Why so little parallelism?
|
From: Erik Magnuson on 9 Nov 2006 00:43 Eugene Miya wrote: > In article <eipla6$rt4$1(a)gemini.csx.cam.ac.uk>, > Nick Maclaren <nmm1(a)cus.cam.ac.uk> wrote: > >>In article <4raspqFq3ffrU1(a)mid.individual.net>, >>=?ISO-8859-1?Q?Jan_Vorbr=FCggen?= <jvorbrueggen(a)not-mediasec.de> writes: >>|> Has anybody claimed that? Even today, there is a place for more than The One >>|> Architecture, and a place for lots of different systems. But for quite some > > Like? (I should have asked) > >>|> time there hasn't been a place for the original Cray approach. >> >>Or rather, the place has shrunk so much that it is no longer large enough >>to contain a viable original Cray approach. The place still exists, even >>if it is solely in the minds of Organisations That We Know Nothing About. > > > My impression is that in some ways, they have somewhat more money than > the average client. They appear progressive in some ways and > conservative in others. > > A lot is in the public domain. They fund lots of research in industry > and universities. Some of this I just saw in an article in US NWP while > waiting in my dentist's office just now. I just heard of Wikis on one of > my CIA visits and there is a big side bar on their use internally now. > Would have been really interesting if your dentist was the late James P. Crutchfield, DDS... Kind of funny in that the last things that CDC that had a significant impact on the computer field were related to small disk drives. The 'IDE/ATA' interface was, IIRC, jointly developed by CDC and Compaq - my Deskpro 386 had a CDC Wren drive. Zoned bit encoding was another. - Erik
From: Jan Vorbrüggen on 9 Nov 2006 04:26 > Not a place for speed? I am not certain what you mean by original? > You mean vectors? (he wasn't the first.) You mean fast scalar? For me, the original Cray approach was your "design systems" but, quite to the contrary, also "design processors" - i.e., he designed the system from the ground up. That's commendable - I'm the holistic, systems engineering type myself - but is no longer viable for a data processing system. Even for the crux of the matter, interconnect, I don't currently see any non-mass market (for a reasonable definition of "mass market" 8-)) succeed. What has happened to Horus, for example? >>For sure. But the goal function has changed: Everybody can jump seven feet >>high nowadays, but doing it carrying 50 kg of baggage is the new game in town. > > They can? I can't jump 7 ft. Well, in a sense you're right. Cray always was about building a balanced system. But in the process, he designed the fastest processing units around. That is not something that will happen again. Thus, it's not in an individual discipline that you have to excel; you have to excel in total performance, i.e., in the decathlon. That would perhaps have been the better analogy. >>>those serious markets won't ever really reveal just what he did for them. >>>Nor ERA nor CDC. >> >>Twaddle. Too much of a conspiracy theory. And do I read "serious" = "national >>security"!? > > Parts of it. [...] So I have to leave it at: > Cray likely worked on things which have not been revealed to the open world. I will willingly believe that the three-letter agencies and other black organizations made good use of whatever Cray et al. had to offer. But for them to drive the product definition and the market, and that in such a way that people wouldn't have noticed? No way. We're talking fairly large amounts of money here, and I don't see a public company hiding those amounts in its balance sheet unnoticed. Jan
From: Jan Vorbrüggen on 9 Nov 2006 04:29 >>|> Has anybody claimed that? Even today, there is a place for more than The One >>|> Architecture, and a place for lots of different systems. But for quite some > > Like? (I should have asked) Power(PC) is still around and going strong, ARM and even MIPS (mostly embedded, to be sure), and it appears there is substantial investment in the S/370 follow-ons from IBM. Not to mention IA-64 8-|. Jan
From: Jan Vorbrüggen on 9 Nov 2006 04:35 >> Well tell us every thing you know about your MI organizations and GCHQ? > I think Nick might be one of the folks who like to make snarky little > remarks about the folks who are trying to protect us from other folks > who would like to kill as many of us as possible. ....with the point being that with all those resources at their disposal, they have failed miserably. And where they didn't fail, the political level failed (cue WMD in the Near East (near for us, guys!)). So why exactly did you, and are you, spending those resources? Yeah, there are likely hundreds of unpublishable success stories. Well, if the tree falls in the forest and nobody heard and saw it - did it really fall? Jan
From: Jan Vorbrüggen on 9 Nov 2006 04:41
> Why would the US be the best place to research the UK's security and > intelligence agencies? While the US has the laudable FOIA - something we are trying to put in place hereabouts - the UK has quite the opposite: the Official Secrets Act, which I have always thought to be unconstitutional in a democracy. But then, as I understand it the UK doesn't _have_ a constitution... Jan |