 19.34 Neil.Franklin.remove.this@ccw.ch wrote: > >Actually with a time machine I would take the entire Linux system > >source and fly back and jump-start the free software movement. (Later) > OK, I would be doing the adapting to whatever hardware I could get > in 1980. But Linux porters don't rename it to Porterux :-) Tom Stepleton mused on this: > ^^^^ >It would be interesting to see you ungzip/untar that 6 MB source code >archive on someone's TRS-80. Error: Disk Full. That is assuming they had real error messages on them. BTW: You are kidding with that 6MB: chonsp:ttyp0:root:~/# du -s /usr/src/kernel-source-2.0.27/ 31547 /usr/src/kernel-source-2.0.27 chonsp:ttyp0:root:~/# > Clearly you would have to look into >places with rarely used minicomputers. I was thinking of a VAX, that came out in 1977 or 1978. With Bill Joy at UCB I would first knock on the door at CMU, I think. > (Maybe that PDP-7 at Bellcore is empty again... :) 18bit computer, somewhere around 1 MIPS, ca 32kword memory, no paging hardware? No, I wouldn't try that one. >Seriously, what kinds of computers would our intrepid time-traveller >want to use? Do we have any of those M68k UNIX workstations around yet? MC68000-5 came out somwhere 1977..79, so it would be around. Sun used it in their first workstations 1981 or 1982. Linux runs usably on an 386-16 (I have one of them around here, note: pre 386DX16 naming). The 4MB RAM and 100MB harddisk will be expensive though :-) >Perhaps something from Xerox? IIRC, they had 32bit at that time. But try getting a raw processor out of them, also their stuff was TTL and ECL so far I know. > Isn't there already an effort to make >a VAX Linux port (or was that just uVAX)? There is, some guy in Australia IIRC. I once had a glimpse at the FAQ for it, IIRC it was a 3xxx model he was using. > APPLE ][ >LILO UNCOMPRESSING LINUX... No uncompressing, please. You will be booting from an kernal floppy, then exchanging for a root floppy. And decompressing will take a lot of time on an 6502-1. The 386-16 wants 20 sec. For the 6501 would have to be ELKS (Embedded Linux Kernal Subset) that is (was? I read about it over 1 year ago) intended for 8bit processors, first 8088 (!), then later to be Z80 when decided on an MMU. -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 21 Jan 1998 04:37:37 +0100 Organization: My own Private Self Lines: 29 Message-ID: <90saa45a.fsf@chonsp.franklin.lugs.ch> References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> X-Newsreader: Gnus v5.3/Emacs 19.34 On 1998-01-19 mjcoyle@erols.com said: -(BTW, the 6809 MMU was the 6829; it could address up to 2MB of RAM -and divided it into 2KB pages. I doubt that having a user program -manipulate 2KB pages is any easier than manipulating the 8088 -"segment" registers.) spam@lisardrock.demon.co.uk added: >no, but having your os do it is probably a hell of a lot easier, because >then it can be done transparently(-ish). also you could remap pages into >any order, virtualise them, etc. - things you just can't do with >segmentation (or at least, the perverted form the 8086 used). On the 8088 the OS can do it transparently also. Simply declare the segment registers tabu for the user software. With CS <> DS = SS = ES you could even do pdp11/70-like D+I adressing. Would make an ideal Unix Verson 6 or 2.xBSD machine. IIRC that is what Xenix did. The real flaws on the 8088 were its inability to: - trap to force the users to leave the segment registers alone. - swapping is awfull, on an pdp11 the OS swaps fixed sized pages, on the 8088 it must rearange variable sized segments. Load on demand is not possible either. OS/2 1.x demonstrated the trouble with this stuff. -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: Mark Borgerson Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 08:56:11 -0800 Organization: WetLabs Inc. Lines: 24 Message-ID: <34C628AB.6308@wetlabs.com> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> NNTP-Posting-Host: 143.227.0.65 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 2.0 (Win95; I; 16bit) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!ais.net!uunet!in3.uu.net!news.pdx.oneworld.com!news John West McKenna wrote: > > Ben Hutchings writes: > > >However, this was not the reason why virtualisation was impossible on > >the 68000. The real problem was that instructions that failed with > >exception 2 (access error) on the 68000 could neither be restarted or > >continued, so any external MMU could protect memory but could not > >support swapping. > > Of course, that didn't stop people doing virtualisation on the 68000. One > of the easiest ways was to have a second processor step in before the fault > happened and make sure the right memory would appear. > I think this was the way that Apple managed virtual memory on the Lisa-- some sort of second processor retained enough information to allow instruction restart. I've still got one up in the attic. I ought to pull the PCBs and check it out. It cost $5K at developer discount, but was the only way to develop Mac code. Strangely enough, my later Mac II and then Quadra systems also cost about $5K. Only in the last rounds with PPC machines has the cost come down to about $2K. Mark Borgerson ###### From: tsw@cagent.com (Tom Watson) Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 10:19:15 -0800 Organization: CagEnt, Inc. Lines: 31 Message-ID: References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> NNTP-Posting-Host: alfred.cagent.com Cache-Post-Path: alfred.cagent.com!unknown@cypher.cagent.com Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!xfer.kren.nm.kr!newshub1.home.com!news.home.com!news.rdc1.sfba.home.net!cypher.cagent.com!user In article <6a4qlb$8a1$1@enyo.uwa.edu.au>, john@ucc.gu.uwa.edu.au (John "West" McKenna) wrote: > Ben Hutchings writes: > > >However, this was not the reason why virtualisation was impossible on > >the 68000. The real problem was that instructions that failed with > >exception 2 (access error) on the 68000 could neither be restarted or > >continued, so any external MMU could protect memory but could not > >support swapping. > > Of course, that didn't stop people doing virtualisation on the 68000. One > of the easiest ways was to have a second processor step in before the fault > happened and make sure the right memory would appear. > The 'other' method was to use 'stack probes'. I know of two machines that do this. I have both of them. A stack probe just did a 'tst' (test) instruction a bit down the stack on every entry to a subroutine. If it required memory to be allocated, it did so, and restarted the instruction (which didn't alter memory). It is a quite reasonable thing to do, and works quite well. One of the machines had simple offset/limit registers for code/bss and stack, the other machine used 68451 mmu's (4 of them). The first ran Xenix, the second ran a System 3 version of Unix. All done around 1981. -- tsw@cagent.com (Home: tsw@johana.com) Please forward spam to: annagram@hr.house.gov (my Congressman), I do. ###### Date: 21 Jan 98 13:46:06 -0800 From: "Charlie Gibbs" Subject: Re: Why IBM chose Intel References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> <90saa45a.fsf@chonsp.franklin.lugs.ch> <6a5he6$2dq$1@news.iastate.edu> Message-ID: <519.325T307T8263148@sky.bus.com> Newsgroups: alt.folklore.computers Lines: 17 X-Newsreader: THOR 2.5a (Amiga;TCP/IP) NNTP-Posting-Host: 204.244.247.7 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news-backup-east.sprintlink.net!news-in-east.sprintlink.net!news.sprintlink.net!Sprint!204.244.4.2!news.westel.com!news.skybus.com!204.244.247.115 In article <6a5he6$2dq$1@news.iastate.edu> rhawkins@iastate.edu (Rick Hawkins) writes: >In article <90saa45a.fsf@chonsp.franklin.lugs.ch>, > wrote: > >>On the 8088 the OS can do it transparently also. Simply declare the >>segment registers tabu for the user software. > >A tabu which experience shows us that no software writers will try to >get around :) Not more than once, anyway. :-) -- cgibbs@sky.bus.com (Charlie Gibbs) Remove the first period after the "at" sign to reply. ###### From: R!ch Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 14:52:02 +0000 Lines: 31 Message-ID: References: <34c76295.231723831@news.netcomuk.co.uk> <885336699.6530.8.nnrp-01.9e98ee68@news.demon.co.uk> NNTP-Posting-Host: paddington.uk.sun.com Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Sender: richardt@paddington In-Reply-To: <885336699.6530.8.nnrp-01.9e98ee68@news.demon.co.uk> Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!news-feed.fnsi.net!news.maxwell.syr.edu!news-peer.gip.net!news.gsl.net!gip.net!news.idt.net!streamer1.cleveland.iagnet.net!qual.net!iagnet.net!btnet-peer!btnet-feed2!btnet!carbon.eu.sun.com!uk-usenet.uk.sun.com!paddington!richardt On Tue, 20 Jan 1998 spam@lisardrock.demon.co.uk wrote: > -Also, most Mac users tend to be running a very small range of apps > -(compared with most PC users) and provided that *those* apps become > -available in PPC binary versions, these people won't be bothered. > > that may have more to do with it - though the amiga might be a very > different matter. Hmm, what %age of PC users run nothing but an office automation suite (eg M$ Office)? Given this, I'd hardly say the PC lusers run a wide range of applications, as implied by the original poster. -- R!ch (Email is flakey at present: use richard.teer@keaton.uk.sun.com) If it ain't analogue, it ain't music. #include \\|// - ? (o o) /==================================oOOo=(_)=oOOo========\ | Richard Teer richard.teer@uk.sun.com | | | | | | WWW: www.rkdltd.demon.co.uk | | .oooO | | ( ) Oooo. | \===================================\ (==( )==========/ \_) ) / (_/ ###### From: btgsch@rmplc.co.uk (ric) Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 18:34:59 +0000 Organization: Those with six coloured blood Lines: 30 Message-ID: <1d37gwc.10x2px8dyn0gN@rm-dynf1-151.rmplc.co.uk> References: <348fa205.3834263@news.compuserve.com> <19980111061100.BAA01354@ladder01.news.aol.com> <34be044d.198495441@news.netcomuk.co.uk> <34ba11c3.8891415@news.compuserve.com> <69d5qt$dqb@examiner.concentric.net> <69dnfv$re4$1@sue.cc.uregina.ca> <69fve3$25q@nnrp1.farm.idt.net> <69gcv2$n5c$1@sue.cc.uregina.ca> <34BBE9CE.646126A1@pec.co.nz> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <19980120.799BD40.E33E@mojaveg.ridgecrest.ca.us> NNTP-Posting-Host: rm-dynf1-151.rmplc.co.uk X-Newsreader: MacSOUP 2.3 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!ais.net!newspeer.monmouth.com!Aladdin!aladdin.net!ns2.aladdin.net!RMplc!rmplc.co.uk!btgsch Everett M. Greene wrote: > In article <6a0pv0$19a$1@max77.public.ox.ac.uk> Ben Hutchings sable.ox.ac.uk> writes: > > In article <34c2d897.0@comnet.co.nz>, Gaven Miller > > wrote: > > >Michael F. Coyle (mjcoyle@erols.com) wrote in alt.folklore.computers: > > >[snip] > > >> The '010 added virtualization, the '020 added memory management on > > >> board. So like the 8086, the 68K has been extended throughout its > > >> lifetime, though the 68K evolution has been *much* less traumatic for > > >> its customers. > > > > > >The 68020 lacked an on-board MMU. > > > > > >The external 68851 MMU chip made up for this. > > > > > >The 68030 was essentially the merging of the 68020 and 68851. > > > > It also added a 256-byte data cache. > > There's both instruction and data cache (1K each?). The '020 had a 256 byte instruction cache, the '030 had a 256 byte data cache also. In a mac, there is precious little difference in benchmark between the two. -- 'ric ###### From: rhawkins@iastate.edu (Rick Hawkins) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 21 Jan 1998 19:14:14 GMT Organization: Iowa State University, Ames, Iowa USA Lines: 14 Message-ID: <6a5he6$2dq$1@news.iastate.edu> References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> <90saa45a.fsf@chonsp.franklin.lugs.ch> NNTP-Posting-Host: pv2086.vincent.iastate.edu Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!news1.chicago.iagnet.net!news2.chicago.iagnet.net!iagnet.net!newsxfer3.itd.umich.edu!newsxfer.itd.umich.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins In article <90saa45a.fsf@chonsp.franklin.lugs.ch>, wrote: >On the 8088 the OS can do it transparently also. Simply declare the >segment registers tabu for the user software. A tabu which experience shows us that no software writers will try to get around :) -- R E HAWKINS rhawkins@iastate.edu These opinions will not be those of ISU until they pay my retainer. ###### Newsgroups: comp.arch.embedded,alt.folklore.computers From: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Subject: Re: Why IBM chose Intel Message-ID: <1998Jan21.215554.4103@lorelei.approve.se> Originator: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson) Organization: [1] + 5934 done /bin/rm -rf ~/ & X-No-Archive: yes X-Newsreader: trn 4.0-test60 (5 October 1997) References: <348fa205.3834263@news.compuserve.com> <69um37$8tc@access4.digex.net> <885195090snz@tnglwood.demon.co.uk> <34c66210.231590990@news.netcomuk.co.uk> Date: Wed, 21 Jan 1998 21:55:54 GMT Lines: 17 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!Cabal.CESspool!bofh.vszbr.cz!newspeer.monmouth.com!news.algonet.se!fci-se!fci!newsfeed.sunet.se!news99.sunet.se!news01.sunet.se!192.71.180.34.MISMATCH!mn6.swip.net!seunet!mn4.swip.net!lorelei!not-for-mail In article <34c66210.231590990@news.netcomuk.co.uk>, Peter wrote: > Not the 8080. The Z80 has the LDIR, CPIR etc. > > These can be extremely useful in optimising certain algorithms. They are slow. One reason why they are slow is that the Z80 fetched the opcode for each repetition. The LDIR, for example, was just like the LDI but with an decrement of BC and an inhibit of the PC increment until BC = 0. Some claim this was done so the repeat instructions could be interrupted by interrupts, but to me it smells like a cheap kludge. -- Goran Larsson hoh AT approve DOT se I was an atheist, http://home1 DOT swipnet DOT se/%7Ew-12153/ until I found out I was God. ###### From: newcomer@flounder.com (Joseph M. Newcomer) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 22:19:13 GMT Organization: Pittsburgh OnLine, Inc. Lines: 66 Message-ID: <34e273c9.94911653@206.210.64.12> References: <348fa205.3834263@news.compuserve.com> <01bd208d$848db940$1bc4accf@mfc> <884810146snz@transcontech.co.uk> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <885040944snz@transcontech.co.uk> NNTP-Posting-Host: ppp42.s9.pgh.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.5/32.451 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.clark.net!europa.clark.net!205.252.116.205!howland.erols.net!fastnet!news.pgh.net!not-for-mail The story I heard was pretty simple. IBM wanted a chip for their PC. They looked at Motorola, Zilog, and Intel, all of whom were candidates. Zilog's chip didn't work, and the company didn't look like it was going to get it working in the time frame required. Motorola couldn't commit to the quantity that IBM wanted to buy. Intel said they could supply the requisite number of chips in the time frame required. Done deal. Technical considerations? Doesn't matter how good it is if you can't get it! joe On Sat, 17 Jan 98 12:42:24 GMT, peb@transcontech.co.uk ("Paul E. Bennett") wrote: >In article <69m7t5$3el$1@max93.public.ox.ac.uk> > worc0223@sable.ox.ac.uk "Ben Hutchings" writes: > >> In article <01bd21f7$848e5100$8cc3accf@mfc>, >> Michael F. Coyle wrote: >> >"Paul E. Bennett" wrote in article >> ><884810146snz@transcontech.co.uk>... >> > >> >> I would have rated the 6809 as the better choice, then being a Forthist I >> >> would (3 byte inner interpreter for Forth, two stacks, lovely). The 6809 >> >> has 16 bit ALU operations and a nicer feel to it. I know that the data bus >> >> was 8 bits but that is not always the performance bottleneck. >> > >> >Can't agree with you here. Although the 6809 was, arguably, the most >> >elegant 8-bit micro around, it was *still* an 8-bitter with a 64K >> >address space. Nifty instructions and addressing modes aren't much >> >help if you're trying to address more memory than the processor can >> >handle. > >If you need the extra RAM Motorola had the MMU to cope with significantly >larger amounts of memory. That too was elegant. Whilst 6809's databus was >8 bit there were several registers and the ALU that were 16 bit and could >perform 16 bit operations. > >> >In an interview in Byte, Don Estridge said that IBM was looking for a ten- >> >year >> >life for the PC architecture. No way could it have been done with an 8-bit >> >CPU. >> >> That's interesting to hear, given that they managed to re-do most of >> the architecture (and get it wrong again) with the AT after somewhat >> less than 10 years, and that Intel had to introduce 2 completely new >> memory models for the x86 in that time. > >To get long life in electronic systems these days you have to have a sound >structure to begin building with and one that will not cripple you in the >long run. The embedded industrial control market seems to have been very >happy with the progress from 6809/6309 to 68008 to 68000 .... 68040 etc. >in much the same time period. Joseph M. Newcomer newcomer@flounder.com http://www3.pgh.net/~newcomer ###### From: newcomer@flounder.com (Joseph M. Newcomer) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Wed, 21 Jan 1998 22:23:41 GMT Organization: Pittsburgh OnLine, Inc. Lines: 57 Message-ID: <34e3749c.95121920@206.210.64.12> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> NNTP-Posting-Host: ppp42.s9.pgh.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.5/32.451 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.ecn.uoknor.edu!feed1.news.erols.com!howland.erols.net!fastnet!news.pgh.net!not-for-mail The first Apollo workstation used two 68K processors. The way it worked was that if the first processor was going to take a page fault, the signal was intercepted, and the bus was paused (infinite wait-states were introduced). Meanwhile, the second processor work up, and brought the requisite page into memory. The first processor was then allowed to proceed, since it could now find the page and not have to fault. When Apollo came out with their workstations using the 68020, Motorola couldn't deliver the MMU in time, so Apollo came up with a kludge board that plugged into the MMU slot and emulated it, albeit slowly, and also covered one of the backplane slots. When Motorola delivered the MMU, it turned out that the kludge box had a slightly different bus spec and you couldn't actually plug the MMU into the MMU slot! I still have the box in the basement. joe On 20 Jan 1998 00:09:04 -0000, Ben Hutchings wrote: >In article <34c2d897.0@comnet.co.nz>, Gaven Miller wrote: >>Michael F. Coyle (mjcoyle@erols.com) wrote in alt.folklore.computers: >>[snip] >>> True, but the original 68000 didn't have any kind of memory management >>> -- this was added later by an external chip (68450?). >> >>The MMU was the 68451, but you needed an '010 - the 68k would do an F-line >>exception if one tried to access a coprocessor, because the 68k lacked >>"talk to coprocessor" brains. > >I never heard of there being a co-processor protocol for the 68010. >However, this was not the reason why virtualisation was impossible on >the 68000. The real problem was that instructions that failed with >exception 2 (access error) on the 68000 could neither be restarted or >continued, so any external MMU could protect memory but could not >support swapping. (The fact that user programs could read the real >status register was an additional smaller problem.) > >>> The '010 added virtualization, the '020 added memory management on >>> board. So like the 8086, the 68K has been extended throughout its >>> lifetime, though the 68K evolution has been *much* less traumatic for >>> its customers. >> >>The 68020 lacked an on-board MMU. >> >>The external 68851 MMU chip made up for this. >> >>The 68030 was essentially the merging of the 68020 and 68851. > >It also added a 256-byte data cache. > > Joseph M. Newcomer newcomer@flounder.com http://www3.pgh.net/~newcomer ###### Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k From: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Subject: Re: Why IBM chose Intel Message-ID: <1998Jan22.001610.5684@lorelei.approve.se> Originator: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson) Organization: [1] + 5934 done /bin/rm -rf ~/ & X-No-Archive: yes X-Newsreader: trn 4.0-test60 (5 October 1997) References: <348fa205.3834263@news.compuserve.com> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> Date: Thu, 22 Jan 1998 00:16:10 GMT Lines: 37 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!fci-se!fci!newsfeed.sunet.se!news99.sunet.se!news01.sunet.se!192.71.180.34.MISMATCH!mn6.swip.net!seunet!mn4.swip.net!lorelei!not-for-mail In article , Tom Watson wrote: > The 'other' method was to use 'stack probes'. I know of two machines that > do this. I have both of them. I have one as well. Luxor ABC-1600 (68008-based) running ABCenix. > A stack probe just did a 'tst' (test) instruction a bit down the stack on > every entry to a subroutine. If it required memory to be allocated, it > did so, and restarted the instruction (which didn't alter memory). It is There is no need to restart the probe instruction. Just fix the stack and continue at the instruction after the probe. > a quite reasonable thing to do, and works quite well. It caused BIG problems with the older versions of bourne shell and korn shell as they managed their memory by catching segmentation violations and brk()-ing for more memory. The instruction that caused the segmentation violations had problems restarting. The source code for korn shell had to be modified to include it's own stack probes... > One of the machines had simple offset/limit registers for code/bss and > stack, the other machine used 68451 mmu's (4 of them). The first ran The ABC machine uses a MMU built from fast RAM, PALs, and simple TTL gates. The MMU has 8 process contexts to reduce process switch time. Even though it only has 1MB RAM (512KB configuration available) it was a usefull machine for its time (some 15 years ago). -- Goran Larsson hoh AT approve DOT se I was an atheist, http://home1 DOT swipnet DOT se/%7Ew-12153/ until I found out I was God. ###### From: spam@lisardrock.demon.co.uk Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Thu, 22 Jan 1998 02:14:47 GMT Message-ID: <885435287.2103.5.nnrp-09.9e98ee68@news.demon.co.uk> References: <90saa45a.fsf@chonsp.franklin.lugs.ch> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 31 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-21 Neil.Franklin.remove.this@ccw.ch said: -On 1998-01-19 mjcoyle@erols.com said: --(BTW, the 6809 MMU was the 6829; it could address up to 2MB of RAM --and divided it into 2KB pages. I doubt that having a user program --manipulate 2KB pages is any easier than manipulating the 8088 --"segment" registers.) -spam@lisardrock.demon.co.uk added: ->no, but having your os do it is probably a hell of a lot easier, ->because then it can be done transparently(-ish). also you could ->remap pages into any order, virtualise them, etc. - things you ->just can't do with segmentation (or at least, the perverted form ->the 8086 used). -On the 8088 the OS can do it transparently also. Simply declare the -segment registers tabu for the user software. With CS <> DS = SS = -ES you could even do pdp11/70-like D+I adressing. Would make an -ideal Unix Verson 6 or 2.xBSD machine. IIRC that is what Xenix did. unfortunately, the 8088 didn't have protection levels (and iwrc, the 6809 *did*) which means you can't actually stop a user from doing this. minix does that too, but it relies on the user not cheating. also the 8088 can't reserve areas of memory for privileged use. -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: spam@lisardrock.demon.co.uk Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Thu, 22 Jan 1998 02:14:48 GMT Message-ID: <885435288.2103.6.nnrp-09.9e98ee68@news.demon.co.uk> References: NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 31 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!news-peer.gip.net!news-lond.gip.net!news.gsl.net!gip.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-21 Neil.Franklin.remove.this@ccw.ch said: ->Perhaps something from Xerox? -IIRC, they had 32bit at that time. But try getting a raw processor -out of them, also their stuff was TTL and ECL so far I know. xerox seemed to be pathologically committed to 16-bit. the alto was (since it was a nova clone, that's not surprising) but iwrc, so were the dorado and the dolphin. (course if they could make it work, it begs the question... :> ) -For the 6501 would have to be ELKS (Embedded Linux Kernal Subset) -that is (was? I read about it over 1 year ago) intended for 8bit -processors, first 8088 (!), then later to be Z80 when decided on an -MMU. hang on, guys - unix v6 was freely available up until at&t declared it commercial and said people couldn't use their source code any more. there's always been a "free unix" tradition. the only advantage of linux was that it was written to make unix available on a particular processor which hadn't supported it terribly well up until then. (also to get mr torvalds his degree. :> ) what about BSD? -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: z80@ds.com (Peter) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Thu, 22 Jan 1998 09:07:50 GMT Organization: x Lines: 56 Message-ID: <34c70502.404376823@news.netcomuk.co.uk> References: <348fa205.3834263@news.compuserve.com> <69um37$8tc@access4.digex.net> <885195090snz@tnglwood.demon.co.uk> <34c66210.231590990@news.netcomuk.co.uk> <1998Jan21.215554.4103@lorelei.approve.se> NNTP-Posting-Host: dialup-18-17.netcomuk.co.uk Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: taliesin.netcom.net.uk 885460022 14681 194.42.232.145 (22 Jan 1998 09:07:02 GMT) X-Complaints-To: abuse@corp.netcom.net.uk X-Newsreader: Forte Agent 1.5/32.451 X-No-Archive: yes Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!newsfeed.internetmci.com!195.99.66.215!news-feed1.eu.concert.net!btnet-peer!btnet!baron.netcom.net.uk!netcom.net.uk!knife.netcom.net.uk!netcom.net.uk!not-for-mail >They are slow. One reason why they are slow is that the Z80 fetched >the opcode for each repetition. The LDIR, for example, was just >like the LDI but with an decrement of BC and an inhibit of the PC >increment until BC = 0. They are a lot faster than the equivalent instructions. In this day of 200MHz pentiums nobody worries about a few extra (or a few thousand extra) opcodes, but in those days every bit mattered. For example, clever use of the INI instruction, and the way it sets the flags, made it possible to service a DD floppy drive with an I/O loop without the use of DMA. And the old LDIR is extremely useful. LDIR instead of LOOP: LD A, (HL) INC HL LD (DE), A INC DE DEC BC LD A, B OR C JR NZ, LOOP I won't bother to dig up the clock cycles. >Some claim this was done so the repeat >instructions could be interrupted by interrupts, but to me it smells >like a cheap kludge. Well, you can't have it both ways. You can't have the speed of continuous DMA, and have DRAM refresh (not to mention interrupt servicing) in between. Not without some serious extra logic. Remember this was 1975. Back then, there were no real design or simulation tools. Everything was designed by hand. The Z80 family were excellent products, much more advanced than any of the other offerings of the day (6800, 8080). Some of the peripheral chips (e.g. the Z80-DMA) were reportedly more complex than the CPU. The people who designed those chips represented the very pinnacle of design ability; the sort of people who today design 10-million-transistor chips. Even today, I would challenge you to design a Z80 (in an FPGA, for example); I think you will soon withdraw the words "cheap kludge". Peter. Return address is invalid to help stop junk mail. E-mail replies to z80@digiXYZserve.com but remove the XYZ. ###### From: Paul Flinders Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 22 Jan 1998 09:37:10 +0000 Organization: UUNet UK server (post doesn't reflect views of UUNet UK) Lines: 33 Message-ID: References: <348fa205.3834263@news.compuserve.com> <69um37$8tc@access4.digex.net> <885195090snz@tnglwood.demon.co.uk> <34c66210.231590990@news.netcomuk.co.uk> <1998Jan21.215554.4103@lorelei.approve.se> NNTP-Posting-Host: users.finobj.com X-Newsreader: Gnus v5.4.66/Emacs 19.34 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!feeder.qis.net!btnet-peer!btnet!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) writes: > In article <34c66210.231590990@news.netcomuk.co.uk>, Peter wrote: > > > Not the 8080. The Z80 has the LDIR, CPIR etc. > > > > These can be extremely useful in optimising certain algorithms. > > They are slow. One reason why they are slow is that the Z80 fetched > the opcode for each repetition. The LDIR, for example, was just > like the LDI but with an decrement of BC and an inhibit of the PC > increment until BC = 0. Some claim this was done so the repeat > instructions could be interrupted by interrupts, but to me it smells > like a cheap kludge. I think that was part of the reason - the other part was that the DRAM refresh output occured in the second two clocks of the M1 cycle so you needed an opcode fetch to ensure that your RAM actually remembered things. LDDR, LDIR etc were faster than an LDI and branch because you didn't have to read the branch (which was two bytes and required IIRC 6 clocks if not taken and 8 if it was). I think that LDIR was 21 clocks per iteration and 19 for the last where the "branch" was not taken (but my Z80 manuals are at my parent's - about 50 miles away so I can't check, I'm sure someone will step in if my memory is wonky). If you wanted really fast moves then unwinding the loop and doing lots of repeated LDIs, followed by a branch. Since screen line lengths tended to be 40 or 80 bytes 40 or 80 LDIs in a row were common. -- Paul ###### From: Paul Flinders Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 22 Jan 1998 09:38:55 +0000 Organization: UUNet UK server (post doesn't reflect views of UUNet UK) Lines: 8 Message-ID: References: <90saa45a.fsf@chonsp.franklin.lugs.ch> <885435287.2103.5.nnrp-09.9e98ee68@news.demon.co.uk> NNTP-Posting-Host: users.finobj.com X-Newsreader: Gnus v5.4.66/Emacs 19.34 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newspeer.monmouth.com!Aladdin!aladdin.net!ns2.aladdin.net!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail spam@lisardrock.demon.co.uk writes: > > unfortunately, the 8088 didn't have protection levels (and iwrc, the > 6809 *did*) No the 6809 didn't, the 68000 did, however. ###### Newsgroups: comp.arch.embedded,alt.folklore.computers From: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Subject: Re: Why IBM chose Intel Message-ID: <1998Jan22.223534.14198@lorelei.approve.se> Originator: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson) Organization: [1] + 5934 done /bin/rm -rf ~/ & X-No-Archive: yes X-Newsreader: trn 4.0-test60 (5 October 1997) References: <348fa205.3834263@news.compuserve.com> <34c66210.231590990@news.netcomuk.co.uk> <1998Jan21.215554.4103@lorelei.approve.se> Date: Thu, 22 Jan 1998 22:35:34 GMT Lines: 13 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!news.algonet.se!fci-se!fci!newsfeed.sunet.se!news99.sunet.se!news01.sunet.se!192.71.180.34.MISMATCH!mn6.swip.net!seunet!mn4.swip.net!lorelei!not-for-mail In article , Paul Flinders wrote: > clocks if not taken and 8 if it was). I think that LDIR was 21 clocks > per iteration and 19 for the last where the "branch" was not taken LDI et.al. use 16 T cycles. LDIR et.al. use 21 T cycles and 16 T cycles for last iteration. -- Goran Larsson hoh AT approve DOT se I was an atheist, http://home1 DOT swipnet DOT se/%7Ew-12153/ until I found out I was God. ###### Newsgroups: comp.arch.embedded,alt.folklore.computers From: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Subject: Re: Why IBM chose Intel Message-ID: <1998Jan22.225005.14318@lorelei.approve.se> Originator: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson) Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson) Organization: [1] + 5934 done /bin/rm -rf ~/ & X-No-Archive: yes X-Newsreader: trn 4.0-test60 (5 October 1997) References: <348fa205.3834263@news.compuserve.com> <34c66210.231590990@news.netcomuk.co.uk> <1998Jan21.215554.4103@lorelei.approve.se> <34c70502.404376823@news.netcomuk.co.uk> Date: Thu, 22 Jan 1998 22:50:05 GMT Lines: 47 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.gip.net!news-raspail.gip.net!news.gsl.net!gip.net!newscore.univie.ac.at!newsfeed.sunet.se!news99.sunet.se!news01.sunet.se!192.71.180.34.MISMATCH!mn6.swip.net!seunet!mn4.swip.net!lorelei!not-for-mail In article <34c70502.404376823@news.netcomuk.co.uk>, Peter wrote: > They are a lot faster than the equivalent instructions. Yes, but slower than what they could have been. > In this day of 200MHz pentiums nobody worries about a few extra (or a > few thousand extra) opcodes, but in those days every bit mattered. For I still do. If you design realtime operating systems that has to cater for applications that requires over 6200 process switches per second. That is 160us for the process switch and for the application to do usefull work. A couple of thousand extra opcodes will be noticed :-) > example, clever use of the INI instruction, and the way it sets the > flags, made it possible to service a DD floppy drive with an I/O loop > without the use of DMA. Did that. The trick is to connect the data available bit and the end of command bit to the proper port bits. > >Some claim this was done so the repeat > >instructions could be interrupted by interrupts, but to me it smells > >like a cheap kludge. > > Well, you can't have it both ways. You can't have the speed of > continuous DMA, and have DRAM refresh (not to mention interrupt > servicing) in between. Not without some serious extra logic. See. Can't afford "some serious extra logic" == cheap. > of the day (6800, 8080). Some of the peripheral chips (e.g. the > Z80-DMA) were reportedly more complex than the CPU. The people who I would think that even the SIO was more complex. > today, I would challenge you to design a Z80 (in an FPGA, for > example); I think you will soon withdraw the words "cheap kludge". I still think that they could have done the 'R instructions more efficient. Reading and decoding the opcode for each iteration has to be classified as a kludge. -- Goran Larsson hoh AT approve DOT se I was an atheist, http://home1 DOT swipnet DOT se/%7Ew-12153/ until I found out I was God. ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 23 Jan 1998 01:26:35 +0100 Organization: My own Private Self Lines: 23 Message-ID: References: <01bd2480$e6d3dca0$bcc3accf@mfc> <885254391.5398.5.nnrp-01.9e98ee68@news.demon.co.uk> <90saa45a.fsf@chonsp.franklin.lugs.ch> <6a5he6$2dq$1@news.iastate.edu> X-Newsreader: Gnus v5.3/Emacs 19.34 In article <90saa45a.fsf@chonsp.franklin.lugs.ch>, wrote: >On the 8088 the OS can do it transparently also. Simply declare the >segment registers tabu for the user software. rhawkins@iastate.edu (Rick Hawkins) noticed rightly: >A tabu which experience shows us that no software writers will try to >get around :) That one is actually simple: just have God (errr, root) declare that "you shalt not use any compiler but mine (/usr/bin/cc)" and make sure that that one can only produce code without MOV xS, xx instructions. IIRC that is what both Xenix and Minix did. It worked. OK, there would never exist lusers who would ever experiment with assembler tricks. On my honour, why would anyone ever do something like that! What? DOS programmers? They do not exist! Pure imagination :-) -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 23 Jan 1998 18:41:33 +0100 Organization: My own Private Self Lines: 30 Message-ID: References: <90saa45a.fsf@chonsp.franklin.lugs.ch> <885435287.2103.5.nnrp-09.9e98ee68@news.demon.co.uk> X-Newsreader: Gnus v5.3/Emacs 19.34 On 1998-01-21 Neil.Franklin.remove.this@ccw.ch said: -On the 8088 the OS can do it transparently also. Simply declare the -segment registers tabu for the user software. With CS <> DS = SS = -ES you could even do pdp11/70-like D+I adressing. Would make an -ideal Unix Verson 6 or 2.xBSD machine. IIRC that is what Xenix did. spam@lisardrock.demon.co.uk remarked correctly: > unfortunately, the 8088 didn't have protection levels As I forgot to write and then added in the next post: Yes, the 8088 simply lacked protection. That only came with the 286, "protected mode" got its name from that. > minix does that too, but it relies on the user not cheating. IIRC that is so (I only read the Tannenbaum book about 8 years ago) on the 8088, on the 286 I am not sure if it may have used the hardware. > (and iwrc, the 6809 *did*) No, it had no protection either. I actually have a 6809 (Dragon 32, similar in design to the TRS CoCo), it is about 2 meters away from where I am now. -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 23 Jan 1998 18:52:56 +0100 Organization: My own Private Self Lines: 32 Message-ID: References: <885435288.2103.6.nnrp-09.9e98ee68@news.demon.co.uk> X-Newsreader: Gnus v5.3/Emacs 19.34 unknown wrote: ->Perhaps something from Xerox? On 1998-01-21 Neil.Franklin.remove.this@ccw.ch said: -IIRC, they had 32bit at that time. But try getting a raw processor -out of them, also their stuff was TTL and ECL so far I know. spam@lisardrock.demon.co.uk > xerox seemed to be pathologically committed to 16-bit. the alto was > (since it was a nova clone, that's not surprising) but iwrc, so were the > dorado and the dolphin. > (course if they could make it work, it begs the question... :> ) Alto was definitely 16bit (actually 64k*16bit (128 kByte) memory. Some of the later ones had a 4* large address space via memory managment. Dorado had up to 16Mbyte momory, that suggests 32 bit (or at least 24bit) architecture. Dolphin had "more momory that Alto". But these ware all non-uP systems. Dandelion (Xerox Star 8010) still was made from 2901 bit slices. Source for this: Adele Goldberg (ed), A History of Personal Workstations ACM Press, 1988, ISBN 0-201-11259-0 Page 287..289, in the Article Charles P Thacker, Personal Distributed Computing: The Alto and Ethernet Hardware, p265..290 -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: spam@lisardrock.demon.co.uk Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 23 Jan 1998 20:57:53 GMT Message-ID: <885589073.26847.7.nnrp-07.9e98ee68@news.demon.co.uk> References: NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 17 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-22 paul.flinders@finobj.com said: -spam@lisardrock.demon.co.uk writes: -> unfortunately, the 8088 didn't have protection levels (and iwrc, ->the 6809 *did*) -No the 6809 didn't, the 68000 did, however. thanks for the correction. but the 6809 did have two stack pointers, which is what confused us. -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: spam@lisardrock.demon.co.uk Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 23 Jan 1998 23:24:54 GMT Message-ID: <885597894.5107.2.nnrp-06.9e98ee68@news.demon.co.uk> References: <1998Jan22.225005.14318@lorelei.approve.se> NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 23 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-22 hoh@approve.se.NO_JUNK_EMAIL(GoranLarsson) said: -> today, I would challenge you to design a Z80 (in an FPGA, for -> example); I think you will soon withdraw the words "cheap kludge". -I still think that they could have done the 'R instructions more -efficient. Reading and decoding the opcode for each iteration has -to be classified as a kludge. hell, maybe they just ran out of transistor space (qv. i8086 aam/aad instructions)... or you could just take the view that since the entire instruction set was a kludge on the 8080, what does one more matter...? :> the best approach might be to look at the z8000; if that does the same, then zilog probably meant to do it that way... -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: spam@lisardrock.demon.co.uk Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 23 Jan 1998 23:24:57 GMT Message-ID: <885597897.5107.3.nnrp-06.9e98ee68@news.demon.co.uk> References: NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 46 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-23 Neil.Franklin.remove.this@ccw.ch said: -spam@lisardrock.demon.co.uk remarked correctly: -> unfortunately, the 8088 didn't have protection levels -As I forgot to write and then added in the next post: Yes, the 8088 -simply lacked protection. That only came with the 286, "protected -mode" got its name from that. we know, we do actually know a reasonable amount about these things... :> the 286 also had segmentation-done-properly in that mode (it actually looked like minicomputer segmentation) but it still didn't have paging, and you still couldn't actively replace chunks of memory with others and keep within a linear memory space. replacing the entire data space isn't really what we had in mind... -> minix does that too, but it relies on the user not cheating. -IIRC that is so (I only read the Tannenbaum book about 8 years ago) -on the 8088, on the 286 I am not sure if it may have used the -hardware. well, it might have done if it noticed it was running on one. the modern versions probably do, but the ancient ones probably don't. in any case, it may well be better to make a completely new design of os - object oriented, perhaps? - for the 286. it's the only chance of using all the available memory and still having a system that doesn't fight the programmer. and the protection comes in handy for enforcing encapsulation there. -No, it had no protection either. I actually have a 6809 (Dragon 32, -similar in design to the TRS CoCo), it is about 2 meters away from -where I am now. hey, we remember those things! wasn't the contents of the case 65% air or something..? :> we always fancied one, just not at the price at which they debuted. (and remember, we're in the uk; the coco was similar in design to the dragon, from our perspective. :P ) -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: spam@lisardrock.demon.co.uk Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Fri, 23 Jan 1998 23:24:59 GMT Message-ID: <885597899.5107.4.nnrp-06.9e98ee68@news.demon.co.uk> References: NNTP-Posting-Host: lisardrock.demon.co.uk X-NNTP-Posting-Host: lisardrock.demon.co.uk [158.152.238.104] X-Everything: Net-Tamer V 1.09.2 Lines: 42 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!dispose.news.demon.net!demon!bullseye.news.demon.net!demon!news.demon.co.uk!demon!lisardrock.demon.co.uk!not-for-mail On 1998-01-23 Neil.Franklin.remove.this@ccw.ch said: -->Perhaps something from Xerox? --IIRC, they had 32bit at that time. But try getting a raw processor --out of them, also their stuff was TTL and ECL so far I know. -> xerox seemed to be pathologically committed to 16-bit. the alto ->was (since it was a nova clone, that's not surprising) but iwrc, ->so were the dorado and the dolphin. -> (course if they could make it work, it begs the question... :> ) -Alto was definitely 16bit (actually 64k*16bit (128 kByte) memory. -Some of the later ones had a 4* large address space via memory -managment. -Dorado had up to 16Mbyte momory, that suggests 32 bit (or at least -24bit) architecture. Dolphin had "more momory that Alto". But these -ware all non-uP systems. Dandelion (Xerox Star 8010) still was made -from 2901 bit slices. that's the danger of trying to guess word length supported from memory space available. the dorado was basically what they dreamed the alto would become; it did stick the the alto's basic architecture in some ways, and the accumulators (actually 256-deep ac stacks) were 16 bits wide. also, dorado smalltalk-80 used 15-bit addresses (well, table entries) and smallints with a flag to discriminate between them. the dolphin was a poor man's dorado, iwrc (we don't know much about the dolphin); it was implemented in ttl as opposed to the dorado's ecl, but was otherwise similar enough to also run smalltalk-80 in the same way. sources: memory of articles read, particularly in "bits of history, words of advice". wonderful book, btw. -- Communa - all at lisardrock.demon.net to be sure we read your reply, send to username `communa' Net-Tamer V 1.09.2 - Test Drive ###### From: shoppa@alph02.triumf.ca (Tim Shoppa) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 23 Jan 1998 23:37:58 GMT Organization: TRIUMF, Canada's National Meson Facility Lines: 26 Message-ID: <6ab9km$7gl$1@nntp.ucs.ubc.ca> References: <348fa205.3834263@news.compuserve.com> <1998Jan21.215554.4103@lorelei.approve.se> <34c70502.404376823@news.netcomuk.co.uk> <1998Jan22.225005.14318@lorelei.approve.se> NNTP-Posting-Host: alph02.triumf.ca Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newspeer.monmouth.com!news-peer.gip.net!news.gsl.net!gip.net!newsfeed.internetmci.com!164.67.42.145!awabi.library.ucla.edu!137.82.194.1!unixg.ubc.ca!alph02.triumf.ca!shoppa In article <1998Jan22.225005.14318@lorelei.approve.se>, Goran Larsson wrote: >In article <34c70502.404376823@news.netcomuk.co.uk>, Peter wrote: >> >Some claim this was done so the repeat >> >instructions could be interrupted by interrupts, but to me it smells >> >like a cheap kludge. >> >> Well, you can't have it both ways. You can't have the speed of >> continuous DMA, and have DRAM refresh (not to mention interrupt >> servicing) in between. Not without some serious extra logic. > >See. Can't afford "some serious extra logic" == cheap. Yeah, so? It's easy to say, 20 years later, what folks back then could've done if they had today's technology. In 1976, the Z80 was quite an accomplishment! And the fact that Zilog and others could churn them out by the millions means that cheap is worth a lot in the real world :-) >I still think that they could have done the 'R instructions more >efficient. Reading and decoding the opcode for each iteration has >to be classified as a kludge. You have extremely high standards - lucky for Zilog that most of the world was satisfied with a cheap fast easy-to-design-around CPU :-) Tim. ###### From: z80@ds.com (Peter) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sat, 24 Jan 1998 20:42:03 GMT Organization: x Lines: 21 Message-ID: <34ca2947.170465@news.netcomuk.co.uk> References: <1998Jan22.225005.14318@lorelei.approve.se> <885597894.5107.2.nnrp-06.9e98ee68@news.demon.co.uk> NNTP-Posting-Host: dialup-20-44.netcomuk.co.uk Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Trace: taliesin.netcom.net.uk 885674449 28855 194.42.233.44 (24 Jan 1998 20:40:49 GMT) X-Complaints-To: abuse@corp.netcom.net.uk X-Newsreader: Forte Agent 1.5/32.451 X-No-Archive: yes Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!nntp.news.xara.net!xara.net!cableinet-uk!news.cableinet.co.uk!baron.netcom.net.uk!netcom.net.uk!knife.netcom.net.uk!netcom.net.uk!not-for-mail >or you could just take the view that since the entire instruction set >was a kludge on the 8080, what does one more matter...? :> the best >approach might be to look at the z8000; if that does the same, then >zilog probably meant to do it that way... I am sure that on the Z80 the repeated opcode fetch was something which was very simple to implement, and it neatly took care of the automatic DRAM refresh which was done during the M1 (opcode fetch) cycle. The Z80 DRAM refresh feature was a major selling point, at a time when the alternative was either a fair bit of logic, or a DRAM controller from e.g. Intel at $50 a time. Peter. Return address is invalid to help stop junk mail. E-mail replies to z80@digiXYZserve.com but remove the XYZ. ###### From: dgy@rtd.com (Don Yuniskis) Newsgroups: comp.arch.embedded,alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 24 Jan 1998 21:09:52 GMT Organization: none Lines: 24 Message-ID: <6adlb0$1qi$1@baygull.rtd.com> References: <1998Jan22.225005.14318@lorelei.approve.se> <885597894.5107.2.nnrp-06.9e98ee68@news.demon.co.uk> <34ca2947.170465@news.netcomuk.co.uk> NNTP-Posting-Host: seagull.rtd.com Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.he.net!rtd.com!dgy In article <34ca2947.170465@news.netcomuk.co.uk>, Peter wrote: > >>or you could just take the view that since the entire instruction set >>was a kludge on the 8080, what does one more matter...? :> the best >>approach might be to look at the z8000; if that does the same, then >>zilog probably meant to do it that way... > >I am sure that on the Z80 the repeated opcode fetch was something >which was very simple to implement, and it neatly took care of the >automatic DRAM refresh which was done during the M1 (opcode fetch) >cycle. Don't forget, also, that it ensures a nice convenient place for interrupts to break into the "extended" instruction sequence. Easier to tie a bag onto an existing architecture than to start over from ground zero... >The Z80 DRAM refresh feature was a major selling point, at a time when >the alternative was either a fair bit of logic, or a DRAM controller >from e.g. Intel at $50 a time. Yes. Ditto with AMD's devices, etc. And the alternative of rolling your own out of LSTTL was big and "if-fy" in terms of guaranteeing worst case performance, etc. ###### From: ab528@FreeNet.Carleton.CA (Heinz W. Wiggeshoff) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 24 Jan 1998 22:33:54 GMT Organization: The National Capital FreeNet Lines: 10 Message-ID: <6adq8i$sni@freenet-news.carleton.ca> References: <1998Jan22.225005.14318@lorelei.approve.se> <885597894.5107.2.nnrp-06.9e98ee68@news.demon.co.uk> <34ca2947.170465@news.netcomuk.co.uk> <6adlb0$1qi$1@baygull.rtd.com> Reply-To: ab528@FreeNet.Carleton.CA (Heinz W. Wiggeshoff) NNTP-Posting-Host: freenet3.carleton.ca X-Given-Sender: ab528@freenet3.carleton.ca (Heinz W. Wiggeshoff) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!cam-news-hub1.bbnplanet.com!su-news-hub1.bbnplanet.com!news.bbnplanet.com!csn!nntp-xfer-1.csn.net!news.acsu.buffalo.edu!srv1.drenet.dnd.ca!crc-news.doc.ca!nott!cunews!freenet-news.carleton.ca!FreeNet.Carleton.CA!ab528 Don Yuniskis (dgy@rtd.com) writes: > Don't forget, also, that it ensures a nice convenient place for > interrupts to break into the "extended" instruction sequence. > Easier to tie a bag onto an existing architecture than to start > over from ground zero... Ie. like the IBM System /370 opcodes MVCL, CLCL - obviously one must allow paging interrupts when operands cross a page boundary. ###### From: William.Hamblen@nashville.com (William Hamblen) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Sat, 24 Jan 1998 23:56:50 GMT Organization: Utterly Disorganized Lines: 15 Message-ID: <34cc7f7b.16089697@news.nashville.com> References: <1998Jan22.225005.14318@lorelei.approve.se> <885597894.5107.2.nnrp-06.9e98ee68@news.demon.co.uk> <34ca2947.170465@news.netcomuk.co.uk> <6adlb0$1qi$1@baygull.rtd.com> Reply-To: William.Hamblen@nashville.com NNTP-Posting-Host: 7327@207.65.180.141 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.5/32.452 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.he.net!Supernews60!supernews.com!Supernews69!not-for-mail On 24 Jan 1998 21:09:52 GMT, dgy@rtd.com (Don Yuniskis) wrote: >In article <34ca2947.170465@news.netcomuk.co.uk>, Peter wrote: > >>The Z80 DRAM refresh feature was a major selling point, at a time when >>the alternative was either a fair bit of logic, or a DRAM controller >>from e.g. Intel at $50 a time. > >Yes. Ditto with AMD's devices, etc. And the alternative of >rolling your own out of LSTTL was big and "if-fy" in terms >of guaranteeing worst case performance, etc. When the Altair 8800 was new the main difference between MIT's dynamic memory board and their static memory board was that the static board worked. ###### From: Tom Stepleton Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Sun, 25 Jan 1998 00:52:39 -0600 Organization: Washington University in St. Louis Lines: 37 Message-ID: <34CAE137.3A5911DC@rototiller.wustl.edu> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> NNTP-Posting-Host: as0-isdn-81.wustl.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01 (X11; I; Linux 2.0.33 i586) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newsxfer.visi.net!news.iag.net!news.sgi.com!nntprelay.mathworks.com!news.mv.net!newspump.wustl.edu!newsreader.wustl.edu!not-for-mail Mark Borgerson wrote: > I think this was the way that Apple managed virtual memory on the Lisa-- > some sort of second processor retained enough information to allow > instruction restart. I've still got one up in the attic. I ought to I have here a paper by a gentleman named Bruce Daniels entitled "The Architecture of the Lisa Personal Computer", photocopied out of "Proceedings of the IEEE" Vol. 72, No. 3, Mar. 1984. In the section entitiled "The Memory Management Unit", Mr. Daniels identifies the MMU as "a section of logic on the processor board". He then goes on to explain the MMU's technical details. Some brief points: The MMU divides the Lisa's logical address space into 128 segments which can vary in size from 512 bytes to 128k. "The upper 7 bits of a 24-bit logical address is the segment number and the remaining 17 bits is the offset within that segment. The offset consists of 8 upper bits which is the logical block number and 9 bits of the displacement within the block." There is more interesting stuff too, including a memory protection scheme employing devices named "contexts". If anyone is interested in hearing about them, let me know and I will be happy to quote more from the article... ....after midterms are over. If anyone is interested in playing "Spot the MMU", I have some previously-unnanounced Lisa schematics online at http://galena.tjs.org/tom/schematics/ Note that their creator, who copied them from the real thing, has informed me that they have a few minor flaws. He didn't elaborate. Tom -- s/rototiller/artsci in my address to e-mail me http://galena.tjs.org ###### From: dpeschel@u.washington.edu (D. Peschel) Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: 25 Jan 1998 08:53:32 GMT Organization: University of Washington, Seattle Lines: 17 Message-ID: <6aeuic$19r6$1@nntp6.u.washington.edu> References: <348fa205.3834263@news.compuserve.com> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> <34CAE137.3A5911DC@rototiller.wustl.edu> NNTP-Posting-Host: saul9.u.washington.edu X-Trace: nntp6.u.washington.edu 885718412 42854 (None) 140.142.64.7 X-Complaints-To: help@cac.washington.edu NNTP-Posting-User: dpeschel Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!news-peer.sprintlink.net!news.sprintlink.net!Sprint!worldnet.att.net!news.u.washington.edu!dpeschel In article <34CAE137.3A5911DC@rototiller.wustl.edu>, Tom Stepleton wrote: >I have here a paper by a gentleman named Bruce Daniels entitled "The >Architecture of the Lisa Personal Computer", photocopied out of >"Proceedings of the IEEE" Vol. 72, No. 3, Mar. 1984. In the section >entitiled "The Memory Management Unit", Mr. Daniels identifies the >MMU as "a section of logic on the processor board". He then goes on >to explain the MMU's technical details. Some brief points: [...] This is very interesting. The more I learn about the Lisa, the more it becomes apparent that the Mac was a big step backward, which Apple hasn't completely undone yet. -- Derek ###### From: "George R. Gonzalez" Newsgroups: alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Sun, 25 Jan 1998 11:26:38 -0600 Organization: SkyPoint Communications, Inc. Lines: 37 Message-ID: <6afsje$28q$1@shadow.skypoint.net> References: <348fa205.3834263@news.compuserve.com> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> <34CAE137.3A5911DC@rototiller.wustl.edu> <6aeuic$19r6$1@nntp6.u.washington.edu> NNTP-Posting-Host: dial337.skypoint.net Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Newsreader: Microsoft Outlook Express 4.71.1712.3 X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.onenet.net!news.oru.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!Supernews60!supernews.com!winternet.com!skypoint.com!not-for-mail D. Peschel wrote in message <6aeuic$19r6$1@nntp6.u.washington.edu>... >In article <34CAE137.3A5911DC@rototiller.wustl.edu>, >Tom Stepleton wrote: > >>I have here a paper by a gentleman named Bruce Daniels entitled "The >>Architecture of the Lisa Personal Computer", photocopied out of >>"Proceedings of the IEEE" Vol. 72, No. 3, Mar. 1984. In the section >>entitiled "The Memory Management Unit", Mr. Daniels identifies the >>MMU as "a section of logic on the processor board". He then goes on >>to explain the MMU's technical details. Some brief points: > >[...] > >This is very interesting. The more I learn about the Lisa, the more >it becomes apparent that the Mac was a big step backward, which Apple >hasn't completely undone yet. > >-- Derek You probably never tried to use one! It was so horribly slow. Bringing up the desk calculator took almost a minute. Asking for 2+2 took maybe 15 seconds to get an answer. And that's on a Lisa with an extra 1Meg of memory ($1200 for the memory with the developer's discount). I tried writing a VT100 emulator for it but gave up as the serial port routines were too slow to work at anything over 300 baud. The application software looked impressive-- but way too slow to use. Allegedly the whole schmeer was written in a top-down overdesigned fashion. As often happens, you end up with something that looks good but lacks several orders of magnitude of speed and ergonomics. ###### From: Tom Stepleton Newsgroups: alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Sun, 25 Jan 1998 15:02:52 -0600 Organization: Washington University in St. Louis Lines: 44 Message-ID: <34CBA87C.5EA8094A@rototiller.wustl.edu> References: <348fa205.3834263@news.compuserve.com> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> <34CAE137.3A5911DC@rototiller.wustl.edu> <6aeuic$19r6$1@nntp6.u.washington.edu> <6afsje$28q$1@shadow.skypoint.net> NNTP-Posting-Host: as0-isdn-98.wustl.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01 (X11; I; Linux 2.0.33 i586) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!news.ecn.uoknor.edu!news.wildstar.net!news.ececs.uc.edu!kragar.kei.com!news.mathworks.com!nntprelay.mathworks.com!news.mv.net!newspump.wustl.edu!newsreader.wustl.edu!not-for-mail George R. Gonzalez wrote: > You probably never tried to use one! It was so horribly slow. > Bringing up the desk calculator took almost a minute. Asking for 2+2 > took maybe 15 seconds to get an answer. And that's on a Lisa with an > extra 1Meg of memory ($1200 for the memory with the developer's > discount). That's a bit of an exaggeration. On my left is a Lisa 2/10 with 1 MB RAM (as opposed to your two megs). Starting cold (i.e. run for the first time in an Office System session) the Office System calculator loads in 17 seconds; starting up a second time it loads in about three seconds. In RPN mode, entering "2 Enter 2 +" and receiving the answer "4" takes about a second: the delay is my typing. There are probably only marginal speed differences between this calculator (supplied with the Lisa Office System 3.1) and the first one on the Lisa 1 (Lisa Office System 1.0). Besides, you can always load the calculator once and leave it running in the background... Here's an honest indicator of the Lisa's floating point computational power, such as it is: LisaMandelbrot, my kludgy, unoptimized Mandelbrot fractal viewer for the Office System, will calculate a 96x64 Mandelbrot Set at 50 iterations in about 25 minutes. It calculates the whole set: it does not take advantage of the Mandelbrot's horizontal symmetry. > I tried writing a VT100 emulator for it but gave up as the serial port > routines were too slow to work at anything over 300 baud. LisaTerminal communicates reliably over a null-modem cable with my Linux box at 2400 bps. The Lisa Pascal Workshop Transfer utility is reliable at 9600 bps. I know you're probably expecting me to shout "Lisa forever!" right about now. I won't: the Lisa was dog-slow compared to its contemporaries. It was far too expensive. Nevertheless, the Office System 3.1 is by far the easiest 'office' environment I have seen. Overall, if you had the patience, the Lisa was a very elegantly designed machine. Tom -- s/rototiller/artsci in my address to email me. http://galena.tjs.org/ ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 25 Jan 1998 19:46:37 +0100 Organization: My own Private Self Lines: 56 Message-ID: References: <885597897.5107.3.nnrp-06.9e98ee68@news.demon.co.uk> X-Newsreader: Gnus v5.3/Emacs 19.34 On 1998-01-23 Neil.Franklin.remove.this@ccw.ch said: -As I forgot to write and then added in the next post: Yes, the 8088 -simply lacked protection. That only came with the 286, "protected -mode" got its name from that. spam@lisardrock.demon.co.uk replied: > we know, we do actually know a reasonable amount about these things... > :> the 286 also had segmentation-done-properly in that mode (it actually > looked like minicomputer segmentation) but it still didn't have paging, > and you still couldn't actively replace chunks of memory with others and > keep within a linear memory space. replacing the entire data space isn't > really what we had in mind... Yes, they waited for the 386 to do that. Insider tip: they needed to keep that as a sales argument for getting people to upgrade ;-) spam@lisardrock.demon.co.uk replied: > in any case, it may well be better to make a completely new design of os > - object oriented, perhaps? - for the 286. it's the only chance of using > all the available memory and still having a system that doesn't fight > the programmer. and the protection comes in handy for enforcing > encapsulation there. Didn't a firm called IBM try to do exactly that with half an OS? They just never found the other half :-) On 1998-01-23 Neil.Franklin.remove.this@ccw.ch said: -No, it had no protection either. I actually have a 6809 (Dragon 32, -similar in design to the TRS CoCo), it is about 2 meters away from -where I am now. spam@lisardrock.demon.co.uk replied: > hey, we remember those things! wasn't the contents of the case 65% air > or something..? :> Yep, exactly that one! For those not there, here a diagram of it: ccccccccccccccccccc c cc cccccc cc kkkkkkkk air (lots) cc ccccc ppp c c mmmmmmmmmmmmmmmmmmmmm ppppppppp c ccccccccccccccccccccccccccccccccccccccccccc kkk = keyboard, mmm = mainboard, ppp = power supply, ccc = case Not to forget that "guaranteed 20'000 presses each key" keyboard which, together with that big resonance chamber, produced the clooonk clooonk..., no keybeep speaker needed :-) -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### Path: ccw.ch!usenet From: Neil.Franklin.remove.this@ccw.ch Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 25 Jan 1998 19:55:03 +0100 Organization: My own Private Self Lines: 35 Message-ID: References: <885597899.5107.4.nnrp-06.9e98ee68@news.demon.co.uk> X-Newsreader: Gnus v5.3/Emacs 19.34 On 1998-01-23 Neil.Franklin.remove.this@ccw.ch said: -Alto was definitely 16bit (actually 64k*16bit (128 kByte) memory. -Some of the later ones had a 4* large address space via memory -managment. -Dorado had up to 16Mbyte momory, that suggests 32 bit (or at least -24bit) architecture. Dolphin had "more momory that Alto". But these -ware all non-uP systems. Dandelion (Xerox Star 8010) still was made -from 2901 bit slices. spam@lisardrock.demon.co.uk corrected this: > that's the danger of trying to guess word length supported from memory > space available. the dorado was basically what they dreamed the alto > would become; it did stick the the alto's basic architecture in some > ways, and the accumulators (actually 256-deep ac stacks) were 16 bits > wide. also, dorado smalltalk-80 used 15-bit addresses (well, table > entries) and smallints with a flag to discriminate between them. 16 bit object pointers mapped by a table to 24 bit physical adresses. Now why does that remind me of the 286? Did Intel copy from them :-) ;-) :-) (yes, that is an animated smiley) That they "forgot" the smallint optimisation is obvious. > sources: memory of articles read, particularly in "bits of history, > words of advice". wonderful book, btw. Sounds interesting. Do you have Author, ISBN? -- Neil.Franklin.remove.this@ccw.ch, http://www.ccw.ch/Neil.Franklin/ for Geek Code, Papernet, Voicenet, PGP public key see http: Mac, 95 and NT users are CLUEless (Command Line User Environment) If I go missing, its once again my newsfeed that has craped ###### From: newcomer@flounder.com (Joseph M. Newcomer) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Mon, 26 Jan 1998 02:58:25 GMT Organization: Pittsburgh OnLine, Inc. Lines: 34 Message-ID: <34edf961.456845646@206.210.64.12> References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> NNTP-Posting-Host: ppp46.s8.pgh.net Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Newsreader: Forte Agent 1.5/32.451 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!howland.erols.net!fastnet!news.pgh.net!not-for-mail One of the questions I put on a PhD hardware qualifier exam was "Design an MMU for a Z80". The ground rules were: (1) You had to run with a stock Z80 (2) You didn't have to support demand paging (Assume an architecture that guaranteed the entire working set was swapped in) (3) You did not have to go down to the gate level to get full credit, but might have to in order to justify a design decision (4) Fixed-size segments were all that was required (5) The resulting system had to have the same performance as a stock Z80 while executing one application (6) An application could not violate protection This was an open-book exam. We provided data sheets for the Z80 and a bunch of support chips, memory chips, etc. Nobody got it right. At least two students redesigned the Z80 chip (I thought that restriction was fairly clear), one said it was impossible because there was no instruction restart on a Z80 (I had actually spent a whole paragraph describing the operating system architecture that caused (2) to be effective). Everybody missed the key trick in the question, which has to do with the Z80 architecture, and as a consequence, nobody got any credit for this question. Surprising, because the Z80 was one of the chips they had studied the semester before (I checked with the faculty member teaching the course!) Anyone want to take a crack at this? Dust off your old Z80 manuals and sketch out a solution? Joseph M. Newcomer newcomer@flounder.com http://www3.pgh.net/~newcomer ###### From: Paul Flinders Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 26 Jan 1998 10:22:46 +0000 Organization: UUNet UK server (post doesn't reflect views of UUNet UK) Lines: 73 Message-ID: References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <34edf961.456845646@206.210.64.12> NNTP-Posting-Host: users.finobj.com X-Newsreader: Gnus v5.4.66/Emacs 19.34 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news.he.net!news-peer.gip.net!news.gsl.net!gip.net!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail newcomer@flounder.com (Joseph M. Newcomer) writes: > One of the questions I put on a PhD hardware qualifier exam was > "Design an MMU for a Z80". The ground rules were: > (1) You had to run with a stock Z80 > (2) You didn't have to support demand paging > (Assume an architecture that guaranteed the > entire working set was swapped in) > (3) You did not have to go down to the gate level > to get full credit, but might have to in order to > justify a design decision > (4) Fixed-size segments were all that was required > (5) The resulting system had to have the same performance > as a stock Z80 while executing one application > (6) An application could not violate protection > > [...] > Anyone want to take a crack at this? Dust off your old Z80 manuals > and sketch out a solution? The details are sketchy - the machine and manuals are gathering dust at my parent's house - it's at least 7 years since I switched the thing on! My old z80 has a 16wordx8bit TTL ram (2x7489 I think) connected to the top 4 bits of the address bus. By writing into the RAM you can map each 4k page in the 64k address space of the z80 into a 4k page within 1M of extended address space. My machine had 512k installed. The problem is writing the RAM locations because you have the *top* 4 address bits going to the RAM - normally you'd have the lower 4 bits going to the RAM with the higer bits being gated for decode. You can solve this by making use of the fact that the z80 I/O space is not 8 bits but 16 - the inport/outport instructions are documented as having the I/O port supplied by the C register but if you go and look at the timing diagrams you can see that the upper 8 bits of address are not undefined during an I/O cycle but contain the contents of the B register (I may have swapped B/C here but you get the idea - the I/O address is BC together, not just B or C). So you gate the write enable from a "normal" port decode using the low 8 bits but set up the top 4 bits of BC before you do the I/O write. The split data in/data out lines of this RAM make the hardware fairly easy (although the RAM locations become write only). I didn't have any form of limit register - I suppose that you could add one if you needed. If you were multitasking under this arrangement you'd probably swap the whole of the 64k space (1M would then get you 16 processes). This would mean that for an application to corrupt the address space of another it would have to manage to do an I/O cycle to the page RAM first. I did once design a user/supervisor mode for the z80 but never built it. Basically the idea was that an external latch contained the user/supervisor bit. You could read the latch using an I/O read and set it with an I/O write, however if you tried to set it to "Supervisor" you'd also get a hardware interrupt thus you'd always end up in "kernel" code. You then needed to gate I/O and writes to the low end of memory (where the NMI and RST vectors were held) so that they did nothing when in User mode but generated an interrupt and set Supervisor mode. Thus no "user" program could do direct I/O or change the vectors. You also might want to protect a region of ram from write access and put the operating system code there. Other hardware interrupts also set "Supervisor". If you combined the above with the page RAM then I think you could almost guarentee full protection between apps although the "normal performance" would only apply if they didn't try to do I/O. Do I get the credit? ###### From: korpela@islay.ssl.berkeley.edu (Eric J. Korpela) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 26 Jan 1998 20:39:08 GMT Organization: Cal Berkeley-- Space Sciences Lab Lines: 26 Message-ID: <6ais9c$871$1@agate.berkeley.edu> References: <348fa205.3834263@news.compuserve.com> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <34edf961.456845646@206.210.64.12> NNTP-Posting-Host: islay.ssl.berkeley.edu Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-out.internetmci.com!newsfeed.internetmci.com!141.211.144.13!newsxfer3.itd.umich.edu!agate!islay.ssl.berkeley.edu!korpela In article <34edf961.456845646@206.210.64.12>, Joseph M. Newcomer wrote: > Anyone want to take a crack at this? Dust off your old Z80 manuals >and sketch out a solution? Hook up an 8 bit latch in the I/O port address space. Use it to control the A16-A23 lines of your address space. Hook it up to 8 1Mx8 SIMMS. You can use circuitry to define a small shared code space to hold the switching code. I think you'll need external refresh circuitry. (Can't remember the Z-80 refresh details offhand.) Another option is to use a couple latches, a 256x8 SRAM and have a programmable MMU. Could even implement shared code/data spaces this way if you use address lines as input to the MMU. You're probably looking for something more complicated than this, though. probably dealing with using the M1 line to select the code space. The real answer depends upon how many segments you want a process to be able to access, how many segments you need to be simultaneously addressable, the size of the segments, etc. Eric -- Eric Korpela | An object at rest can never be korpela@ssl.berkeley.edu | stopped. Click here for more info. ###### From: werme@alingo.zk3.dec.com (Eric Werme) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 27 Jan 98 00:43:56 GMT Organization: Digital Equipment Corporation Lines: 26 Message-ID: References: <348fa205.3834263@news.compuserve.com> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <34edf961.456845646@206.210.64.12> <6ais9c$871$1@agate.berkeley.edu> Reply-To: werme@zk3.dec.com NNTP-Posting-Host: alingo.zk3.dec.com Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!howland.erols.net!fastnet!news-xfer.netaxs.com!xfer.kren.nm.kr!su-news-hub1.bbnplanet.com!news.bbnplanet.com!news1.digital.com!pa.dec.com!crl.dec.com!nntpd.lkg.dec.com!alingo.zk3.dec.com!werme korpela@islay.ssl.berkeley.edu (Eric J. Korpela) writes: >In article <34edf961.456845646@206.210.64.12>, >Joseph M. Newcomer wrote: >> Anyone want to take a crack at this? Dust off your old Z80 manuals >>and sketch out a solution? >Hook up an 8 bit latch in the I/O port address space. Use it to control >the A16-A23 lines of your address space. Hook it up to 8 1Mx8 SIMMS. >You can use circuitry to define a small shared code space to hold the >switching code. But Joe said: (6) An application could not violate protection You need to disable access to that I/O space while in "user mode", and some system call to get back to exec mode. Or am I reading too much into Joe's criteria? It's also been a long time since I was familiar with Z80 internals -- -- <> ROT-13 addresses: <> The above is unlikely to contain official <> <> <> claims or policies of Digital Equipment Corp. <> <> <> Eric (Ric) Werme <> ###### From: korpela@islay.ssl.berkeley.edu (Eric J. Korpela) Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 27 Jan 1998 01:03:44 GMT Organization: Cal Berkeley-- Space Sciences Lab Lines: 24 Message-ID: <6ajbpg$et9$1@agate.berkeley.edu> References: <348fa205.3834263@news.compuserve.com> <34edf961.456845646@206.210.64.12> <6ais9c$871$1@agate.berkeley.edu> NNTP-Posting-Host: islay.ssl.berkeley.edu Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!awabi.library.ucla.edu!128.32.155.1!agate!islay.ssl.berkeley.edu!korpela In article , Eric Werme wrote: >But Joe said: > > (6) An application could not violate protection > >You need to disable access to that I/O space while in "user mode", and some >system call to get back to exec mode. > >Or am I reading too much into Joe's criteria? It's also been a long time >since I was familiar with Z80 internals I think that's reading too much into it. You could gate IORD and IOWR to generate an interrrupt while in "user mode," but I don't know how you'd be able to change into "supervisor mode" without use of IO instructions. Then there's the question of whether a user mode program would have write privleges in code space. Eric -- Eric Korpela | An object at rest can never be korpela@ssl.berkeley.edu | stopped. Click here for more info. ###### From: Paul Flinders Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 27 Jan 1998 09:18:04 +0000 Organization: UUNet UK server (post doesn't reflect views of UUNet UK) Lines: 26 Message-ID: References: <348fa205.3834263@news.compuserve.com> <34edf961.456845646@206.210.64.12> <6ais9c$871$1@agate.berkeley.edu> <6ajbpg$et9$1@agate.berkeley.edu> NNTP-Posting-Host: users.finobj.com X-Newsreader: Gnus v5.4.66/Emacs 19.34 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail korpela@islay.ssl.berkeley.edu (Eric J. Korpela) writes: > I think that's reading too much into it. You could gate IORD and IOWR > to generate an interrrupt while in "user mode," I sketched out a possible design (described in another post) but never actually implemented it. You need to ignore IORD/IOWR whilst in "user mode" as the Z80 will complete the current instruction before taking the interrupt. > but I don't know how you'd be able to change into "supervisor mode" > without use of IO instructions. You could use M1 plus a comparator on the data bus to decode one of the single byte software interrupt instructions but it's probably easier to define the action to get nto supervisor mode as simply "set the supervisor bit" - doing the I/O cycle to write the latch will get caught, set the latch and generate an interrupt. > Then there's the question of whether a user mode program would have write > privleges in code space. Not if you want to implement shared text - you could hive off one line of the page RAM to be "write protect". ###### From: Paul Flinders Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 27 Jan 1998 09:21:43 +0000 Organization: UUNet UK server (post doesn't reflect views of UUNet UK) Lines: 20 Message-ID: References: <348fa205.3834263@news.compuserve.com> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <34edf961.456845646@206.210.64.12> <6ais9c$871$1@agate.berkeley.edu> NNTP-Posting-Host: users.finobj.com X-Newsreader: Gnus v5.4.66/Emacs 19.34 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-dc-3.sprintlink.net!news-dc-1.sprintlink.net!news-east.sprintlink.net!news-peer.sprintlink.net!news.sprintlink.net!Sprint!newsfeed.internetmci.com!158.43.192.17!rill.news.pipex.net!pipex!bore.news.pipex.net!pipex!not-for-mail korpela@islay.ssl.berkeley.edu (Eric J. Korpela) writes: > In article <34edf961.456845646@206.210.64.12>, > Joseph M. Newcomer wrote: > > Anyone want to take a crack at this? Dust off your old Z80 manuals > >and sketch out a solution? > > Hook up an 8 bit latch in the I/O port address space. Use it to control > the A16-A23 lines of your address space. Hook it up to 8 1Mx8 SIMMS.... Think in the time frame when Z80s were "serious computing" - I think that 256k DRAMs were just coming into circulation as the 8086 and 68000 were making it clear that the Z80s days were numbered. I don't recall SIMMs in that time frame. These days you'd just use one of the Z80 clones with a built in MMU (is it the Z800?) -- Paul ###### From: David Wragg Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 27 Jan 1998 14:06:11 +0000 Organization: Dept. of Computing, Imperial College, University of London, UK. Lines: 58 Message-ID: References: <348fa205.3834263@news.compuserve.com> <01bd21f7$848e5100$8cc3accf@mfc> <69m7t5$3el$1@max93.public.ox.ac.uk> <01bd23b5$70471000$a6c3accf@mfc> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <34edf961.456845646@206.210.64.12> NNTP-Posting-Host: outoften.doc.ic.ac.uk X-Newsreader: Gnus v5.4.37/XEmacs 19.15 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newspeer.monmouth.com!Aladdin!aladdin.net!ns2.aladdin.net!nntp.news.xara.net!xara.net!server6.netnews.ja.net!server4.netnews.ja.net!doc.ic.ac.uk!not-for-mail newcomer@flounder.com (Joseph M. Newcomer) writes: > One of the questions I put on a PhD hardware qualifier exam was > "Design an MMU for a Z80". The ground rules were: > (1) You had to run with a stock Z80 > (2) You didn't have to support demand paging > (Assume an architecture that guaranteed the > entire working set was swapped in) > (3) You did not have to go down to the gate level > to get full credit, but might have to in order to > justify a design decision > (4) Fixed-size segments were all that was required > (5) The resulting system had to have the same performance > as a stock Z80 while executing one application > (6) An application could not violate protection > To satisfy 1-5 the you have to decide which address bits select the segment, and what permissions bits would be nice. You don't say if a per-process address space should be more than 2^32 bytes, but if so you also need to get the extra address bits to the external logic, but there are simple enough ways to do that. So far, no great leap of the imagination, and nothing that you couldn't do on any other 8-bit microprocessor. > [...] > > Everybody missed the key trick in the question, which has to do with > the Z80 architecture, and as a consequence, nobody got any credit for > this question. Surprising, because the Z80 was one of the chips they > had studied the semester before (I checked with the faculty member > teaching the course!) Okay, so there's a Z80 trick concerned, and I guess it's to do with (6). The external logic clearly needs a latch saying whether it's in "supervisor" mode, where the segment registers can be set, or "user" mode, where they can't. Well, the Z80 has the one byte RST instructions. External logic can easily watch for these in the first instruction fetch cycle, and go to "supervisor" mode when it sees one. It probably should watch for interrupts, too. (The 8080 had RSTs too, right?) But that's hardly full protection. Instructions like DI, IM, LD I, A etc. should not be available in "user" mode. You can work around the lack of restartable instructions by have the external logic fake NOPs during the instruction fetch, but things can get hairy (the supervisor code would have to fix up the PC if it wanted to return to the user code; no two byte NOPs - well, there are two byte instructions with reversible effects; very hairy indeed. But if returning to a naughty user process isn't required then it's doable). Was full protection needed for credit? Can it be done in a simpler way? The alternate register set probably makes life easier as well. Without that, a user process might wreak havoc by putting a ridiculous value in its SP. So what was the trick? ###### From: Robert Billing Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: Tue, 27 Jan 98 21:37:59 GMT Message-ID: <885937079snz@tnglwood.demon.co.uk> References: <348fa205.3834263@news.compuserve.com> <34edf961.456845646@206.210.64.12> <6ais9c$871$1@agate.berkeley.edu> <6ajbpg$et9$1@agate.berkeley.edu> Reply-To: unclebob@tnglwood.demon.co.uk X-Mail2News-User: unclebob@tnglwood.demon.co.uk X-Mail2News-Path: post-10.mail.demon.net!post.mail.demon.net!tnglwood.demon.co.uk X-Trace: mail2news.demon.co.uk 885940935 29172 unclebob tnglwood.demon.co.uk X-Complaints-To: abuse@demon.net X-Newsreader: Demon Internet Simple News v1.29 Lines: 20 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newsxfer.visi.net!rill.news.pipex.net!pipex!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!tnglwood.demon.co.uk!unclebob In article paul.flinders@finobj.com "Paul Flinders" writes: > You could use M1 plus a comparator on the data bus to decode one of the > single byte software interrupt instructions but it's probably easier to Curiously I worked this out for the 6800 years ago. The software and hardware interrupts generated 7 consecutive write cycles, far more than any other combination of instructions. If you sit a synchronous counter across the bus, wired to count up on a write or preload one on a read, then a carry out of bit 2 indicates an interrupt, and a switch to supervisor state. -- I am Robert Billing, Christian, inventor, traveller, cook and animal lover, I live near 0:46W 51:22N. http://www.tnglwood.demon.co.uk/ "Bother," said Pooh, "Eeyore, ready two photon torpedoes and lock phasers on the Heffalump, Piglet, meet me in transporter room three" ###### From: Ken Smith Newsgroups: alt.folklore.computers Subject: Re: Why IBM chose Intel Date: 28 Jan 1998 05:35:14 GMT Organization: a2i network Lines: 21 Message-ID: <6amg2i$s5a$1@samba.rahul.net> References: <348fa205.3834263@news.compuserve.com> <6ais9c$871$1@agate.berkeley.edu> <6ajbpg$et9$1@agate.berkeley.edu> NNTP-Posting-Host: waltz.rahul.net NNTP-Posting-User: kensmith Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!newsxfer.visi.net!peerfeed.ncal.verio.net!bug.rahul.net!rahul.net!a2i!samba.rahul.net!rahul.net!a2i!kensmith.a2i!kensmith In article <6ajbpg$et9$1@agate.berkeley.edu>, Eric J. Korpela wrote: [....] >I think that's reading too much into it. You could gate IORD and IOWR >to generate an interrrupt while in "user mode," but I don't know how >you'd be able to change into "supervisor mode" without use of IO >instructions. If you do an I/O in "user" mode, it causes an interrupt and flips you into "super". The code for the supervisor looks at the instruction you did and the values in the registers. Based on this it can (a) Do the I/O for real and return to "user" mode, (b) do some operation for you and return to "user" mode, or (c) abort your program. A clever enough version of this could even run a copy of its self by similating all of the I/Os for the copy that is really in "user" mode. -- -- kensmith@rahul.net forging knowledge ###### From: David L Pearce Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Date: Fri, 06 Feb 1998 13:26:25 -0500 Organization: North Carolina State University Lines: 7 Message-ID: <34DB55D0.3C982A47@eos.ncsu.edu> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> <34D2A915.3CED9370@rototiller.wustl.edu> <34DADF40.6191@concentric.net> NNTP-Posting-Host: cfr3034pc1.cfr.ncsu.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.03 [en] (WinNT; I) Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!newsfeed.usit.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!nntprelay.mathworks.com!gatech!nntp-xfer.ncsu.edu!not-for-mail > Sorry, the 68450 was a four channel DMA controller. > I spent part of my youth making a 68450 talk to a z8530. Wouldn't know who sells these in small quantities would you? David ###### From: ard@p850ug1.demon.co.uk (Tony Duell) Newsgroups: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Subject: Re: Why IBM chose Intel Followup-To: comp.arch.embedded,alt.folklore.computers,comp.sys.m68k Date: 6 Feb 1998 23:47:06 -0000 Organization: P850 User Group Message-ID: <6bg7dq$vk@p850ug1.demon.co.uk> References: <348fa205.3834263@news.compuserve.com> <69s7qb$df2$1@max89.public.ox.ac.uk> <01bd247c$b5930de0$bcc3accf@mfc> <34c2d897.0@comnet.co.nz> <6a0pv0$19a$1@max77.public.ox.ac.uk> <6a4qlb$8a1$1@enyo.uwa.edu.au> <34C628AB.6308@wetlabs.com> <34D2A915.3CED9370@rototiller.wustl.edu> <34DADF40.6191@concentric.net> NNTP-Posting-Host: p850ug1.demon.co.uk X-NNTP-Posting-Host: p850ug1.demon.co.uk [158.152.97.199] X-Newsreader: TIN [version 1.2 PL2] Lines: 15 Path: ccw.ch!aetna.dolphins.ch!news.planetc.com!leto.ou.edu!hammer.uoregon.edu!logbridge.uoregon.edu!nntprelay.mathworks.com!woodstock.news.demon.net!demon!news.demon.co.uk!demon!p850ug1.demon.co.uk!not-for-mail Michael L. Drew (mldrew@concentric.net) wrote: : Sorry, the 68450 was a four channel DMA controller. Correct, but there was also an MMU called the 68451. I don't know much about it, but there's one in my Torch XXX, connected to a 68010 CPU. : I spent part of my youth making a 68450 talk to a z8530. : The 68851 was the Paged Memory Management Unit. : It was a monster. I remember waiting for them to come out. : Mike -tony