How Many Processor Cores Are Enough?
|
From: Tim Bradshaw on 12 Oct 2006 18:40 On 2006-10-12 19:31:04 +0100, kenney(a)cix.compulink.co.uk said: > I get upset when basic physics is confused. C should not be used in > regard to propagation delays in anything but vacuum. For a start in > most cases we are not talking about EMR but electrons or holes. Until > you get optical chips operating in a vacuum C is wishful thinking. My basic physics was not confused, thanks. What I said was what I intended to say: *whatever* you do, and *however* you propagate your signals, they will not propagate faster than c (bar some fairly significant changes to our understanding of how the universe works). So the latency calculated using c is a lower bound. --tim
From: Tim Bradshaw on 12 Oct 2006 18:41 On 2006-10-12 19:31:03 +0100, kenney(a)cix.compulink.co.uk said: > In article <egl6l3$69q$1$8300dec7(a)news.demon.co.uk>, tfb(a)tfeb.org (Tim > Bradshaw) wrote: > >> I'm sure Dennis meant the speed of light in vacuo: I certainly did. > > It is not something that should be used for calculating delays on > chips. That depends on a lot of factors. In all cases it will be > considerably less than C. It's a lower bound on the delay that any technology could have. Which was my point, sigh.
From: Nick Maclaren on 13 Oct 2006 04:24 In article <1160677731.207222.94960(a)m73g2000cwd.googlegroups.com>, "ranjit_mathews(a)yahoo.com" <ranjit_mathews(a)yahoo.com> writes: |> Nick Maclaren wrote: |> > In article <1160664104.209124.194720(a)b28g2000cwb.googlegroups.com>, |> > "ranjit_mathews(a)yahoo.com" <ranjit_mathews(a)yahoo.com> writes: |> > |> |> > |> Very perspicacious! What potential do you forsee for a Software IC |> > |> implemented on a grid of processors and usable from serial codes |> > |> (speedups from parallelism would last only for the duration of each |> > |> call to the Software IC)? |> > |> > What on earth is a Software Integrated Circuit? |> |> It varied from one touter to the next. Think in terms of a design tool |> that can emit logic for either an FPGA or for a grid of processors. |> (Perhaps the tool would come with a runtime system to implement a |> dataflow machine on a grid of processors and emit logic that invokes |> the runtime system) Oh, hell - THAT. Preciously little. Possibly damn all. Yes, there are a few actions that can be speeded up significantly, but there are very few potential uses where those dominate the time. And, in virtually every case, 90% of the performance gain could be provided by exposing more of the existing hardware function to the ISA. Regards, Nick Maclaren.
From: Peter Dickerson on 13 Oct 2006 06:46 "Dennis M. O'Connor" <dmoc(a)primenet.com> wrote in message news:1160688863.532157(a)nnrp1.phx1.gblx.net... > <kenney(a)cix.compulink.co.uk> wrote ... > > In article <k2l104-v7b.ln1(a)osl016lin.hda.hydro.com>, > > terje.mathisen(a)hda.hydro.com (Terje Mathisen) wrote: > > > >> I am sure Dennis really appreciated being told about the speed > >> of light, and how it varies in different materials. > > > > I get upset when basic physics is confused. C should not be used > > in regard to propagation delays in anything but vacuum. > > No one here used it for anything but that. > You just assumed that people were talking > about the propagation speed of signals in other > materials. You assumed wrong. > > Have you forgotten that the "speed of light" in a > material is IIRC equal to C/SQRT(dielectric_constant) ? > > > For a start in most cases we are not talking about EMR > > but electrons or holes. > > Only in the devices. But device delays no longer > dominate in cutting-edge designs. Wire delays do, > and that's EM fields, not electron movement. > > > Until you get optical chips operating in a vacuum C is > > wishful thinking. > > No, C is the speed of light in a vacuum, regardless of > whether optical chips are operating in a vacuum or not. > It is not, as you claim, "wishful thinking". Where I come from the speed of light in vacuo is denoted by (lower case) c not (upper case) C. Hang on, I thing that was a different thread... Peter
From: kenney on 13 Oct 2006 12:56
In article <1160687715.522605(a)nnrp1.phx1.gblx.net>, dmoc(a)primenet.com (Dennis M. O'Connor) wrote: > All the college freshman are discovering USENET. > And I wonder, should I just kill-file the lot of them ? I apologise if I am telling someone how to suck eggs. However there are a lot of factors involved. C may be the upper limit on speed but there are a lot of other factors involved including at Ghz frequencies the inductance of the wire. Ken Young |