From: "Philo" Newsgroups: alt.folklore.computers Subject: CRAY 1 Date: Thu, 10 Aug 2000 03:39:33 -0000 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2314.1300 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 Lines: 36 Message-ID: <39926ac5$0$159$39368dfe@news.twtelecom.net> NNTP-Posting-Host: 207.250.72.192 X-Trace: 965896903 news.twtelecom.net 159 207.250.72.192 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news.uni-stuttgart.de!rz.uni-karlsruhe.de!schlund.de!newsfeed01.sul.t-online.de!t-online.de!blackbush.xlink.net!npeer.kpnqwest.net!news-peer.gip.net!news.gsl.net!gip.net!cyclone2.usenetserver.com!news-out.usenetserver.com!newsfeed.onecall.net!nntp.inc.net!news.twtelecom.net!news.twtelecom.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:61702 Perhaps someone may be interested...perhaps someone may correct my memory...but I had a friend who worked for Cray around 1980+ and I was lucky enough to have him give me a tour of the facilities. He had previously been employed at GE and had worked on the design of the cat-scanner. (i think he was involved with the FFT design) anyway... at the time i took the informal tour i had little knowledge of computers . so my memories of that day may be slightly skewed... but as i recall, the Cray1 had used TTL technology and had been surprised it was not running CMOS (and lower power consumption) another thing which surprised me was how standard the construction was. the boards (though very closely spaced) used discrete TTL logic not too terribly different (to my eyes) than standard technology. The packages were smaller...perhaps similar to smt...if i recall...but the devices themselves were all quite familiar 74xx (Don't recall TTL any more) But there was one thing which did *impress* me. .. the board traces... there was often a slight , extra loop going to a gate here and there. When I asked Ed why, he told me it was to allow for equal propigation time to the gates. (Probably standard design practice but to someone like me with low frequency design experience...impressive) So...twenty years later...anyone have comments on my memory of the day. Also does anyone know what the buss speeed and processor speed would have been of that machine, as I have lost track of my friend over the years? -- Philo website www.plazaerth.com/philo ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 10 Aug 2000 21:51:11 +0200 Organization: My own Private Self Lines: 53 Message-ID: <6uog30gaj4.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 965937071 1016 10.0.3.2 (10 Aug 2000 19:51:11 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 10 Aug 2000 19:51:11 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:61714 "Philo" writes: > at the time i took the informal tour i had little knowledge of computers . > so my memories of that day may be slightly skewed... > but as i recall, the Cray1 had used TTL technology and had been surprised it > was not running CMOS (and lower power consumption) It used ECL. CMOS and TTL were both too slow. ECL wants even more power than TTL, and 3 voltages to feed it. > another thing which surprised me was how standard the construction was. Seymour Cray was a conservative designer. And chips look like chips. The radical stuff was in how they were connected together. > the boards (though very closely spaced) used discrete TTL logic not too > terribly different (to my eyes) than standard technology. > The packages were smaller...perhaps similar to smt...if i recall...but the > devices themselves were all quite familiar 74xx (Don't recall TTL any more) Nope. ECLs. To be more precise: - dual NAND (first 5 and second 4 inputs) - octal D-FF (with common strobe) - 1kBit SRAM At least that is what the Cray-1 Hardware manual I read about 15 years ago said. And yes, I am annoyed at myself for not copying it. > But there was one thing which did *impress* me. .. the board traces... there > was often a slight , extra loop going to a gate here and there. > When I asked Ed why, he told me it was to allow for equal propigation time > to the gates. (Probably standard design practice but to someone like me with > low frequency design experience...impressive) Only in very fast stuff, from about 30-100MHz on up (depends on distance). > So...twenty years later...anyone have comments on my memory of the day. Also > does anyone know what the buss speeed and processor speed would have been of > that machine, as I have lost track of my friend over the years? Bus I do not know. But the processor was clocked somewhere around 70MHz, doing one floating point Op per cycle. This is of cource only when processing vectors. Else it would be factor 10 lower. -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### Newsgroups: alt.folklore.computers From: dp@world.std.com (Jeff DelPapa) Subject: Re: CRAY 1 Message-ID: Date: Thu, 10 Aug 2000 21:04:21 GMT References: <39926ac5$0$159$39368dfe@news.twtelecom.net> Organization: Chaos and Confusion Lines: 108 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!informatik.tu-muenchen.de!news.csl-gmbh.net!blackbush.xlink.net!newsfeed.ision.net!ision!cpk-news-hub1.bbnplanet.com!news.gtei.net!newshub.northeast.verio.net!verio!news-feeds.jump.net!world!dp Xref: chonsp.franklin.ch alt.folklore.computers:61719 In article <39926ac5$0$159$39368dfe@news.twtelecom.net>, Philo wrote: >at the time i took the informal tour i had little knowledge of computers . >so my memories of that day may be slightly skewed... >but as i recall, the Cray1 had used TTL technology and had been surprised it >was not running CMOS (and lower power consumption) Actually they used ECL. Runs the transistors as (essentially) class A amplifiers. Very power hungry, as they never went into "cutoff". But the power consumption bought you speed you couldn't get anywhere else. It was not TTL/CMOS compatible, it used a single -5.2V power supply, with a much smaller swing for the logic signal. (there were level shifting buffers available to get you to the TTL universe -- don't forget to slow down the clock in the interface, so the other chips could keep up) At the time the Cray 1 was built, CMOS was the slowest but densest option (things like 2mhz 8080's were built using it), TTL was fast enough for most people (things like Vaxen were built from it), and ECL was only if cost really wasn't a concern. Besides all the heat, the clocks were fast enough that you needed RF designers to do your circut board layouts. It wasn't a bus you were building, it was a bunch of parallel transmission lines. Some designs used teflon dielectric, partly for its heat resistance, partly for its more uniform electrical properties. (and the heat and speed meant usual breadboard and debugging methods were not available -- putting a board on an "extender" was out of the question.) >another thing which surprised me was how standard the construction was. >the boards (though very closely spaced) used discrete TTL logic not too >terribly different (to my eyes) than standard technology. Well at the time, most machines were using some sort of dual inline package, with pins-thru-holes board mounting. The cray's used surface mount, which was pretty much restricted to MIL spec projects back then. Not sure if it was the original, but some of the machines used a heavy copper ground plane to conduct heat to the chilled water cooled board mounting rails/segment dividers. There was far too much heat for just air cooling. >The packages were smaller...perhaps similar to smt...if i recall...but the >devices themselves were all quite familiar 74xx (Don't recall TTL any more) Perhaps in the I/O interfaces, but the processor core was all ECL. > >But there was one thing which did *impress* me. .. the board traces... there >was often a slight , extra loop going to a gate here and there. >When I asked Ed why, he told me it was to allow for equal propigation time >to the gates. (Probably standard design practice but to someone like me with >low frequency design experience...impressive) The prop delay management didn't end with the boards. Every single wire pair in that curved backplane was a calibrated length. A standard part of a S. Cray lecture was handing out 30cm bits of wire. (aka: 1 ns long) I remember looking at the backplane of a Dec machine undergoing a repair some years ago (around the time you visited Cray). I looked at the printed circut board that it was made from, and pointed out to the technician some solder splashes that bridged a couple of traces, wondering how they got there and if they were responsible for its current toes up state... Turns out they were tuning stubs. You would hook up a scope, and move them with a soldering iron to get some of the internal timing signals to line up. >So...twenty years later...anyone have comments on my memory of the day. Also >does anyone know what the buss speeed and processor speed would have been of >that machine, as I have lost track of my friend over the years? Like all benchmarks between very different designs, the processor speed was pretty hard to relate to modern practices. Your best bet would be to compare "best case" floating point performance, with other machines. (some of his designs used the FP adder for integer math, as they operated in one cycle either way). They were "vector" machines, with multiple arithmetic units. They also were of the "very long instruction word" style that some still under development X86 successor processors will be using. Instructions were packed up to 4 to a word, and managing the utilization of various resources was part of the programmers workload. (the compilers of the day generally weren't up to the task. Normal practice was to examine what the Fortran compiler did, and hand tune the loops) One of the interesting speedups you got to think about was prefetch of data. All computations were register to register, and the load and store to memory instructions operated asynchronous to execution. Good programs had the load request several instructions prior to using the value fetched in a computation. Get it right, the data arrived before you used it, and lots of instructions got to overlap their execution. Get it wrong, and the instruction that used the result would stall. Next, instead of a more general cache, there was a special purpose one for instructions only, and if you could get your inner loop to fit into it (it was only about 80 instructions deep on the early ones) you went very fast indeed. -dp- Organizer, The New England Rubbish Deconstruction Society; The NERDS. We are the first US team to compete in the British Scrapheap Challenge series. (called "Junkyard Wars" when shown in the US) The NERDS New this week: FAQ, and "top ten" comparisons with Survivor and Iron Chef. This planet needs a lot more kids that think taking the lawnmowers' engine apart is more fun than playing nintendo. ###### Message-ID: <399319E9.F05066@plazaearth.com> Date: Thu, 10 Aug 2000 16:08:57 -0500 From: philo X-Mailer: Mozilla 4.51 [en] (X11; I; Linux 2.2.5-15 i586) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 18 NNTP-Posting-Host: 207.250.72.54 X-Trace: 965941790 news.twtelecom.net 156 207.250.72.54 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news.uni-stuttgart.de!fu-berlin.de!news.maxwell.syr.edu!nntp.inc.net!news.twtelecom.net!news.twtelecom.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:61727 thank you for your corrections Neil... i recalled so clearly that cmos was *not* used that in my mind i assumed it to be TTL or hTTL. another thing i recall now was that during testing, it was someone's job to run a wooden popsicle stick up and down the rows of boards to make sure there were no intermittant connections. since this was done with the case open and uncooled, the machine must have had some low power settings...although i have no idea how long the "popsicle stick' test was run or how long it would have taken the machine to overheat without the cooling. anyway i'm sure glad i got to see a bit of real history and will be able to tell the story more accurately. Philo btw: i recall how i pretty much *stopped* reading science fiction in 1964 when Isaac Asimov mentioned a desk-top computer. Interstellar travel , sure why not> Desk-top computer>>>holy cow, he's sure gone too far here:) ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39932ec0$1@news.ucsc.edu> Date: 10 Aug 2000 15:37:52 -0800 X-Trace: 10 Aug 2000 15:37:52 -0800, sundance.cse.ucsc.edu Lines: 61 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.stanford.edu!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:61715 I started in the X-MP era; I am a software type even though I have EE as a job title with a math background. In article <39926ac5$0$159$39368dfe@news.twtelecom.net>, Philo wrote: >but as i recall, the Cray1 had used TTL technology and had been surprised it >was not running CMOS (and lower power consumption) I believe that it was TTL, but it was a combination of emitter-coupled logic (ECL) and bi-polar logic for the speed. Why are you surprised? Elaborate. The machine was over 5 times faster at that time than any previous machine. If you were in the market for such a machine, you didn't quibble about cost (power), you went to IBM and ran on slower machines. CMOS was still being refined in many ways at that time. I think the ECL came from Fairchild like the ILLIAC ECL. >another thing which surprised me was how standard the construction was. Well, you didn't look close up. Not many machines used Freon to cool them. But lots of other aspects of the machine were off the shelf. That was deliberate. CRI had customers willing to pay, and the technology had not yet reached the refined level it has. Vaccuum tube computers were still in use during that era (1975). Reliability and quality were factors. >the boards (though very closely spaced) used discrete TTL logic not too >terribly different (to my eyes) than standard technology. >The packages were smaller...perhaps similar to smt...if i recall...but the >devices themselves were all quite familiar 74xx (Don't recall TTL any more) > >But there was one thing which did *impress* me. .. the board traces... there >was often a slight , extra loop going to a gate here and there. >When I asked Ed why, he told me it was to allow for equal propagation time >to the gates. (Probably standard design practice but to someone like me with >low frequency design experience...impressive) Quite a few machines of that period had little loops like that. I have photos of a Cyber 205 with even longer loops. The ILLIAC IV made miles of equilength ribbon cable. Nano seconds count. This attention to detail is what made CRI machines the best in the world. >So...twenty years later...anyone have comments on my memory of the day. Also >does anyone know what the buss speeed and processor speed would have been of >that machine, as I have lost track of my friend over the years? You have a generally decent memory of your observations. Bus? What's a bus? The clock was 12.5 ns for the earliest machines with 3-5 ticks for a fetch. The CPU had a direct connect to memory. Cray no need no 'tinkin' bus. 8^) I bet the next thing you will ask for is the cache size or virtual memory page size. And if you could afford, you could get any color you asked for. Almost. Ref: R. Russell CACM 1978. ###### From: bauer@shell3.ba.best.com (Jerry Bauer) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 10 Aug 2000 23:40:02 GMT Organization: Best Internet Communications, Inc. Lines: 61 Message-ID: <8mvegi$8c6$1@nntp1.ba.best.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> NNTP-Posting-Host: shell3.ba.best.com X-Trace: nntp1.ba.best.com 965950802 8582 206.184.139.134 (10 Aug 2000 23:40:02 GMT) X-Complaints-To: abuse@best.com NNTP-Posting-Date: 10 Aug 2000 23:40:02 GMT X-Newsreader: trn 4.0-test72+ked.01 (05 September 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!feed2.news.luth.se!luth.se!nntp.primenet.com!nntp.gblx.net!feeder.via.net!newshub2.rdc1.sfba.home.com!news.home.com!news2.best.com!nntp1.ba.best.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:61741 In article , Jeff DelPapa wrote: >In article <39926ac5$0$159$39368dfe@news.twtelecom.net>, >Philo wrote: >>at the time i took the informal tour i had little knowledge of computers . >>so my memories of that day may be slightly skewed... >>but as i recall, the Cray1 had used TTL technology and had been surprised it >>was not running CMOS (and lower power consumption) > <<< snipped >>> > >At the time the Cray 1 was built, CMOS was the slowest but densest >option (things like 2mhz 8080's were built using it), TTL was fast >enough for most people (things like Vaxen were built from it), and ECL >was only if cost really wasn't a concern. > <<< snipped >>> Actually, CMOS was not the most dense. LSI of that era was mainly done in NMOS. Chips like the 8080 and 6800 were done in a single-polysilicon, single-metal, NMOS process. Some others (mostly single-chip microcomputers) were done in PMOS. Possibly the most well-known CMOS processor family was the RCA 1800. CMOS lost out in the density department because, while half the transistors could be embedded in the substrate silicon, the other half had to be embedded in oppositely-doped "wells" in the substrate silicon. In PMOS and NMOS, all transistors were embedded in the substrate silicon. CMOS had the advantage in power consumption. In PMOS and NMOS, the driven output of each logic element is always "on" through a resistive element. (The primitive function of a resistive element is to convert current to heat.) Logic state changes from "driven" to "not driven" are accomplished by shunting the driving current to the substrate reference voltage (VSS). CMOS, because it had active elements, rather than resistive, changed from "driven" to "not driven" by turning off and on the appropriate transistor switches. PMOS and NMOS therefore had a relative high and constant power consumption (because typically, half of all logic elements are '1' and half are '0' at any given time). CMOS had a low and operating-frequency-dependent power consumption because power was only consumed briefly when a logic element changed state. The power consumption of a CMOS device is an almost linear function of the operating frequence. The driving force for density was chip size, a consequence of silicon defect density. As manufacturing processes matured and defect density decreased, larger chips became practical; silicon area became cheap, and heat dissipation became the greater problem. If you think a 1GHz CMOS Pentium is hot, consider what it would be like if half its transistors were replaced with resistors (i.e. heaters). There's a topic for the next "Hot Chips" symposium! Jerry Randal Bauer ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 11 Aug 2000 22:53:53 +0200 Organization: My own Private Self Lines: 61 Message-ID: <6ug0obpli6.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 966027233 628 10.0.3.2 (11 Aug 2000 20:53:53 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 11 Aug 2000 20:53:53 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:61791 dp@world.std.com (Jeff DelPapa) writes: Actually they used ECL. Runs the transistors as (essentially) class A > amplifiers. Very power hungry, as they never went into "cutoff". But > the power consumption bought you speed you couldn't get anywhere else. Yes. > It was not TTL/CMOS compatible, it used a single -5.2V power supply, Or in the case of the Cray1 an -1.3V an -5.2V dual power supply. IIRC the Cray 1 made its -1.3V by using assymmetric 3-pin dual resistors external to its chips (most likely so that the chips stayed cooler). The diagram I have here (about 10k ECL) also shows resistors between the output transistors emmiters and -5.2V, these must be the other symmetric 3-pin dual resistors that were mentioned in the Cray 1 manual. > with a much smaller swing for the logic signal. 0 = -0.9+-0.1V, 1 = -1.7+-0.1V, makes 0.8V swing. > At the time the Cray 1 was built, CMOS was the slowest but densest > option (things like 2mhz 8080's were built using it), 8080, z80, 6800, 6502 were all NMOS. 8080 and 4004 were PMOS. CMOS was used for 1802 and later only became mainstream when temperature problems grounded NMOS somewhere around 100-300'000 transistors. > then. Not sure if it was the original, but some of the machines used > a heavy copper ground plane to conduct heat to the chilled water > cooled board mounting rails/segment dividers. There was far too much > heat for just air cooling. That was already so in the Cray 1. > >The packages were smaller...perhaps similar to smt...if i recall...but the > >devices themselves were all quite familiar 74xx (Don't recall TTL any more) > > Perhaps in the I/O interfaces, but the processor core was all ECL. IIRC the Cray 1 had no I/O. It was connected as an auxillary processor to an VAX or similar size box. > This planet needs a lot more kids that think taking the lawnmowers' > engine apart is more fun than playing nintendo. All too true. -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39947fb8$1@news.ucsc.edu> Date: 11 Aug 2000 15:35:36 -0800 X-Trace: 11 Aug 2000 15:35:36 -0800, sundance.cse.ucsc.edu Lines: 35 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:61793 In article , Jeff DelPapa wrote: >ECL. >A standard part of a S. Cray lecture was handing out 30cm bits of >wire. (aka: 1 ns long) That was an Adm. Grace Hopper lecture. And her picoseconds are salt grains. >Like all benchmarks between very different designs, the processor >speed was pretty hard to relate to modern practices. The Cray architecture is one of the cleanest, easy to understand, architectures there. Few modes and minimal states. >They were "vector" machines, with multiple arithmetic units. They >also were of the "very long instruction word" style that some still >under development X86 successor processors will be using. No, the architecture while vector, is quite far from VLIW. I have a Multiflow Trace. A far more complex architecture from that era would be the CDC Star-100, Cyber203/205. >Organizer, The New England Rubbish Deconstruction Society; The >NERDS. We are the first US team to compete in the British Scrapheap >Challenge series. (called "Junkyard Wars" when shown in the US) href="http://www.the-nerds.org/">The NERDS New this week: FAQ, and >"top ten" comparisons with Survivor and Iron Chef. Very interesting. >This planet needs a lot more kids that think taking the lawnmowers' >engine apart is more fun than playing nintendo. Need to get them taking electronics apart. ###### From: Dennis Ritchie Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Sat, 12 Aug 2000 01:28:39 +0000 Organization: Bell Labs / Lucent Technologies Lines: 39 Message-ID: <3994A847.F120A8D9@bell-labs.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6ug0obpli6.fsf@chonsp.franklin.ch> NNTP-Posting-Host: trux.cs.bell-labs.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.61 [en] (WinNT; U) X-Accept-Language: en Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-koe1.dfn.de!news-han1.dfn.de!news-nue1.dfn.de!uni-erlangen.de!newsfeeds.belnet.be!news.belnet.be!colt.net!newspeer.clara.net!news.clara.net!news5-gui.server.ntli.net!ntli.net!uunet!ffx.uu.net!nntphub.cb.lucent.com!news Xref: chonsp.franklin.ch alt.folklore.computers:61800 Neil Franklin wrote (I snip much): > IIRC the Cray 1 had no I/O. It was connected as an auxillary processor > to an VAX or similar size box. > No, it had 12 quite fast (for the time) channels to disks. Later Cray 1 models used a peripheral-processor box akin to the CDC6600 PP. The Cray 1 I saw at LANL (early on; perphaps the first, though one never really knows about NSA deliveries), had a DG Nova as the system controller and console. But of course the real I/O never went through that. I won't try to recall out what's right and what's wrong about its logic design discussed earlier in this thread, except to say that I agree that it was ECL, mostly double-rail, and there were only a very few chip types. Also that inter-board timing was important (those blue-and-white twisted wires in the "waterfall" inside the C-shaped chassis were of carefully considered length.) The cooling, Cray legendarily and perhaps truthfully observed, was the important technical breakthrough. The boards were indeed two-sided with a central copper plate that connected thermally to the cast aluminum frame; this in turn was cooled by cold freon. What was the technical breakthrough? Well, cast aluminum is rather porous to freon, so the engineering was finding out how to bond, thermally, the stainless steel tubes that carried the freon to the aluminum frame. I've heard a story about a Cray 1 whose cooling freon supply failed, tragically coincidental with malfunction of the over-temperature sensors. I gather that it took only a couple of minutes before most of the chips had desoldered themselves. Dennis ###### From: "Philo" Newsgroups: alt.folklore.computers References: <39926ac5$0$159$39368dfe@news.twtelecom.net> Subject: Re: CRAY 1 Date: Sat, 12 Aug 2000 11:08:07 -0500 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2314.1300 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 Lines: 10 Message-ID: <399576a7$0$155$39368dfe@news.twtelecom.net> NNTP-Posting-Host: 207.250.192.197 X-Trace: 966096551 news.twtelecom.net 155 207.250.192.197 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!nntp.inc.net!news.twtelecom.net!news.twtelecom.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:61817 pricing? i know at the time ed mentioned the price of a cray1 but i forgot... don't know how many millions it was... but we used to joke how great ot would have been to have been a salesman for cray... we used to say: yeah at that price all you'd have to do is sell one a day... then you could just go home and goof off for the rest of the afternoon ;) ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399576a7$0$155$39368dfe@news.twtelecom.net> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399889a2$1@news.ucsc.edu> Date: 14 Aug 2000 17:06:58 -0800 X-Trace: 14 Aug 2000 17:06:58 -0800, sundance.cse.ucsc.edu Lines: 27 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!fu-berlin.de!cpk-news-hub1.bbnplanet.com!news.gtei.net!sunqbc.risq.qc.ca!falcon.america.net!nntp-relay.ihug.net!ihug.co.nz!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:61965 In article <399576a7$0$155$39368dfe@news.twtelecom.net>, Philo wrote: >pricing? >i know at the time ed mentioned the price of a cray1 but i forgot... >don't know how many millions it was... The first Cray 1 sold for just over $8M (8.8 costing 8.6). >but we used to joke how great to would have been to have been a salesman for >cray... >we used to say: >yeah at that price all you'd have to do is sell one a day... >then you could just go home and goof off for the rest of the afternoon ;) Were things so simple. One ends up buying or leasing systems. After the firm proved itself, and the CPU orders were assured the real costs had disk drive systems coming in far more expensive. The IBIS drives had formatted performance four times that of IBM Channel rates. There became a break even point where if you bought enough drives, you could have the Cray thrown in for free. I walked one of our PRs for a Y-MP and that PR for the entire system was $24M. And that's typical (more is even common). Our DEC salesman left DEC after selling his first VAX 9000. He stopped bothering to try to sell us one, and then he quit and moved to Siicon Graphics. That was an interesting shock to see him play the field. ###### From: "dls2" Newsgroups: alt.folklore.computers References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399576a7$0$155$39368dfe@news.twtelecom.net> <399889a2$1@news.ucsc.edu> Subject: Re: CRAY 1 Lines: 16 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Message-ID: Date: Tue, 15 Aug 2000 11:01:34 GMT NNTP-Posting-Host: 24.13.83.35 X-Complaints-To: abuse@home.net X-Trace: news1.rdc1.nj.home.com 966337294 24.13.83.35 (Tue, 15 Aug 2000 04:01:34 PDT) NNTP-Posting-Date: Tue, 15 Aug 2000 04:01:34 PDT Organization: @Home Network Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news-fra.pop.de!newsfeed.direct.ca!look.ca!newshub2.rdc1.sfba.home.com!news.home.com!news1.rdc1.nj.home.com.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:61950 "Eugene Miya" wrote: > Our DEC salesman left DEC after selling his first VAX 9000. Why? > He stopped bothering to try to sell us one, and then he quit > and moved to Siicon Graphics. That was an interesting shock > to see him play the field. Wow! -- Derrick Shearer ###### From: mwandel@nortel.ca (Markus Wandel) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 17 Aug 2000 19:13:57 GMT Organization: Nortel Lines: 26 Message-ID: <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> NNTP-Posting-Host: wcars0mc.ca.nortel.com Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-mue1.dfn.de!news.augsburg.net!news.idt.net!xfer10.netnews.com!xfe11.netnews.com!netnews.com!howland.erols.net!torn!qcarh002.nortelnetworks.com!bcarh189.ca.nortel.com!bmerhc5e.ca.nortel.com!bcarh8ab.ca.nortel.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62155 In article <399319E9.F05066@plazaearth.com>, philo wrote: >another thing i recall now was that during testing, it was someone's >job to run a wooden popsicle stick up and down the rows of boards to >make sure there were no intermittant connections. since this was done >with the case open and uncooled, the machine must have had some low >power settings...although i have no idea how long the "popsicle stick' >test was run or how long it would have taken the machine to overheat >without the cooling. This thing was not air cooled. As far as I recall reading somewhere, the liquid cooling lines ran in the wedge shaped parts between the columns of boards, and the boards were cooled by direct thermal conduction through the mounting rails. So running it with the covers off would have made no difference (actually made it run a little cooler due to air convection cooling). At least that's my take on things. Now my question... was there any reason for the CRAY-1 to actuallly be shaped like it was, rather than, for example, "flattened out"? Other than appearance that is. What a cool looking machine. The closest I've been to one is a dead one in a museum (Deutsches Museum in Munich, Germany... as I recall you can go right up to it and touch it.) Markus ###### Message-ID: <399C40B6.A19E1B12@uchicago.edu> From: Simon Allaway Organization: University of Chicago X-Mailer: Mozilla 4.74 [en] (Windows NT 5.0; U) X-Accept-Language: en MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 17 NNTP-Posting-Host: 128.135.145.16 X-Trace: uchinews 966541279 128.135.145.16 (Thu, 17 Aug 2000 14:41:19 CDT) NNTP-Posting-Date: Thu, 17 Aug 2000 14:41:19 CDT Date: Thu, 17 Aug 2000 14:44:54 -0500 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!loops.cs.wisc.edu!uchinews!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62106 Markus Wandel wrote: > > Now my question... was there any reason for the CRAY-1 to actuallly > be shaped like it was, rather than, for example, "flattened out"? > Other than appearance that is. What a cool looking machine. It struck me that having all the boards 'converge' in a central area kept the wires shorter. Therefore faster and potentially cooler. Kind of like all the plumbing being in the same spot within a (modern) house. Simon -- Simon Allaway | "It's not a firewall, University of Chicago | it's a leather pouch." Haskell Hall - M136 | - Anon. ###### From: wiss@eelwing.arda (Jonas Wissting) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Thu, 17 Aug 2000 23:19:45 +0200 Organization: Utfors AB Lines: 35 Message-ID: References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> NNTP-Posting-Host: md4692813.utfors.se Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: yggdrasil.utfors.se 966586229 18333 212.105.40.19 (18 Aug 2000 08:10:29 GMT) X-Complaints-To: abuse@utfors.se NNTP-Posting-Date: 18 Aug 2000 08:10:29 GMT X-Newsreader: knews 1.0b.1 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!feed2.news.luth.se!luth.se!newsfeeds.belnet.be!news.belnet.be!newsfeed00.sul.t-online.de!t-online.de!newsfeed.wirehub.nl!news.algonet.se!algonet!nntp.se.dataphone.net!newsfeed.sunet.se!news01.sunet.se!news.utfors.se!luthien.arda!nobody Xref: chonsp.franklin.ch alt.folklore.computers:62121 In article <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com>, mwandel@nortel.ca (Markus Wandel) writes: ... > Now my question... was there any reason for the CRAY-1 to actuallly > be shaped like it was, rather than, for example, "flattened out"? > Other than appearance that is. What a cool looking machine. > > The closest I've been to one is a dead one in a museum (Deutsches Museum > in Munich, Germany... as I recall you can go right up to it and touch it.) I walk past one -X/MP every day, and one -1 about once a week. There is a "cut" in the circle, it is not closed, so I guess one could make the box linear (to fit along a wall) with just a little amount of violence. Jonas -- http://wiss.unx.nu http://linux.unx.nu Another Glitch in the Call We don't need no indirection We don't need no flow control No data typing or declarations Did you leave the lists alone? Hey! Hacker! Leave those lists alone! Chorus: All in all, it's just a pure-LISP function call. All in all, it's just a pure-LISP function call. ###### From: bhk@dsl.co.uk (Brian {Hamilton Kelly}) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Fri, 18 Aug 2000 09:12:29 GMT Organization: Dragonhill Systems Ltd Message-ID: <966589949snz@dsl.co.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6ug0obpli6.fsf@chonsp.franklin.ch> X-Trace: mail2news.demon.co.uk 966618102 mail2news:17383 mail2news mail2news.demon.co.uk X-Complaints-To: abuse@demon.net X-Mail2News-Path: news.demon.net!dsl.demon.co.uk X-Newsreader: Demon Internet Simple News v1.31 Lines: 20 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!informatik.tu-muenchen.de!news.csl-gmbh.net!blackbush.xlink.net!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62186 In article <6ug0obpli6.fsf@chonsp.franklin.ch> neil@franklin.ch.remove "Neil Franklin" writes: > IIRC the Cray 1 had no I/O. It was connected as an auxillary processor > to an VAX or similar size box. ISTR that the Cray at Aldermaston had *two* 370/195 machines to handle all its I/O; however, I don't know what model this was. (Roughly 1976 +/-1 year) To my mind, the Cray is the only computer I've ever seen that looked like it belonged on the bridge of the Enterprise, with its padded horseshoe seat over the PSUs :-) -- Brian {Hamilton Kelly} bhk@dsl.co.uk "We have gone from a world of concentrated knowledge and wisdom to one of distributed ignorance. And we know and understand less while being incr- easingly capable." Prof. Peter Cochrane, BT Labs ###### From: mwandel@nortel.ca (Markus Wandel) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 18 Aug 2000 12:05:02 GMT Organization: Nortel Lines: 19 Message-ID: <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> NNTP-Posting-Host: wcars0mc.ca.nortel.com Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-koe1.dfn.de!news-fra1.dfn.de!news0.de.colt.net!colt.net!news.maxwell.syr.edu!logbridge.uoregon.edu!!nrchh45.us.nortel.com!zcarh46f.ca.nortel.com!bcarh8ac.ca.nortel.com!bcarh8ab.ca.nortel.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62154 In article <399C40B6.A19E1B12@uchicago.edu>, Simon Allaway wrote: >Markus Wandel wrote: >> >> Now my question... was there any reason for the CRAY-1 to actuallly >> be shaped like it was, rather than, for example, "flattened out"? >> Other than appearance that is. What a cool looking machine. > >It struck me that having all the boards 'converge' in a central area >kept the wires shorter. Therefore faster and potentially cooler. Kind of >like all the plumbing being in the same spot within a (modern) house. Yes, but from the look of the backplane (loose rat's nest of twisted pair wiring) the fractional inch of length that could be saved by curving the backplane rather than making a straight one, was not utilized. And of course as has been noted the machine is not a whole cylinder so you don't have the advantage of it being circular either. Markus ###### From: "Dave Babcock" Newsgroups: alt.folklore.computers References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> Subject: Re: CRAY 1 Lines: 24 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.50.4133.2400 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Message-ID: Date: Fri, 18 Aug 2000 13:43:13 GMT NNTP-Posting-Host: 209.204.154.28 X-Complaints-To: abuse@sonic.net X-Trace: typhoon.sonic.net 966606193 209.204.154.28 (Fri, 18 Aug 2000 06:43:13 PDT) NNTP-Posting-Date: Fri, 18 Aug 2000 06:43:13 PDT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-koe1.dfn.de!news-fra1.dfn.de!news0.de.colt.net!colt.net!newspeer1.nac.net!netnews.com!howland.erols.net!newsfeed.cwix.com!typhoon.sonic.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62153 "Markus Wandel" wrote in message news:8nhdhl$4vq$1@bcarh8ab.ca.nortel.com... > In article <399319E9.F05066@plazaearth.com>, > philo wrote: > > Now my question... was there any reason for the CRAY-1 to actuallly > be shaped like it was, rather than, for example, "flattened out"? > Other than appearance that is. What a cool looking machine. VIewed from the top both the Cray 1 & 2 are in the shape of a 'C' for Cray. > The closest I've been to one is a dead one in a museum (Deutsches Museum > in Munich, Germany... as I recall you can go right up to it and touch it.) As I think it's been mentioned in this thread before, the Computer Museum History Center in Mountain View California (www.computerhistory.org) has both a Cray 1 & Cray 2 on (static) display. They also have Cray's earlier CDC 6600 and 7600. DaveB ###### From: Tom Ivar Helbekkmo Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 19 Aug 2000 00:05:36 +0200 Lines: 21 Message-ID: <86g0o2dy33.fsf@athene.i.eunet.no> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com> NNTP-Posting-Host: athene.i.eunet.no Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: oslo-nntp.eunet.no 966636336 28044 193.71.2.52 (18 Aug 2000 22:05:37 GMT) X-Complaints-To: newsmaster@eunet.no NNTP-Posting-Date: 18 Aug 2000 22:05:37 GMT X-No-Archive: Yes User-Agent: Gnus/5.0807 (Gnus v5.8.7) Emacs/20.3 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news.nextra.ch!newsfeed1.online.no!newsfeed.online.no!sol.no!Norway.EU.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62144 mwandel@nortel.ca (Markus Wandel) writes: > >It struck me that having all the boards 'converge' in a central area > >kept the wires shorter. Therefore faster and potentially cooler. Kind of > >like all the plumbing being in the same spot within a (modern) house. > > Yes, but from the look of the backplane (loose rat's nest of twisted > pair wiring) the fractional inch of length that could be saved by > curving the backplane rather than making a straight one, was not > utilized. I've been told (during a tour of a machine room containing a Cray 1, during which I actually got to sit on the thing while it was running), that the point was that the shape allowed them to adjust individual wire lengths in order to get the right timing of signals travelling between different physical parts of the computer. -tih -- The basic difference is this: hackers build things, crackers break them. --Eric S. Raymond ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399da5ab$1@news.ucsc.edu> Date: 18 Aug 2000 14:07:55 -0800 X-Trace: 18 Aug 2000 14:07:55 -0800, sundance.cse.ucsc.edu Lines: 34 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62098 In article <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com>, Markus Wandel wrote: >This thing was not air cooled. As far as I recall reading somewhere, >the liquid cooling lines ran in the wedge shaped parts between the Freon. >columns of boards, and the boards were cooled by direct thermal conduction >through the mounting rails. So running it with the covers off would have >made no difference (actually made it run a little cooler due to air >convection cooling). At least that's my take on things. Temperature diffs could change the speed of the machine in multiple 0.1 ns. Not necessarily for the favorable. One did not want to run any of these machines w/o skins for sustained periods of time. The results would typically be inconsistencies in memory recovery. You had to running on these machines to appreciate the differences. (There's papers.) >Now my question... was there any reason for the CRAY-1 to actuallly >be shaped like it was, rather than, for example, "flattened out"? >Other than appearance that is. What a cool looking machine. Well the 7600 was a square C about 12 feet on a side. The overriding concern was the ability to work on these machines. But speed is also a concern. The time to repair was a major concern (the 2 had something like a 20-30 minute repair time which was amazing considering that this was immersion cooling). Seymour was a very functional guy. When you are the architect of the machine, you basically can make it look like anything you want. >The closest I've been to one is a dead one in a museum (Deutsches Museum >in Munich, Germany... as I recall you can go right up to it and touch it.) We keep a list in c.s.s. on places to see them. It's the 18th, so that panel of the FAQ went out today. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399da7d8$1@news.ucsc.edu> Date: 18 Aug 2000 14:17:12 -0800 X-Trace: 18 Aug 2000 14:17:12 -0800, sundance.cse.ucsc.edu Lines: 9 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!logbridge.uoregon.edu!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62095 In article <399C40B6.A19E1B12@uchicago.edu>, Simon Allaway wrote: >It struck me that having all the boards 'converge' in a central area >kept the wires shorter. Therefore faster and potentially cooler. Kind of >like all the plumbing being in the same spot within a (modern) house. There are other computers using a radial topology. I think first photos of the IBM 801 (ask Lynn) and the Convex C-3 are both radial. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399da982$1@news.ucsc.edu> Date: 18 Aug 2000 14:24:18 -0800 X-Trace: 18 Aug 2000 14:24:18 -0800, sundance.cse.ucsc.edu Lines: 23 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62089 In article , Jonas Wissting wrote: >I walk past one -X/MP every day, and one -1 about once a week. > >There is a "cut" in the circle, it is not closed, so I guess one could >make the box linear (to fit along a wall) with just a little amount of >violence. Not all X-MPs are C shaped. As CRI refined their engineering, and they were able to add multiple CPUs and more memory, the architecture shrunk in the available space. As their serious paying clients slowly saturated, they started to consider less serious markets who would pay for slightly smaller machines. The first Xs were C shaped, but we got S/N 313 X-MP (actually Cray-1M) which was a 12 model: 1 CPU, 2 MW memory, this was a half truncated C (no wings). The point's all moot now. Most current models are more boxy and rectangular. Js, Ts, and special projects are largely rectanular to accomodate their boards. The real trick was finding people, all women, in the Chippewa Falls and Mendotta Heights area who could fit into these inner area. CRI basically hired all they could in quite a radius of their area. Robotic assembly of Ys made their assembly jobs moot. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399da9cb$1@news.ucsc.edu> Date: 18 Aug 2000 14:25:31 -0800 X-Trace: 18 Aug 2000 14:25:31 -0800, sundance.cse.ucsc.edu Lines: 7 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.direct.ca!look.ca!hub.org!hub.org!news.gv.tsc.tdk.com!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62099 In article <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com>, Markus Wandel wrote: >And of course as has been noted the machine is not a whole >cylinder so you don't have the advantage of it being circular either. Y-MP/C-90s and the precursor to the C-1 were solid cylinders. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6ug0obpli6.fsf@chonsp.franklin.ch> <966589949snz@dsl.co.uk> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <399dac18$1@news.ucsc.edu> Date: 18 Aug 2000 14:35:20 -0800 X-Trace: 18 Aug 2000 14:35:20 -0800, sundance.cse.ucsc.edu Lines: 24 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!sunqbc.risq.qc.ca!falcon.america.net!news.gv.tsc.tdk.com!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62093 In article <966589949snz@dsl.co.uk>, Brian {Hamilton Kelly} wrote: >In article <6ug0obpli6.fsf@chonsp.franklin.ch> > neil@franklin.ch.remove "Neil Franklin" writes: >> IIRC the Cray 1 had no I/O. It was connected as an auxillary processor >> to an VAX or similar size box. A VAX is at best a batch submission engine. A DG Nova was the typical IOP until CRI made their own IOP and it was faster than a VAX. >ISTR that the Cray at Aldermaston had *two* 370/195 machines to handle >all its I/O; however, I don't know what model this was. (Roughly 1976 >+/-1 year) That's pretty damning of the 195. >To my mind, the Cray is the only computer I've ever seen that looked like >it belonged on the bridge of the Enterprise, with its padded horseshoe >seat over the PSUs :-) Well that's 25 years ago. Looks don't count for much unless you have the fastest functioning machine. It's not really a word processing engine. There are more interesting Cray projects than this older technology. ###### From: Dennis Ritchie Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Sat, 19 Aug 2000 06:36:50 +0000 Organization: Bell Labs / Lucent Technologies Lines: 30 Message-ID: <399E2B02.E6C35ED9@bell-labs.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399da982$1@news.ucsc.edu> NNTP-Posting-Host: trux.cs.bell-labs.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 4.61 [en] (WinNT; U) X-Accept-Language: en Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!wn4feed!wn3feed!worldnet.att.net!207.24.196.41!nntphub.cb.lucent.com!news Xref: chonsp.franklin.ch alt.folklore.computers:62115 Eugene Miya wrote, about construction of Cray 1 and (at least earlier X-MPs): > The point's all moot now. Most current models are more boxy and rectangular. > Js, Ts, and special projects are largely rectanular to accomodate their boards. > > The real trick was finding people, all women, in the Chippewa Falls and > Mendotta Heights area who could fit into these inner area. CRI > basically hired all they could in quite a radius of their area. > Robotic assembly of Ys made their assembly jobs moot. Some time ago, Peter Weinberger and I visited Cray Research in Mendota Heights (Minneapolis suburb, and far from rural) and also at Chippewa Falls, Wisconsin (much more rural) on a date I can't remember but could be radiocarbon aged at a point at which X-MPs were in full production along with some Cray 2s, and the Y-MP prototypes were being board-tested. A memorable visit, not least in seeing how the machines were actually constructed. Cray X-MPs (and presumably C-Is) assembled in the C shape had several vertical sub-chassis subtending 6 or 8 fractions of the circle, and most of the wiring of the backplane was within these. So during wiring between the pins of the backplane, it was like several people embroidering a fabric on the inside of small canoe. I talked to a couple of the workers--the ones on the XMP seemed cheerier, because a group of 3 or 4 would work together, and they could chat: the chassis was ~2m long. The Cray 2 folks, doing the same thing on a much smaller scale, had to work individually. Dennis ###### From: Charles Richmond Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Sat, 19 Aug 2000 11:01:43 -0700 Organization: Cannine Computer Center Lines: 27 Message-ID: <399ECB87.B86581AF@ev1.net> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com> <86g0o2dy33.fsf@athene.i.eunet.no> Reply-To: richmond@ev1.net X-Complaints-To: newsabuse@supernews.com X-Mailer: Mozilla 4.7C-CCK-MCD {C-UDP; EBM-APPLE} (Macintosh; I; PPC) X-Accept-Language: en MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news.uni-stuttgart.de!logbridge.uoregon.edu!newsfeed.stanford.edu!sn-xit-01!supernews.com!sn-inject-01!corp.supernews.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62179 Tom Ivar Helbekkmo wrote: > > mwandel@nortel.ca (Markus Wandel) writes: > > > >It struck me that having all the boards 'converge' in a central area > > >kept the wires shorter. Therefore faster and potentially cooler. Kind of > > >like all the plumbing being in the same spot within a (modern) house. > > > > Yes, but from the look of the backplane (loose rat's nest of twisted > > pair wiring) the fractional inch of length that could be saved by > > curving the backplane rather than making a straight one, was not > > utilized. > > I've been told (during a tour of a machine room containing a Cray 1, > during which I actually got to sit on the thing while it was running), > that the point was that the shape allowed them to adjust individual > wire lengths in order to get the right timing of signals travelling > between different physical parts of the computer. > Also, IMHO Seymour Cray had a great sense of style with his computers. I have heard Steve Jobs say that Bill Gates has *no* style...I do *not* think Mr. Jobs would say this about Seymour Cray... -- +-------------------------------------------------------------+ | Charles and Francis Richmond | +-------------------------------------------------------------+ ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 19 Aug 2000 20:45:27 +0200 Organization: My own Private Self Lines: 45 Message-ID: <6un1i984zc.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399da5ab$1@news.ucsc.edu> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 966710727 868 10.0.3.2 (19 Aug 2000 18:45:27 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 19 Aug 2000 18:45:27 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:62195 eugene@cse.ucsc.edu (Eugene Miya) writes: > In article <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com>, > Markus Wandel wrote: > > >Now my question... was there any reason for the CRAY-1 to actuallly > >be shaped like it was, rather than, for example, "flattened out"? > >Other than appearance that is. What a cool looking machine. Shorter connecting wires. They would be about twice as long in "linear" arrangement. > concern. The time to repair was a major concern (the 2 had something > like a 20-30 minute repair time which was amazing considering that this > was immersion cooling). This is also the reason for the C instead of full circle: so that someone could get in to the wires (while production and service). > >The closest I've been to one is a dead one in a museum (Deutsches Museum > >in Munich, Germany... as I recall you can go right up to it and touch it.) > > We keep a list in c.s.s. on places to see them. It's the 18th, so that > panel of the FAQ went out today. You can add this to the FAQ: The Deutsches Museum Cray ia a 1S, serial# 26, IIRC (seeing it in summer 1989). Also to add: the university I work for (Swiss Federal Institute of Technology in Zuerich) has a disused Cray standing in the entrance hall of the RZ building (contains the computer centre). We "traded it in" for the present C90 and then Cray decided it was not worth them transporting it back home, so they donated it as an exposition object. Unfortunately I do not have the exact type (XMP or YMP) nor serial, I will look the next time I am there (Thursday in 12 days). -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### From: bhk@dsl.co.uk (Brian {Hamilton Kelly}) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Fri, 18 Aug 2000 17:08:07 GMT Organization: Dragonhill Systems Ltd Message-ID: <966618487snz@dsl.co.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> X-Trace: mail2news.demon.co.uk 966716105 mail2news:18714 mail2news mail2news.demon.co.uk X-Complaints-To: abuse@demon.net X-Mail2News-Path: news.demon.net!dsl.demon.co.uk X-Newsreader: Demon Internet Simple News v1.31 Lines: 19 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!fu-berlin.de!newscore.gigabell.net!news0.de.colt.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62230 In article dave@kaleidosoft.com "Dave Babcock" writes: > As I think it's been mentioned in this thread before, the Computer Museum > History Center in Mountain View California (www.computerhistory.org) has > both a Cray 1 & Cray 2 on (static) display. They also have Cray's earlier > CDC 6600 and 7600. Was the latter the one that was built as a hollow cube, with a room inside for the engineer's console? ISTR hearing about such a beastie being brought in to supplant the IBM7094 at Imperial College, shortly after I stopped working there (in 1966). -- Brian {Hamilton Kelly} bhk@dsl.co.uk "We have gone from a world of concentrated knowledge and wisdom to one of distributed ignorance. And we know and understand less while being incr- easingly capable." Prof. Peter Cochrane, BT Labs ###### Message-ID: <399F71F6.5FC6@bellsouth.net> From: futurebots Reply-To: fuboco@bellsouth.net Organization: Future-Bot Components X-Mailer: Mozilla 3.01C-BLS20 (Win95; U; 16bit) MIME-Version: 1.0 Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Lines: 32 Date: Sat, 19 Aug 2000 22:51:50 -0700 NNTP-Posting-Host: 209.214.15.71 X-Trace: news3.mia 966743682 209.214.15.71 (Sat, 19 Aug 2000 23:54:42 EDT) NNTP-Posting-Date: Sat, 19 Aug 2000 23:54:42 EDT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news.uni-stuttgart.de!rz.uni-karlsruhe.de!schlund.de!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.atl.bellsouth.net.MISMATCH!newsfeed.atl!news3.mia.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62199 Hello all, I have a few of an INTEL i80860XP-50 processor, it had 2.5 million transistors this was a 64 bit Risc processor that can out back in the early 90's. This chip was refured as a CRAY on a chip by BYTE or one other of the magazines at that time, it was fast !! The Navy and other military units used it for Radar and Sonar stuff. It's in a 168 pin PGA package, it can do 100 MFLOPS single precision, 64 bit data bus, 32 bit address bus, 64 bit 3D graphics unit, 4 K instruction cache, 8 K dat cache, can do 3 operations per clock, and will run Unix.. I don't understand why it took INTEL almost 10 years more to just come out with it's new 64 bit Withrolet? Windows processor at what 1 Ghz ? I would have liked to see the 80860XP progress through the years running at 1 GHZ or more today, running let's say LINUX, WOW, but iNTEL stop producing it back in the mid 90's asfare as I know, or still produce it but never increased it clock from the 50 Mhz max.. If anybody know more about this processor I would love to know more.. Thanks -- Dan Mathias -------------------------- Future-Bot Components Phone/Fax (561) 575-1487 106 Commerce way, A8 http://www.futurebots.com Jupiter, Fl. 33458 USA Email: fuboco@bellsouth.net ----------------------------------------------------------------------- Robotic and Electronic Components for the Hobbyist and Professional.. ###### From: andrew@cucumber.demon.co.uk (Andrew Gabriel) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 20 Aug 2000 09:26:47 GMT Organization: home Message-ID: <8no88n$bb5@cucumber.demon.co.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> NNTP-Posting-Host: cucumber X-NNTP-Posting-Host: cucumber.demon.co.uk:158.152.58.86 X-Trace: news.demon.co.uk 966770081 nnrp-01:25269 NO-IDENT cucumber.demon.co.uk:158.152.58.86 X-Complaints-To: abuse@demon.net X-Newsreader: knews 0.9.6 Lines: 19 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!newsgate.cistron.nl!bullseye.news.demon.net!demon!news.demon.co.uk!demon!cucumber.demon.co.uk!usenet Xref: chonsp.franklin.ch alt.folklore.computers:62212 In article <399F71F6.5FC6@bellsouth.net>, futurebots writes: >I have a few of an INTEL i80860XP-50 processor, it had 2.5 million >transistors this was a 64 bit Risc processor that can out back in the >early 90's. >This chip was refured as a CRAY on a chip by BYTE or one other of the >magazines at that time, it was fast !! The Navy and other military units >used it for Radar and Sonar stuff. >It's in a 168 pin PGA package, it can do 100 MFLOPS single precision, 64 >bit data bus, 32 bit address bus, 64 bit 3D graphics unit, 4 K >instruction cache, 8 K dat cache, can do 3 operations per clock, and >will run Unix.. But will it run MS-DOS, MS-flight Simulator, ... ? -- Andrew Gabriel Consultant Software Engineer ###### From: Luis Fernandes Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 20 Aug 2000 10:11:52 -0400 Organization: Ryerson Polytechnic University Lines: 11 Sender: elf@genesis Message-ID: References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399C40B6.A19E1B12@uchicago.edu> <8nj8pe$bbo$1@bcarh8ab.ca.nortel.com> <86g0o2dy33.fsf@athene.i.eunet.no> <399ECB87.B86581AF@ev1.net> NNTP-Posting-Host: sulfur.ee.ryerson.ca X-Newsreader: Gnus v5.6.42/Emacs 19.34 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!torn!news.ryerson.ca!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62210 >>>>> "richmond" == Charles Richmond writes: richmond> Also, IMHO Seymour Cray had a great sense of style with richmond> his computers. I have heard Steve Jobs say that Bill richmond> Gates has *no* style...I do *not* think Mr. Jobs would richmond> say this about Seymour Cray... In an interview that Cray did for the Smithsonian, he said this, "I've enjoyed the aesthetics part of building computers because it's any extra little thing you add that is clearly your own personality being projected in the product." ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 20 Aug 2000 22:43:44 +0200 Organization: My own Private Self Lines: 100 Message-ID: <6uu2cfhddr.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 966804224 472 10.0.3.2 (20 Aug 2000 20:43:44 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 20 Aug 2000 20:43:44 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:62232 futurebots writes: > I have a few of an INTEL i80860XP-50 processor, it had 2.5 million > transistors this was a 64 bit Risc processor that can out back in the > early 90's. 1 mio transistors. It came out in 1989 a few months before the 486 (which has 1.08 mio transistors). Originally ran at 33MHz. > This chip was refured as a CRAY on a chip by BYTE [search index and magazine] Byte December 1989 page 333 says: See "Intel's Cray-on-a-Chip" in the May Byte. > magazines at that time, it was fast !! The Navy and other military units > used it for Radar and Sonar stuff. NeXT used it for the graphics accelerator in its NeXTdimension video card. I had one of them. > It's in a 168 pin PGA package, it can do 100 MFLOPS single precision, To be more precise: 1 FLOP per clock cycle (=50 MFLOPs) but with one possible operation being the combination of multiplication and addition that gives a max 100 MFLOPs when doing jobs where all multiplicaions are followed by an addition of their result to something else. > instruction cache, 8 K dat cache, can do 3 operations per clock, The 2 combined float ops plus one "linked" integer op (usually an address calculation to fetch the next operand). The 860 was a vector processor. So it only reached this speed when ploughing through an vector. Actually the 850 floating unit was not larger than the 486es one, the difference 2..3->33..50 MFLOP was the result of the vector organisation of the 860. Actually all this (I)SSE stuff of the newer PII processors is a reappearance of this vectoring in the x86 architecture. > and will run Unix.. Unix will run on anything. And actually any OS runs lousily on an 860. The multi-100-cycle interrupt handling was truly awfull. Intel actually specified an 10 page (!) code in their manual, which any interupt had to handle to unroll all possble states of vector processing without state loss. Deviating from it would introduce suble bugs into some random type of calculation. P.S: AFAIK Windos NT was named so because it was originally intended to run on the N10 (=860) processor. MS gave up on it. Later some firms combined 486+860 on one motherboard running MS-DOS on 486 and calculations on 860. > I don't understand why it took INTEL almost 10 years more to just come > out with it's new 64 bit Withrolet? Windows processor at what 1 Ghz ? 64bit is really irrelevant, so long you have no data that requires 64bit calculations. Today processes larger than 4GByte are making >32bit address calculations neccessary. Before that only graphics manipulation (the 860s original intention) and scientific calculations (the 860s usage) needed this width. > I would have liked to see the 80860XP progress through the years running > at 1 GHZ or more today, running let's say LINUX, WOW, but iNTEL stop > producing it back in the mid 90's asfare as I know, or still produce it > but never increased it clock from the 50 Mhz max.. It didn't sell in millions each month. > If anybody know more about this processor I would love to know more.. From my books database: [Dewar] Robert Dewar/Matthew Smosna, Microprocessors - A Programmers View McGraw-Hill, 1990, ISBM 0-07-016638-2 A detailed, thorough, but easy to read text comparing todays 32 bit proc. It covers 80386/486, 68030/40, Mips, Sparc, 80860, Power and Transputer. -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### From: abuse@cabal.org.uk (Peter Corlett) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 21 Aug 2000 01:30:11 GMT Organization: None Lines: 19 Message-ID: <8nq0n3$njn$1@vindaloo.cabal.org.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> <6uu2cfhddr.fsf@chonsp.franklin.ch> NNTP-Posting-Host: modem-51.amlach.dialup.pol.co.uk X-Trace: newsg1.svr.pol.co.uk 966878574 1540 62.136.104.179 (21 Aug 2000 17:22:54 GMT) NNTP-Posting-Date: 21 Aug 2000 17:22:54 GMT X-Complaints-To: abuse@theplanet.net X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!diablo.theplanet.net!news.theplanet.net!newspost.theplanet.net!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62248 Neil Franklin wrote: [...] > To be more precise: 1 FLOP per clock cycle (=50 MFLOPs) but with one > possible operation being the combination of multiplication and addition > that gives a max 100 MFLOPs when doing jobs where all multiplicaions are > followed by an addition of their result to something else. To be fair, Multiply Accumulate is rather a useful operation in DSP work, so for some such algorithms, you may find it done more often than not. Can the processor also clip calculations at the largest positive or negative result instead of wrapping? [...] > Unix will run on anything. I'm not aware of any form of Unix that runs on the Z80 boxen I have to hand, although scarily, I've got the source code to a TCP/IP stack for a ZX Spectrum. ###### Newsgroups: alt.folklore.computers From: null@lycosmail.com (Fred Wedemeier) Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> X-Newsreader: News Xpress 2.01 Lines: 25 Message-ID: X-Abuse-Info: Please be sure to forward a copy of ALL headers X-Abuse-Info: Otherwise we will be unable to process your complaint properly X-Complaints-To: support@usenetserver.com NNTP-Posting-Date: Mon, 21 Aug 2000 01:34:05 EDT Organization: WebUseNet Corp http://www.usenetserver.com - Home of the fastest NNTP servers on the Net. Date: Mon, 21 Aug 2000 05:33:42 GMT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news.uni-stuttgart.de!rz.uni-karlsruhe.de!schlund.de!newsfeed01.sul.t-online.de!t-online.de!blackbush.xlink.net!npeer.kpnqwest.net!news-peer.gip.net!news.gsl.net!gip.net!cyclone2.usenetserver.com!news-out.usenetserver.com!cyclone1.usenetserver.com!news-west.usenetserver.com.POSTED!fcw138 Xref: chonsp.franklin.ch alt.folklore.computers:62235 In article <399F71F6.5FC6@bellsouth.net>, fuboco@bellsouth.net wrote: > >I don't understand why it took INTEL almost 10 years more to just come >out with it's new 64 bit Withrolet? Windows processor at what 1 Ghz ? > >I would have liked to see the 80860XP progress through the years running >at 1 GHZ or more today, running let's say LINUX, WOW, but iNTEL stop >producing it back in the mid 90's asfare as I know, or still produce it >but never increased it clock from the 50 Mhz max.. > >If anybody know more about this processor I would love to know more.. Andrew Grove wrote a few paragraphs about the situation in "Only the Paranoid Survive." Intel was selling three processor lines and got concerned about the cost to sustain all three and with the overlap between the x86 and 860 potentially confusing their customers. The 860 lost. Given that the business exists to make $$$ rather than fast processors for the sake of making fast processors, it probably wasn't a bad decision. -- --------------------------------------------------- best regards, Fred Wedemeier null@lycosmail.com ###### Path: chonsp.franklin.ch!not-for-mail From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 21 Aug 2000 22:01:58 +0200 Organization: My own Private Self Lines: 42 Message-ID: <6ud7j2ids9.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> <6uu2cfhddr.fsf@chonsp.franklin.ch> <8nq0n3$njn$1@vindaloo.cabal.org.uk> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 966888118 678 10.0.3.2 (21 Aug 2000 20:01:58 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 21 Aug 2000 20:01:58 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:62268 abuse@cabal.org.uk (Peter Corlett) writes: > Neil Franklin wrote: > [...] > > To be more precise: 1 FLOP per clock cycle (=50 MFLOPs) but with one > > possible operation being the combination of multiplication and addition > > that gives a max 100 MFLOPs when doing jobs where all multiplicaions are > > followed by an addition of their result to something else. > > To be fair, Multiply Accumulate is rather a useful operation in DSP work, so > for some such algorithms, you may find it done more often than not. Sure. Also some matrix stuff. > Can the > processor also clip calculations at the largest positive or negative result > instead of wrapping? I have no information on that. > [...] > > Unix will run on anything. > > I'm not aware of any form of Unix that runs on the Z80 boxen I have to hand, > although scarily, I've got the source code to a TCP/IP stack for a ZX > Spectrum. Unix was developed on an pdp11/20 (with only 24k RAM), which is roughly in the same ballpark as the Z80. I have actually seen an Unix-like OS with TCP/IP for an 6502 (GeckOS http://www.6502.org/users/andre/osa/index.html). And the developers of the ELKS 8088 Linux once announced that the were going for MMUed Z80s (Z280?), but their web site says nothing about it. -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39a16f34$1@news.ucsc.edu> Date: 21 Aug 2000 11:04:36 -0800 X-Trace: 21 Aug 2000 11:04:36 -0800, sundance.cse.ucsc.edu Lines: 23 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-mue1.dfn.de!news-nue1.dfn.de!uni-erlangen.de!newsfeeds.belnet.be!news.belnet.be!news.tele.dk!171.64.14.106!newsfeed.stanford.edu!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62274 In article <966618487snz@dsl.co.uk>, Brian {Hamilton Kelly} wrote: >In article > dave@kaleidosoft.com "Dave Babcock" writes: >> As I think it's been mentioned in this thread before, the Computer Museum >> History Center in Mountain View California (www.computerhistory.org) has >> both a Cray 1 & Cray 2 on (static) display. They also have Cray's earlier >> CDC 6600 and 7600. > >Was the latter the one that was built as a hollow cube, with a room >inside for the engineer's console? ISTR hearing about such a beastie >being brought in to supplant the IBM7094 at Imperial College, shortly >after I stopped working there (in 1966). Well the 7600 didn't have a ceiling to really be called a room. We only have two corners of the square out, and the copper tubing is exposed. The walls/sides on this are clear blue tint plexiglas. There is a photo of the installed machine in place, and I might have seen this machine in 1982 on my first visit to LCC. And the S/N is 1. It was LLNL's in its past life. And the operator console wasn't in the center. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399da5ab$1@news.ucsc.edu> <6un1i984zc.fsf@chonsp.franklin.ch> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39a172b6$1@news.ucsc.edu> Date: 21 Aug 2000 11:19:34 -0800 X-Trace: 21 Aug 2000 11:19:34 -0800, sundance.cse.ucsc.edu Lines: 53 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.stanford.edu!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62273 >> concern. The time to repair was a major concern (the 2 had something >> like a 20-30 minute repair time which was amazing considering that this >> was immersion cooling). In article <6un1i984zc.fsf@chonsp.franklin.ch>, Neil Franklin wrote: >This is also the reason for the C instead of full circle: so that >someone could get in to the wires (while production and service). Well, I'm not certain that's fully precisely the case. Neither the 3 nor the 4 were C shaped and the successor SRC-6 is quite rectangular. The 3 was an octagon. The 4 was a cube about the size of the NeXT cube I am told (I know a little more than this, brochures were produced, but I did not see the final prototype, but we are in touch, but I also have reasons for not elaborating). >> >The closest I've been to one is a dead one in a museum (Deutsches Museum >> >in Munich, Germany... as I recall you can go right up to it and touch it.) >> >> We keep a list in c.s.s. on places to see them. It's the 18th, so that >> panel of the FAQ went out today. > >You can add this to the FAQ: The Deutsches Museum Cray ia a 1S, It's there. Sitting on C 1s is very popular. >serial# 26, IIRC (seeing it in summer 1989). I added this. Thanks. >Also to add: the university I work for (Swiss Federal Institute of >Technology in Zuerich) has a disused Cray standing in the entrance >hall of the RZ building (contains the computer centre). Depending on time and travel requirements I might be able to check this machine in December. Right now the facility in Lugano is the primary place that I'm being asked to visit followed by Lausanne and then Zurich. >We "traded it in" for the present C90 and then Cray decided it was not >worth them transporting it back home, so they donated it as an >exposition object. > >Unfortunately I do not have the exact type (XMP or YMP) nor serial, I >will look the next time I am there (Thursday in 12 days). If you get the serial; it is also sometimes useful to not the color of "the skins." There's a slew of different Cray chassies in Lausanne. About 4 of them. ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <399da982$1@news.ucsc.edu> <399E2B02.E6C35ED9@bell-labs.com> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39a173b0$1@news.ucsc.edu> Date: 21 Aug 2000 11:23:44 -0800 X-Trace: 21 Aug 2000 11:23:44 -0800, sundance.cse.ucsc.edu Lines: 35 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-mue1.dfn.de!news.augsburg.net!nntp.primenet.com!nntp.gblx.net!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62270 In article <399E2B02.E6C35ED9@bell-labs.com>, Dennis Ritchie wrote: >Eugene Miya wrote, about construction of Cray 1 and (at least earlier X-MPs): ... >Some time ago, Peter Weinberger and I visited Cray Research in Mendota Heights >(Minneapolis suburb, and far from rural) and also at Chippewa Falls, Wisconsin >(much more rural) on a date I can't remember but could be radiocarbon >aged at a point at which X-MPs were in full production along with >some Cray 2s, and the Y-MP prototypes were being board-tested. > >A memorable visit, not least in seeing how the machines were actually >constructed. Cray X-MPs (and presumably C-Is) assembled in the C shape >had several vertical sub-chassis subtending 6 or 8 fractions of the >circle, and most of the wiring of the backplane was within these. >So during wiring between the pins of the backplane, it was like several >people embroidering a fabric on the inside of small canoe. I talked >to a couple of the workers--the ones on the XMP seemed cheerier, because >a group of 3 or 4 would work together, and they could chat: the chassis >was ~2m long. The Cray 2 folks, doing the same thing on a much smaller >scale, had to work individually. We were just discussing some of that c.s.s. Search on the X-MP thread about Rollwagon's contracted book on the manufacturing facilities in C.F. I recall that a few post cards were also reproduced, but the book on the assembly people at CF was B&W. The count was running 4-5 to 1 against the book. But that a copy sold for a lot of money on eBay. That there was 2s says that you visited in the mid-80s. I never had a chance to visit C.F. or M.H. but I did do a Pascal Standards meeting in the Twin Cities. CDC, even as a small company was quite a visible presence. ###### From: Jim Stewart Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Tue, 22 Aug 2000 10:49:18 -0700 Organization: http://www.jkmicro.com Lines: 34 Message-ID: <6882E44C3DC0151C.C0D7AAF356069DB2.E1F892CEA38AF2AE@lp.airnews.net> X-Orig-Message-ID: <39A2BD1D.4157DDA0@jkmicro.com> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> Reply-To: jstewart@jkmicro.com Abuse-Reports-To: abuse at airmail.net to report improper postings NNTP-Proxy-Relay: 204.181.96.50 NNTP-Posting-Time: Tue Aug 22 12:54:07 2000 NNTP-Posting-Host: !a`Ja1k-Vss-34&&7g-9(LqMP (Encoded at Airnews!) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: alt.folklore.computers X-Mailer: Mozilla 4.74 [en] (Win98; U) X-Accept-Language: en Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!cpk-news-hub1.bbnplanet.com!news.gtei.net!howland.erols.net!news-out.nntp.airnews.net.MISMATCH!cabal10.airnews.net!news.airnews.net!cabal12.airnews.net!news.airnews.net!cabal14.airnews.net!news.airnews.net!cabal2.airnews.net!news-f.iadfw.net!jkmicro.com!anonymous Xref: chonsp.franklin.ch alt.folklore.computers:62395 futurebots wrote: > > Hello all, > > I have a few of an INTEL i80860XP-50 processor, it had 2.5 million > transistors this was a 64 bit Risc processor that can out back in the > early 90's. > This chip was refured as a CRAY on a chip by BYTE or one other of the > magazines at that time, it was fast !! The Navy and other military units > used it for Radar and Sonar stuff. > It's in a 168 pin PGA package, it can do 100 MFLOPS single precision, 64 > bit data bus, 32 bit address bus, 64 bit 3D graphics unit, 4 K > instruction cache, 8 K dat cache, can do 3 operations per clock, and > will run Unix.. > > I don't understand why it took INTEL almost 10 years more to just come > out with it's new 64 bit Withrolet? Windows processor at what 1 Ghz ? > > I would have liked to see the 80860XP progress through the years running > at 1 GHZ or more today, running let's say LINUX, WOW, but iNTEL stop > producing it back in the mid 90's asfare as I know, or still produce it > but never increased it clock from the 50 Mhz max.. > > If anybody know more about this processor I would love to know more.. If you're *REALLY* interested in it, I'll sell you my Intel ISPC 860. It's a phone booth sized 'supercomputer' with 36 i80860 processors in a hypercube configuration, along with DAT tape, 2 SCSI disk drives, and a BSD Unix console running on a 386 PC. You'll probably need an electrician to power it up, it needs 220 volts at 30 amperes. Jim ###### From: bhk@dsl.co.uk (Brian {Hamilton Kelly}) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: Wed, 23 Aug 2000 17:21:17 GMT Organization: Dragonhill Systems Ltd Message-ID: <967051277snz@dsl.co.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <39a16f34$1@news.ucsc.edu> X-Trace: mail2news.demon.co.uk 967070877 mail2news:7205 mail2news mail2news.demon.co.uk X-Complaints-To: abuse@demon.net X-Mail2News-Path: news.demon.net!dsl.demon.co.uk X-Newsreader: Demon Internet Simple News v1.31 Lines: 19 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.tesion.net!news.belwue.de!news-stu1.dfn.de!news-mue1.dfn.de!news-nue1.dfn.de!uni-erlangen.de!newsfeeds.belnet.be!news.belnet.be!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:62522 In article <39a16f34$1@news.ucsc.edu> eugene@cse.ucsc.edu "Eugene Miya" writes: [apropos the 7600] > And the operator console wasn't in the center. I didn't say "operator's console", I said "engineer's console". IIRC, that machine used to use spare cycles (which would otherwise have been wasted by a NULL Process) in running diagnostics: stories (probably apocryphal) had it that it would detect a potential failure mode and say "diode xyz123 on board nnn is " and that this came out on a separate teletype used by the engineer(s). -- Brian {Hamilton Kelly} bhk@dsl.co.uk "We have gone from a world of concentrated knowledge and wisdom to one of distributed ignorance. And we know and understand less while being incr- easingly capable." Prof. Peter Cochrane, BT Labs ###### From: andrew@cucumber.demon.co.uk (Andrew Gabriel) Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 26 Aug 2000 18:33:18 GMT Organization: home Message-ID: <8o92he$i4h@cucumber.demon.co.uk> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6uog30gaj4.fsf@chonsp.franklin.ch> <399319E9.F05066@plazaearth.com> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <966618487snz@dsl.co.uk> <399F71F6.5FC6@bellsouth.net> NNTP-Posting-Host: cucumber X-NNTP-Posting-Host: cucumber.demon.co.uk:158.152.58.86 X-Trace: news.demon.co.uk 967319137 nnrp-14:1076 NO-IDENT cucumber.demon.co.uk:158.152.58.86 X-Complaints-To: abuse@demon.net X-Newsreader: knews 0.9.6 Lines: 15 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!newsfeed.Austria.EU.net!nmaster.kpnqwest.net!blackbush.xlink.net!bignews.mediaways.net!diablo.netcom.net.uk!netcom.net.uk!dispose.news.demon.net!demon!news.demon.co.uk!demon!cucumber.demon.co.uk!usenet Xref: chonsp.franklin.ch alt.folklore.computers:62626 In article <399F71F6.5FC6@bellsouth.net>, futurebots writes: >Hello all, > >I have a few of an INTEL i80860XP-50 processor, it had 2.5 million >transistors this was a 64 bit Risc processor that can out back in the >early 90's. Stratus used it to build SVR4 unix systems, running their FTX2. Later they switched to hp-pa (e.g. Continuum Model 610) running FTX3. -- Andrew Gabriel Consultant Software Engineer ###### From: Neil Franklin Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 Date: 01 Sep 2000 01:27:47 +0200 Organization: My own Private Self Lines: 58 Message-ID: <6uk8cxf1u4.fsf@chonsp.franklin.ch> References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <8nhdhl$4vq$1@bcarh8ab.ca.nortel.com> <399da5ab$1@news.ucsc.edu> <6un1i984zc.fsf@chonsp.franklin.ch> <39a172b6$1@news.ucsc.edu> NNTP-Posting-Host: chonsp.franklin.ch X-Trace: chonsp.franklin.ch 967764467 328 10.0.3.2 (31 Aug 2000 23:27:47 GMT) X-Complaints-To: news@chonsp.franklin.ch NNTP-Posting-Date: 31 Aug 2000 23:27:47 GMT X-Newsreader: Gnus v5.7/Emacs 20.4 Xref: chonsp.franklin.ch alt.folklore.computers:62802 eugene@cse.ucsc.edu (Eugene Miya) writes: > Neil Franklin wrote: > > >Also to add: the university I work for (Swiss Federal Institute of > >Technology in Zuerich) has a disused Cray standing in the entrance > >hall of the RZ building (contains the computer centre). > > Depending on time and travel requirements I might be able to check this > machine in December. Right now the facility in Lugano is the primary > place that I'm being asked to visit followed by Lausanne and then Zurich. > > >We "traded it in" for the present C90 and then Cray decided it was not > >worth them transporting it back home, so they donated it as an > >exposition object. > > > >Unfortunately I do not have the exact type (XMP or YMP) nor serial, I > >will look the next time I am there (Thursday in 12 days). > > If you get the serial; it is also sometimes useful to not the color of > "the skins." OK. It is (OK, was) thursday. I have been there, looked it up. Type: Cray X-MP/28 Serial No: 420 Colour: the "tower" is yellow with maroon "seatbench" From the 3 info boards there: Board 1 (on "tower"): Processor: 8.5ns clock 16 gate array technology (whatever that is) 3 sets of interprocessor registers for clustering 2 CPUs Memory: MOS, 8MWords, 64bit words, 32 banks, 76ns 24bit Addresses, 4 ports per CPU Board 2 (on "tower"): 3 IOPs with 128k local memory and 4MWords MOS common buffer memory 100 MByte/s transfer IOP->X-MP 3 front ends: DEC VAX 8350 Cyber 180-853 Sun 3 gateway to TCP/IP Mass Storage: DD49 1.2GByte, 9.8MByte/s 16ms DD40 21GByte, 12MByte/s 16ms Board 3 (at entrance): In Use: 1988.07.04-1991.04.10 Replaced by: Cray Y-MP4D/264 (so not the preset C90 or J90). -- Neil Franklin, neil@franklin.ch.remove http://neil.franklin.ch/ Nerd, Geek, Hacker, Unix Guru, Sysadmin, Roleplayer, LARPer, Mystic ###### Newsgroups: alt.folklore.computers Subject: Re: CRAY 1 References: <39926ac5$0$159$39368dfe@news.twtelecom.net> <6un1i984zc.fsf@chonsp.franklin.ch> <39a172b6$1@news.ucsc.edu> <6uk8cxf1u4.fsf@chonsp.franklin.ch> Organization: UC Santa Cruz CIS/CE From: eugene@cse.ucsc.edu (Eugene Miya) NNTP-Posting-Host: sundance.cse.ucsc.edu Message-ID: <39aef457$1@news.ucsc.edu> Date: 31 Aug 2000 17:12:07 -0800 X-Trace: 31 Aug 2000 17:12:07 -0800, sundance.cse.ucsc.edu Lines: 6 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newsmaster-01.atnet.at!newsmaster-03.atnet.at!atnet.at!newsrouter.chello.at!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.stanford.edu!newsfeed.berkeley.edu!ucberkeley!news.ucsc.edu!eugene Xref: chonsp.franklin.ch alt.folklore.computers:62818 In article <6uk8cxf1u4.fsf@chonsp.franklin.ch>, Neil Franklin wrote: ... I amended the c.s.s FAQ with your data and trust Deja to archive the rest of your post.