Is it time to stop research in Computer Architecture ?
|
From: Jean on 15 Oct 2009 00:44 In last couple of decades the exponential increase in computer performance was because of the advancements in both computer architecture and fabrication technology. What will be the case for future ? Can I comment that the next major leap in computer performance will not because of breakthroughs in computer architecture but rather from new underlying technology ?
From: Ken Hagan on 15 Oct 2009 07:22 On Thu, 15 Oct 2009 05:44:29 +0100, Jean <alertjean(a)rediffmail.com> wrote: > In last couple of decades the exponential increase in computer > performance was because of the advancements in both computer > architecture and fabrication technology. Call me cynical, but I can't think of any advances in computer architecture that have entered the mainstream in the last twenty years. Surely its *all* been in fabrication. > What will be the case for future ? Can I comment that the next major > leap in computer performance will not because of breakthroughs in > computer architecture but rather from new underlying technology ? I see no reason for the fabrication not to improve, but whether those improvements will boost single-core speed seems, umm, less certain. Right now, everyone is going parallel, despite the fact that no-one has much idea of how to use that parallel power. I'd say the need for research has never been greater.
From: nmm1 on 15 Oct 2009 07:36 In article <op.u1t9vpynss38k4(a)khagan.ttx>, Ken Hagan <K.Hagan(a)thermoteknix.com> wrote: >On Thu, 15 Oct 2009 05:44:29 +0100, Jean <alertjean(a)rediffmail.com> wrote: > >> In last couple of decades the exponential increase in computer >> performance was because of the advancements in both computer >> architecture and fabrication technology. > >Call me cynical, but I can't think of any advances in computer >architecture that have entered the mainstream in the last twenty years. >Surely its *all* been in fabrication. Agreed. Most of the complications of the past 20 years have either been old techniques reintroduced, or dubiously advantages. That isn't to denigrate the engineers, but the objectives they were given. >I see no reason for the fabrication not to improve, but whether those >improvements will boost single-core speed seems, umm, less certain. Right >now, everyone is going parallel, despite the fact that no-one has much >idea of how to use that parallel power. I'd say the need for research has >never been greater. I can. Until now, the problems were (hard) engineering ones; they are now (hard) physics ones. The killer is to obtain the advantages of smaller processes without the major downsides: power consumption, unreliability and fab cost. It's got a LOT tougher. I agree about parallelism, of course. Regards, Nick Maclaren.
From: Mayan Moudgill on 15 Oct 2009 09:28 Ken Hagan wrote: > On Thu, 15 Oct 2009 05:44:29 +0100, Jean <alertjean(a)rediffmail.com> wrote: > >> In last couple of decades the exponential increase in computer >> performance was because of the advancements in both computer >> architecture and fabrication technology. > > > Call me cynical, but I can't think of any advances in computer > architecture that have entered the mainstream in the last twenty years. > Surely its *all* been in fabrication. > Instruction set architecture: multi-media extensions micro-architecture: 2-bit branch prediction
From: ChrisQ on 15 Oct 2009 10:04
Mayan Moudgill wrote: > Ken Hagan wrote: >> On Thu, 15 Oct 2009 05:44:29 +0100, Jean <alertjean(a)rediffmail.com> >> wrote: >> >>> In last couple of decades the exponential increase in computer >>> performance was because of the advancements in both computer >>> architecture and fabrication technology. >> >> >> Call me cynical, but I can't think of any advances in computer >> architecture that have entered the mainstream in the last twenty >> years. Surely its *all* been in fabrication. >> > > Instruction set architecture: multi-media extensions > micro-architecture: 2-bit branch prediction Yes, but utimately boring and really just rearranging the deck chairs. Compared to the 70's and 80's the pace of development is essentially static. Then there were dozens ? of architectures. Now we have what intel and amd deem good for us. ie: what makes them the most money for as long as possible. In this respect, there's a basic conflict between the aims of science / pursuit of excellence and the aims of business. Don't expect any advance from the established order. They know far too much already :-)... Regards, Chris |