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: Anne & Lynn Wheeler on 7 Sep 2009 13:38 Mayan Moudgill <mayan(a)bestweb.net> writes: > 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). don't forget los gatos vlsi lab ... did chips for various disk division products (like jib prime for 3880 disk controller) . also put in lots of work on blue iliad (1st 32bit 801 ... never completed). then there was stuff going outside the US. one might point out that the number of circuits going into many current processors drawfs the aggregate number of circuits in all of those chips from the past. los gatos also did the LSM (los gatos state machine ... renamed the logic simulation machine). it was used for logic verification of some number of chip designs ... not just Los Gatos chips. One of the things that differentiated LSM (from most of the other, similar hardware logic simulators of the period) was that it had a "clock" which provided for handling non-syncrhonous designs and/or digital chips that included analog circuits (possibly what one might find in disk r/w heads). part of the 801/iliad risc effort in the late 70s & early 80s was converge the large number of different processor chips to 801 (significant numbers were "embedded) ... it wasn't just the chip design .... but each chip tended to have customized software development and programming environment. -- 40+yrs virtualization experience (since Jan68), online at home since Mar1970
From: Kai Harrekilde-Petersen on 7 Sep 2009 13:44 Mayan Moudgill <mayan(a)bestweb.net> writes: > 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. Actually, you have just mentioned all of the things we *aren't* doing in our hearing aid DSP (the GN Coyote, which was co-developed with Audiologic, now part of Cirrus Logic). One simple reason: power consumption (hint: <1mW power envelope). Whether we are doing architecture work, or just micro-architecture, these days is a matter of interpretation as I see it. But we definitely change the thing from generation to generation. But I will agree that the days of the 80'ies where every semiconductor with worth it's salt had to have their own processor series are gone and buried (and good riddance too). I wouldn't be surprised if we see a minor revival in processor design centered around low power consumption and "acceptable" performance (who needs a multi-core-multi-GHz processor to show webpages, edit text and do a bit of email/blogging/twitter?). Kai -- Kai Harrekilde-Petersen <khp(at)harrekilde(dot)dk>
From: nmm1 on 7 Sep 2009 14:00 In article <uzl96y7i0.fsf(a)harrekilde.dk>, Kai Harrekilde-Petersen <khp(a)harrekilde.dk> wrote: > >I wouldn't be surprised if we see a minor revival in processor design >centered around low power consumption and "acceptable" performance >(who needs a multi-core-multi-GHz processor to show webpages, edit >text and do a bit of email/blogging/twitter?). Because you are running Macrosloth Bloatware - and even Linux seems to be competing on that front :-( I would like to believe that you are right, but have been trying to persuade people that there is a Real Market for such things for some years. And look at what happened to SiCortex. It may happen, but is unlikely in the near future. Regards, Nick Maclaren.
From: nmm1 on 7 Sep 2009 15:38 In article <u1vmilgi9.fsf(a)harrekilde.dk>, Kai Harrekilde-Petersen <khp(a)harrekilde.dk> wrote: >Robert Myers <rbmyersusa(a)gmail.com> writes: >> On Sep 7, 1:44�pm, Kai Harrekilde-Petersen <k...(a)harrekilde.dk> wrote: >> >>> I wouldn't be surprised if we see a minor revival in processor design >>> centered around low power consumption and "acceptable" performance >> >> What a *very* odd comment, as the shift of design emphasis and >> acccompanying frenetic activity has been apparent for years. > >And they use the lowered power envelope to stuff even more idle cores >onto the die. Not exactly forward progress in my book (although Nick >might approve of getting more cores). Not the way that they are currently being done :-( I agree with you in this respect. My rule of thumb is that existing software can make use of 2-4 cores on a desktop system, but 8 is a waste. >It's all a matter of perspective and perception. Not entirely. My radical proposal is clearly intended to kick-start innovation on the software front, because the current direction is so clearly leading into a dead-end. Only a complete loon would expect current software to do anything useful on large numbers of processors, let alone with a new architecture! Regards, Nick Maclaren.
From: Jim Haynes on 7 Sep 2009 16:40
I think one thing that has happened is that the distinction between architecture and implementation has blurred considerably. The term "computer architecture" came into use at the time the IBM System/360 was introduced, to stress the fact that there was one instruction set across six different implementations of widely varying performance. Today we don't see any demand for that kind of performance variation in the single processor. There's no market for brand new PCs with 486 CPUs. Software grows to soak up every increase in performance. To get really high power you use hundreds or thousands of off-the-shelf CPUs because they are mass-produced at large quantity prices. (And you couldn't make a higher power CPU by doing anything other than what designers of mass-market CPUs are already doing.) And what they are doing is very much limited by the manufacturing processes, so you can't propose a different architecture without considering how it is to be manufactured. And there is the momentum of what is already being done. You can't plan to make an architecture and build one or a few machines to test the water - anything you do has to sell thousands and millions to be worth the cost of getting ready to produce it. |