IA64 and emulator performance
|
From: Nick Maclaren on 22 Oct 2006 10:39 In article <m3ods48mie.fsf(a)lhwlinux.garlic.com>, Anne & Lynn Wheeler <lynn(a)garlic.com> writes: |> jgd(a)cix.co.uk (John Dallman) writes: |> |> > Well, I have the impression that the price of Itanium chips is only 3-4 |> > times the price of x86, for vastly smaller production volumes - but |> > can;t find the Itanium prices online at present, so the difference could |> > be bigger. If that's true, I feel reasonably confident that Intel are, |> > overall, loosing money on it. The cutbacks in development will have |> > ameliorated that problem somewhat, but mean that Itanium is gradually |> > losing performance competitiveness. |> > |> > One strongly doubts that HP are paying Intel /more/ than the rate at |> > which anyone can buy Itaniums. |> |> other than the upfront design costs ... the costs are per wafer ... |> modulo some stuff about the number of steps/layers. if the chips share |> same process/line ... and the cost of the line is covered ... then |> the wafer price is relatively the same as long as you have at least a |> minimum sized wafer lot run. |> |> so first level approximation comparison can be number of chips per |> wafer. Assuming a very low failure rate. If the failure rate is high (as it is reported to be for the Itanium), the proportion of failures is likely to be pro rata to the size, so it is the square of that. Whatever. I don't thing that the production costs are a big deal, in the overall scheme of that CPU. Regards, Nick Maclaren.
From: Anne & Lynn Wheeler on 22 Oct 2006 10:49 nmm1(a)cus.cam.ac.uk (Nick Maclaren) writes: > Assuming a very low failure rate. If the failure rate is high (as it > is reported to be for the Itanium), the proportion of failures is likely > to be pro rata to the size, so it is the square of that. > > Whatever. I don't thing that the production costs are a big deal, > in the overall scheme of that CPU. so i should have mentioned *working* chips per wafer ... however, if volume is sufficiently low ... amortizing the upfront design and other ancillary costs per chip ... will dominate the per chip production costs.
From: Andy Freeman on 22 Oct 2006 16:04 > amortizing the upfront design and other > ancillary costs per chip ... will dominate the per chip production > costs. Stopping production won't reduce design or other sunk costs. The only costs that matter in the "should we kill it?" decision are the costs incurred going forward. Intel could be in a "Itanium will never make money because of sunk costs, but it's better to keep making what we've got as long as folks will pay our average production cost plus profit" position. I suspect that Intel's biggest sunk cost is the opportunity cost; AMD got into the game while Intel was distracted by Itanium.
From: Greg Lindahl on 22 Oct 2006 17:31 In article <memo.20061021155512.2712B(a)jgd.compulink.co.uk>, John Dallman <jgd(a)cix.co.uk> wrote: >Absolutely. It's serving as a demonstration of the "supercomputing >fallacy", the belief that almost everyone has problems that are amenable >to the techniques of the category of machines that have been known as >"supercomputers". Perhaps the "supercomputing fallacy" is that people have no clue what chip features make good supercomputers? Certainly the Itanium is missing the most imortant one, "significantly better performance than the rest", and the next important one, "great price performance". The action in supercomputing is mostly in the interconnect these days. The attack of the killer micros is sooo yesterday. -- greg
From: Nick Maclaren on 23 Oct 2006 04:57
In article <453be339$1(a)news.meer.net>, lindahl(a)pbm.com (Greg Lindahl) writes: |> In article <memo.20061021155512.2712B(a)jgd.compulink.co.uk>, |> John Dallman <jgd(a)cix.co.uk> wrote: |> |> >Absolutely. It's serving as a demonstration of the "supercomputing |> >fallacy", the belief that almost everyone has problems that are amenable |> >to the techniques of the category of machines that have been known as |> >"supercomputers". |> |> Perhaps the "supercomputing fallacy" is that people have no clue what |> chip features make good supercomputers? Certainly the Itanium is |> missing the most imortant one, "significantly better performance than |> the rest", and the next important one, "great price performance". It did have "significantly better performance than the rest" - for a while, and for the very small number of Itanium-friendly applications! Unfortunately, as with many other specialised architectures, there weren't enough of them to make a market. And I agree that, even for them, its price/performance was poor. |> The action in supercomputing is mostly in the interconnect these days. |> The attack of the killer micros is sooo yesterday. Indeed. Provided that you include memory access as being part of the interconnect problem, which it is nowadays. Regards, Nick Maclaren. |