VLIW pre-history
|
From: Stephen Fuld on 21 May 2007 19:59 Eugene Miya wrote: >> Eugene Miya wrote: >>> "In the beginning was the pipe. And the pipe was good. >>> And Ken blessed the pipe..... >>> 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). > > In article <dlP3i.16047$p47.5679(a)bgtnsc04-news.ops.worldnet.att.net>, > Stephen Fuld <S.Fuld(a)PleaseRemove.att.net> wrote: >> I thought that pipelining went back at least to Richard Feynman's use of >> pipelining his "computers", the human kind, in the work on the Manhattan >> Project. :-) > > Functionally this is true, but one might also argue that this was an > assembly line technique which goes back to Henry Ford and others. Sure. Mostly, I was trying to be funny and point out an perhaps the first "computer" use of pipelining. I *did* put in a smiley after all. -- - Stephen Fuld (e-mail address disguised to prevent spam)
From: Mark Smotherman on 21 May 2007 20:23 eugene(a)cse.ucsc.edu (Eugene Miya) writes: >I thought this was more a question on wide (horizontal) instruction >systems than pipelining. Yeah, I should have changed the subject line. >So do you think the Stretch was VLIW or a precursor? I had meant to reply to your earlier posting... Stretch's lookahead is a precusor to DAE (decoupled access execute), not VLIW. (E.g., Jim Smith's Astronautics ZS-1 - starting loads as early as possible.) There was no VLIW-like multiple issue in Stretch. Instead, I think the IBM SSEC is a VLIW precursor. (To tie in yet another thread: John Backus was a top SSEC programmer. And Amdahl and Backus proposed Stretch's lookahead.)
From: Jan Vorbrüggen on 22 May 2007 02:51 >>> I thought that pipelining went back at least to Richard Feynman's use >>> of pipelining his "computers", the human kind, in the work on the >>> Manhattan Project. :-) >> Functionally this is true, but one might also argue that this was an >> assembly line technique which goes back to Henry Ford and others. > Sure. Mostly, I was trying to be funny and point out an perhaps the > first "computer" use of pipelining. I *did* put in a smiley after all. However, noteworthy in this respect is his description of exception/fault handling. Once in a while, a computer made an error. When this was detected, instead of stopping the whole computation and restarting it, a special set of data (using a different colour of punched cards) was prepared containing the data before the error occured and surrounding it (in space). This was rushed through the computation until it caught up with the original but faulty data. Because they were doing physical simulation, the spreading of the faulty data was fairly easy to predict. The climax comes when Feynman returns from the hospital in Santa Fe, where his wife has just died from tuberculosis. Nobody pays any attention to him, because his computers are doing an error correction to an error correction... Jan
From: Al Kossow on 22 May 2007 12:44 On May 18, 4:27 pm, eug...(a)cse.ucsc.edu (Eugene Miya) wrote: > "In the beginning was the pipe. And the pipe was good. > And Ken blessed the pipe..... > > In article <f2kue0$af...(a)joe.rice.edu>, > Jason Lee Eckhardt <j...(a)forest.owlnet.rice.edu> wrote: > > >Alan Charlesworth <dontlikes...(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. Early GE CAT scanners used AP-120B's with a special convolution board attached to a DG Eclipse. Apparently my first message never made it out. I used a follow-on product (FPS-100) in the early 80's and have most of the software for it, except for a few things that I couldn't pull from tape. The tools are in FORTRAN. I'm pretty sure I have APASM and APSIM, will have to did out the bits to be sure.
From: Eugene Miya on 22 May 2007 15:51
In article <f2td57$9p4$1(a)hubcap.clemson.edu>, Mark Smotherman <mark(a)clemson.edu> wrote: >eugene(a)cse.ucsc.edu (Eugene Miya) writes: >>I thought this was more a question on wide (horizontal) instruction >>systems than pipelining. > >Yeah, I should have changed the subject line. Well it still is VLIW pre-hist. We're here among the converted. My first boss at Ames was uninterested and unimpressed by software pipelining. >>So do you think the Stretch was VLIW or a precursor? > >I had meant to reply to your earlier posting... Stretch's lookahead >is a precusor to DAE (decoupled access execute), not VLIW. (E.g., >Jim Smith's Astronautics ZS-1 - starting loads as early as possible.) >There was no VLIW-like multiple issue in Stretch. Yeah, I know where Jim's one surviving ZS-1 is. Maybe in time it will migrate to CHM where I will have to explain it to the senior curator. I have harder machines to locate. >Instead, I think the IBM SSEC is a VLIW precursor. Hmmmm. I'm not certain what to think. For instance, one of my inputs is Burton's opinion and the work he did before Cray, Inc. and MTA. Burton isn't a fan of ILP. It's merely limited. >(To tie in yet another thread: John Backus was a top SSEC programmer. >And Amdahl and Backus proposed Stretch's lookahead.) I know a little about the code which motivated the first look ahead. It's still a classified code and likely will forever remain so. And that's why part of it is part of the LFK. -- |