Why so little parallelism?
|
From: Del Cecchi on 10 Nov 2006 07:22 "BDH" <bhauth(a)gmail.com> wrote in message news:1163143162.988205.80440(a)e3g2000cwe.googlegroups.com... >> >> Now Eugene, he doesn't need a few billion. He can design his own >> >> and >> >> get help with the gory details, as well as access to a world class >> >> fab >> >> for much less than a few billion. >> >> >> >> I would be happy to facilitate the transaction. A few million as >> >> starters would be sufficient. >> > >> > Its not even that much. If your a student and smart, you can get a >> > small IC fabed at a major vendors fab's for _FREE_. >> >> That would be Mosis most likely and you don't get the latest stuff. >> Or >> very many square mm. >> If you want 65nm or even maybe 45nm SOI you gotta pay. > > You need a real chip even for proof of concept of something only very > helpful on a large scale. How much does a real mask cost nowadays? > It's a set, what with all the device types and about 9 levels of metal. A million bucks or two should be in the ball park. I don't really know. but it is up there.
From: prep on 10 Nov 2006 09:37 eugene(a)cse.ucsc.edu (Eugene Miya) writes: > The customer for the first RAMAC I thought was the Fort. Certainly > close if not the first. CDC had a line of pretty drives resold as > the RP06 and other models. No, the RP06 was a Memorex 677(?) with a massbus `DCL' wart on the side. It was good enought that another company moved heaven and earth to drive Memorex to the wall. CDC spat off their disk biz early by going into a join venture with NCR called MAgnetic Peripherals. CDC disks meets NCR printers. That was about the only thing NCR coud do near right... -- Paul Repacholi 1 Crescent Rd., +61 (08) 9257-1001 Kalamunda. West Australia 6076 comp.os.vms,- The Older, Grumpier Slashdot Raw, Cooked or Well-done, it's all half baked. EPIC, The Architecture of the future, always has been, always will be.
From: prep on 10 Nov 2006 09:44 eugene(a)cse.ucsc.edu (Eugene Miya) writes: > Hey, I speak the truth. Unless I am trolling for data. ;^) Yes, I > have to find a Data Vault for the Museum, a big bank of 80 MB > Fujitsu disks later the 160 MB Eagles. Now I carry 2 GB on my Swiss > Army Knife. The 80 MB was an 8", the Eagle is 486MB, 10" and I think unkillable! The 160 was a 14" with a transparent cover on the HDA. Not real fast AIR. -- Paul Repacholi 1 Crescent Rd., +61 (08) 9257-1001 Kalamunda. West Australia 6076 comp.os.vms,- The Older, Grumpier Slashdot Raw, Cooked or Well-done, it's all half baked. EPIC, The Architecture of the future, always has been, always will be.
From: BDH on 10 Nov 2006 19:05 > > You need a real chip even for proof of concept of something only very > > helpful on a large scale. How much does a real mask cost nowadays? > > > It's a set, what with all the device types and about 9 levels of metal. > A million bucks or two should be in the ball park. I don't really know. > but it is up there. So it's possible, though still not affordable to me. So here's another question - what would the reaction of capital sources and companies be to some unknown tiny company with a working prototype with, say, 10x speedup on matrix multiplies, FFTs, and sorting vs GPUs? If they also had a compiler making ordinary programs work with it? Or what if it only worked with a new, APLish language?
From: Andy Freeman on 10 Nov 2006 20:02
BDH wrote: > > > You need a real chip even for proof of concept of something only very > > > helpful on a large scale. How much does a real mask cost nowadays? > > > > > It's a set, what with all the device types and about 9 levels of metal. > > A million bucks or two should be in the ball park. I don't really know. > > but it is up there. > > So it's possible, though still not affordable to me. > > So here's another question - what would the reaction of capital sources > and companies be to some unknown tiny company with a working prototype > with, say, 10x speedup on matrix multiplies, FFTs, and sorting vs GPUs? > > If they also had a compiler making ordinary programs work with it? > > Or what if it only worked with a new, APLish language? They'd probably wonder why anyone would bother with such a small improvement. (Special purpose machines for those purposes have been built many times.) http://www.google.com/search?sa=N&tab=gw&q=fft%20processor |