What happened to computer architecture (and comp.arch?)
|
Prev: VMWare tools killed my Mac OS X ?
Next: Software vs hardware floating-point [was Re: What happened ...]
From: Terje Mathisen on 7 Sep 2009 03:25 Mayan Moudgill wrote: > > I've been reading comp.arch off and on for more than 20 years now. In > the past few years the SNR has deteriorated considerably, and I was > wondering why. Maybe people who used to post at comp.arch are on other > formums? Maybe its that I've gotten a little harder to impress? Then I > thought about the quality of most papers at ISCA and Micro, the fact > that both EDF and MPF have gone away, and I think the rot is not > confined to just comp.arch. > > So, whats going on? I'm sure part of it is that the latest generation of > architects is talking at other sites. > > However, equally important is that there are far fewer of them. The > number of companies designing processors has gone down and there are > fewer startups doing processors. So, less architects. Maybe. It might also be that the number of _good_ architects are more or less constant, and the required minimum size of a team has gone up. I.e. less possible teams? [snip] > A lot of the architecture that is being done is application-specific. Yes, like all the SIMD and/or manycore stuff? Even if Larrabee has been targeted initially as a graphics pipeline, it is obvious that the same pipeline will in the future be used for amany other forms of processing that can use a _lot_ of fp performance. > Consequently, its probably more apt to be discussed in > comp.<application> than comp.arch. A lot of the trade-offs will make > sense only in that context. > > Basically, I think the field has gotten more complicated and less > accessible to the casual reader (or even the gifted well read amateur). > The knowledge required of a computer architect have increased to the > point that its probably impossible to acquire even a *basic* grounding > in computer architecture outside of actually working in the field > developing a processor or _possibly_ studying with one of a few PhD > programs. The field has gotten to the point where it _may_ require > architects to specialize in different application areas; a lot of the > skills transfer, but it still requires retraining to move from, say, > general-purpose processors to GPU design. > > I look around and see a handful of guys posting who've actually been > doing computer architecture. But its a shrinking pool.... > > Ah, well - I guess I can always go hang out at alt.folklore.computers. No! Terje -- - <Terje.Mathisen at tmsw.no> "almost all programming can be viewed as an exercise in caching"
From: zzbunker on 7 Sep 2009 03:25 On Sep 6, 9:44 pm, Mayan Moudgill <ma...(a)bestweb.net> wrote: > I've been reading comp.arch off and on for more than 20 years now. In > the past few years the SNR has deteriorated considerably, and I was > wondering why. Maybe people who used to post at comp.arch are on other > formums? Maybe its that I've gotten a little harder to impress? Then I > thought about the quality of most papers at ISCA and Micro, the fact > that both EDF and MPF have gone away, and I think the rot is not > confined to just comp.arch. > > So, whats going on? I'm sure part of it is that the latest generation of > architects is talking at other sites. > > However, equally important is that there are far fewer of them. The > number of companies designing processors has gone down and there are > fewer startups doing processors. So, less architects. > > Within those processors there is less architecture (or micro > architecture) being done; instead, the imperative that clock cycle has > to be driven down leaves less levels of logic per cycle, which in turn > means that the "architecture" has to be simpler. So, less to talk about. > > There is less low-hanging fruit around; most of the simpler and > obviously beneficial ideas are known, and most other ideas are more > complex and harder to explain/utilize. > > A larger number of decisions are being driven by the details of the > process, libraries and circuit families. This stuff is less accessible > to a non-practitioner, and probably propietary to boot. > > A lot of the architecture that is being done is application-specific. > Consequently, its probably more apt to be discussed in > comp.<application> than comp.arch. A lot of the trade-offs will make > sense only in that context. > > Basically, I think the field has gotten more complicated and less > accessible to the casual reader (or even the gifted well read amateur). > The knowledge required of a computer architect have increased to the > point that its probably impossible to acquire even a *basic* grounding > in computer architecture outside of actually working in the field > developing a processor or _possibly_ studying with one of a few PhD > programs. The field has gotten to the point where it _may_ require > architects to specialize in different application areas; a lot of the > skills transfer, but it still requires retraining to move from, say, > general-purpose processors to GPU design. > > I look around and see a handful of guys posting who've actually been > doing computer architecture. But its a shrinking pool.... Well, the computer architects essentially limited computer architectute to Von Neumann architectute, and the 21st Centiiry People are mostly working on Atomic Clock Wristwatches, Post AT&T Fiber Optics, GPS, HDTV, Electric, On-Line Publishing, the only you can really tell them anymore is that X windows will only live on in GM. > > Ah, well - I guess I can always go hang out at alt.folklore.computers.
From: nmm1 on 7 Sep 2009 04:29 In article <a_CdnXqAUf36KDnXnZ2dnUVZ8oCdnZ2d(a)lyse.net>, Terje Mathisen <Terje.Mathisen(a)tmsw.no> wrote: >Mayan Moudgill wrote: >> >> So, whats going on? I'm sure part of it is that the latest generation of >> architects is talking at other sites. >> >> However, equally important is that there are far fewer of them. The >> number of companies designing processors has gone down and there are >> fewer startups doing processors. So, less architects. > >Maybe. It might also be that the number of _good_ architects are more or >less constant, and the required minimum size of a team has gone up. Not really. I don't think that the number of top architects needed on a team is likely to be very different. The reasons are almost certainly that architecture is now dominated by 'safe' designs (i.e. those with proven markets), and a few very large companies. The active architects cannot post about what they are working on, and there isn't a lot of peripheral activity. Look at what's happened to memory (and, no, RAMBUS was NOT exciting), the range of architectural designs and SiCortex. >> A lot of the architecture that is being done is application-specific. > >Yes, like all the SIMD and/or manycore stuff? > >Even if Larrabee has been targeted initially as a graphics pipeline, it >is obvious that the same pipeline will in the future be used for amany >other forms of processing that can use a _lot_ of fp performance. There isn't all that much of it, once you factor in the low bandwidth of Larrabee. The codec market could collapse overnight if the current chaos is replaced by a de facto standard. QCD isn't exactly a major market, and isn't likely to become so in the forseeable future. Cryptanalysis is a market, at least while the USA and its lackeys follow the current political path, but not as much of one as is often claimed. And I can't think of much else. While I agree that experience with the ICL DAP, BBN Butterfly etc. indicate that applications can be developed for more scalable architectures, neither Larrabee nor Niagara/etc. are either scalable or large enough to encourage such adaptation. As you know, I am radical among radicals, but what I should like to see is a 1,024 core chip, with an enhanced 2-D grid memory topology, back-to-back with its NON-shared memory, NO hardware floating-point, first-class support for emulation and virtual architectures, and so on. Available in desktops at a desktop price. It's all proven technology, and several companies could do it. And, yes, I know that it would need a slow clock speed to deliver that - but not as slow as many people think. If that didn't kick-start some innovation, then the IT industry is beyond redemption. It would also beat Larrabee, Niagara and even Tesla into a pulp for the very processor-intensive applications. Regards, Nick Maclaren.
From: Mayan Moudgill on 7 Sep 2009 12:13 Terje Mathisen wrote: > Mayan Moudgill wrote: > >> >> I've been reading comp.arch off and on for more than 20 years now. In >> the past few years the SNR has deteriorated considerably, and I was >> wondering why. Maybe people who used to post at comp.arch are on other >> formums? Maybe its that I've gotten a little harder to impress? Then I >> thought about the quality of most papers at ISCA and Micro, the fact >> that both EDF and MPF have gone away, and I think the rot is not >> confined to just comp.arch. >> >> So, whats going on? I'm sure part of it is that the latest generation of >> architects is talking at other sites. >> >> However, equally important is that there are far fewer of them. The >> number of companies designing processors has gone down and there are >> fewer startups doing processors. So, less architects. > > > Maybe. It might also be that the number of _good_ architects are more or > less constant, and the required minimum size of a team has gone up. > > I.e. less possible teams? Consider this: at one time, IBM had at least 7 teams developing different processors: Rochester, Endicott, Poughkeepsie/Fishkill, Burlington, Raliegh, Austin & Yorktown Heights (R&D). Now, I think the number is down to 3 processors over maybe 4 sites? DEC is gone, of course. I don't know what the status is at HP. AMD, Intel & Via - well, AMD/ATI has probably got 2-3 development efforts on-going, Intel hmmm, I would guess 2 or 3 (Israel and a couple in USA?), Via 1? Sun/Fujitsu? Who knows. Is Freescale (nee Motorola) developing new Power stuff? ARM and its liscencees - Qualcomm, Samsung, Broadcom et.al. probably about 5 or 6 processors. Other embedded (MIPS, SH-n): ? GPU: nVidia, maybe some startups? DSP: TI's probably not doing much in the way of new architecture, more like evolutionary work on the C55x and C6x lines. ADI? dunno. Motorola's still pushing Starcore, though I don't know how much extra they've done. Lucent/Agere are out of the game. A few startups (including Sandbridge) Odds & sods: Tensilica I'm sure there are other processor or processor-like groups I've forgotten out there.
From: nmm1 on 7 Sep 2009 12:43
In article <PfOdnX-Ctu-irDjXnZ2dnUVZ_oKdnZ2d(a)bestweb.net>, Mayan Moudgill <mayan(a)bestweb.net> wrote: > >I'm sure there are other processor or processor-like groups I've >forgotten out there. Hitachi, for a start. There are almost certainly other active groups in Japan, possibly Korea, probably China, and perhaps India. Regards, Nick Maclaren. |