VLIW pre-history
|
From: Eugene Miya on 18 May 2007 13:20 In article <dontlikespan-EF71FE.06082818052007(a)newsgroups.comcast.net>, Alan Charlesworth <dontlikespan(a)nowhere.com> wrote: >Nope. I stayed at FPS until it went under and its assets were bought by >Cray in 1992. Those assets later turned into the Starfire E10K, which >were sold to Sun when SGI bought Cray in 1996 -- but that is a different >story. As for the i860, FPS used in a matrix coprocessor board, which >was an add-in to the FPS-164, a 64-bit follow-on to the AP-120B. Alan, you are the kind of person, like Josh Fisher, the CHM needs to take your history, oral and written (we got Josh's), and get your help to locate and preserve samples of all these boxes (well, we have enough i860s). Don't forget the Tesseract (which was Steve Stevenson's basis for starting comp.hypercube) and the 264. One friend from CRI went to SUN and took the E10K line. Small world. --
From: Eugene Miya on 18 May 2007 13:21 In article <1179466666.901608.324210(a)o5g2000hsb.googlegroups.com>, Quadibloc <jsavard(a)ecn.ab.ca> wrote: >I'll agree that this isn't true VLIW, just "vaguely VLIW-ish", as >you've put it. But with a dataflow architecture, the program counter >is a means and not an end, so the relationship may be closer than it >seems. What kinds of "dataflow" architectures are you talking about? The two kinds I am familiar don't have program counters. --
From: Jason Lee Eckhardt on 18 May 2007 15:22 In article <dontlikespan-EF71FE.06082818052007(a)newsgroups.comcast.net>, Alan Charlesworth <dontlikespan(a)nowhere.com> wrote: > >> I'm curious if Alan Charlesworth (who posted in this thread >> as well) had any contact with the i860 designers wrt to the >> FP unit? Or did the designers borrow the ideas independently? >> Does Les Kohn lurk here, or Sai Wai Fu? I've always wanted >> to know the design details-- the IEEE Spectrum article in 1989 >> talks more about staffing, as opposed to the technical details. >> >Nope. I stayed at FPS until it went under and its assets were bought by >Cray in 1992. >Those assets later turned into the Starfire E10K, which >were sold to Sun when SGI bought Cray in 1996 -- but that is a different >story. As for the i860, FPS used in a matrix coprocessor board, which >was an add-in to the FPS-164, a 64-bit follow-on to the AP-120B. Thanks Alan. Did any assets, especially the AP-120B math libraries or dev. tools (APAL, etc.), survive the asset sale to Cray? I'd like to resurrect an AP or 164 by writing an emulator and running some original, unmodified binaries-- as well as just general preservation of anything related to those machines (I've already done this for my other favorite manually-advanced pipeline machine, the i860). There are so many historical questions I'd love to learn about the AP design, about software pipelining[*], and the influence of FPS on other designs and techniques. I agree with Eugene, we need a massive braindump :) jason. [*] In the compiler literature, SWP is generally credited to Bob Rau, due to his 1981 MICRO-14 paper. But clearly the programming examples written by Alan in the "How to Program the AP-120B" manual in 1976 are software pipelined loops (though the phrase doesn't actually appear there), showing the idea to pre-date Rau's article. I suppose Rau could still be legitimately credited with having made the idea more systematic (the "modulo constraint", etc), and therefore easier to incorporate into a compiler.
From: Nick Maclaren on 18 May 2007 16:07 In article <f2kue0$afg$1(a)joe.rice.edu>, jle(a)forest.owlnet.rice.edu (Jason Lee Eckhardt) writes: |> |> [*] In the compiler literature, SWP is generally credited to Bob |> Rau, due to his 1981 MICRO-14 paper. But clearly the programming |> examples written by Alan in the "How to Program the AP-120B" manual |> in 1976 are software pipelined loops (though the phrase doesn't |> actually appear there), showing the idea to pre-date Rau's article. |> I suppose Rau could still be legitimately credited with having made |> the idea more systematic (the "modulo constraint", etc), and |> therefore easier to incorporate into a compiler. That, I regret, is because the modern generation of computer scientists refuse to accept that anything happened before the Unix revolution. Though the ability to publish old results as new inventions gives them an incentive. Bob Rau may well have done important new work, and even been the first to write it up, but software pipelining as a compilation technique dates from no later than the 1960s. It was old hat when I came on the scene - in that context, c. 1971 - and the methodology was well understood, too. In those days, we Just Did It. Regards, Nick Maclaren.
From: Eugene Miya on 18 May 2007 19:27
"In the beginning was the pipe. And the pipe was good. And Ken blessed the pipe..... In article <f2kue0$afg$1(a)joe.rice.edu>, Jason Lee Eckhardt <jle(a)forest.owlnet.rice.edu> wrote: >Alan Charlesworth <dontlikespan(a)nowhere.com> wrote: > Thanks Alan. > > Did any assets, especially the AP-120B math libraries or dev. > tools (APAL, etc.), survive the asset sale to Cray? I'd like > to resurrect an AP or 164 by writing an emulator and running some > original, unmodified binaries-- as well as just general preservation > of anything related to those machines (I've already done this > for my other favorite manually-advanced pipeline machine, the i860). My thinking is that some place with some obsolete obscure 120/164/264 processor attached to something like an MRI or CT will be found. Meanwhile I'm not holding my breath. I have other hard to find machines to locate. > There are so many historical questions I'd love to learn about > the AP design, about software pipelining[*], and the influence > of FPS on other designs and techniques. I agree with Eugene, we > need a massive braindump :) 8^) The problem taking brain dumps is that you have to have a really knowledgeable person asking the questions. Right now it's mostly being done by well meaning novices typically in a 2 hour period. They ask about things like career and education, and family, etc. I took 8 days over many months to interview an officemate (the Univac I was his first computer). And we covered things his family never knew about. I saw Knuth being interviewed: the CHM asked Feigenbaum to do it (possibly a reasonable choice, but likely too short). I would not be a good candidate as a 120B user. Someone more architecturally knowledgeable would be best. > jason. > > [*] In the compiler literature, SWP is generally credited to Bob > Rau, due to his 1981 MICRO-14 paper. You mean: %A B. Ramakrishna Rau %A C. D. Glaeser %T Some Scheduling Techniques and an Easily Schedulable Horizontal Architecture for High Performance Scientific Computing %J Proceedings of the 14th Annual Workshop on Microprogramming (14th MICRO'81) %D October 1981 %P 183-198 %K RBBRS1786, Rhighnam, compiler, optimizer, VLIW, superscalar compiler techniques, polycyclic architecture, The first obscure keywords tell me that Bernutat-Buchmann, Rudolph, & ScholBer and Peter Highnam in their bibliographies as well as me cite them. Levine: you still lurking? Does the compiler community as Jason says believe software pipelining started with Bob? But I also know Gao Guang-Rong worked on software pipelinesas he did his thesis under Jack Dennis (I shared an office with a fellow tea drinker and only later found out that he was into trains), graphics pipelines existed, and I sure know compiler people thought of compilation as a stream (another overloaded term), and all this while the followers of Ken and Dennis were trying to get others into pipes (am I forgetting the DSP community?). Pipelining goes back to 1964 in my biblio alone, but there are subtle terminology distnctions between all the different ways people use it (ask how errors propagate backward, that's a good first order question). > But clearly the programming > examples written by Alan in the "How to Program the AP-120B" manual > in 1976 are software pipelined loops (though the phrase doesn't > actually appear there), showing the idea to pre-date Rau's article. > I suppose Rau could still be legitimately credited with having made > the idea more systematic (the "modulo constraint", etc), and > therefore easier to incorporate into a compiler. What does credit entail? 8^) More money? -- |