But your CMP scaling is signficantly worse

Article: The Battle in 64 bit Land: Merchant Chips on the Rise
By: IlleglWpns (asarpeshkar.delete@this.hotmail.com), May 3, 2004 1:21 pm
Room: Moderated Discussions
For example your CMP SpecInt results show 65% scaling. The published opteron results show >90% scalability going from 1 to 2 processors.

Is it your belief that the lack of a second memory controller relative to a non-CMP system will hurt scalability that much? Or is it just that at 3.2 ghz the memory wait will be that much higher (% wise).

Also, the current available Pathscale compiler will probably account for a ~100 point increase in SpecFP compared to your predictions, but that's a minor quibble.

< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
New Silicon Insider Article OnlineDavid Kanter2004/05/02 11:25 PM
  TDP questionSingh, S.R.2004/05/03 05:18 AM
    TDP questionWouter Tinus2004/05/03 07:11 AM
      TDP questionanonymous2004/05/03 08:12 AM
        TDP questionWouter Tinus2004/05/03 08:51 AM
          TDP questionanonymus2004/05/04 07:26 AM
    TDP questionPaul DeMone2004/05/03 07:16 AM
      TDP questionAlberto2004/05/03 08:16 AM
        TDP questionPaul DeMone2004/05/03 10:11 AM
          TDP questionAlberto2004/05/03 04:25 PM
      TDP questionMaarten2004/05/04 04:22 AM
        TDP questionPaul DeMone2004/05/04 06:54 AM
  New Silicon Insider Article OnlineBob Koz2004/05/03 07:58 AM
    New Silicon Insider Article OnlinePaul DeMone2004/05/03 08:21 AM
      Does your 'evidence' include backstabing corps?Interested in Illinois2004/05/03 09:00 AM
      Does your 'evidence' include backstabing corps? (NT)Interested in Illinois2004/05/03 09:00 AM
        Does your medication cost alot?waitressInGaza2004/05/03 10:16 AM
          I don't need that poison. We have you. (NT)Interested in Illinois2004/05/03 10:21 PM
            I think this can stop now (NT)David Kanter2004/05/04 12:57 AM
    New Silicon Insider Article OnlineDaniel2004/05/03 01:34 PM
  core size -- x86 vs. IPFanonymous2004/05/03 08:02 AM
    core size -- x86 vs. IPFPaul DeMone2004/05/03 08:24 AM
      core size -- x86 vs. IPFanonymus2004/05/04 07:35 AM
  Corrections aboundInterested in Illinois2004/05/03 08:19 AM
    StalkingDean Kent2004/05/03 12:20 PM
      StalkingSalem2004/05/05 06:28 PM
        More irrational people. :-)Dean Kent2004/05/05 06:48 PM
          More irrational people. :-)Interested in Illinois2004/05/05 09:07 PM
            More irrational people. :-)Heath Provost2004/05/05 09:27 PM
            More irrational people. :-)aap2004/05/05 11:27 PM
              More irrational people. :-)Dean Kent2004/05/06 12:45 AM
            More irrational people. :-)Dean Kent2004/05/06 12:38 AM
              More irrational people. :-)Interested in Illinois2004/05/06 05:32 AM
                More irrational people. :-)Beelzebubba92004/05/06 06:49 AM
                http://www.sacklunch.net/Latin/V/venividivici.html (NT)mas2004/05/07 03:56 PM
              More irrational people. :-)Beelzebubba92004/05/06 06:41 AM
            More funBeelzebubba92004/05/06 06:33 AM
              OT: Even More Funhobold2004/05/07 03:56 AM
                OT: Even More FunPaul DeMone2004/05/07 08:10 AM
                  OT: Even More Funhobold2004/05/08 10:59 AM
  New Silicon Insider Article Onlinemas2004/05/03 09:44 AM
    New Silicon Insider Article OnlinePaul DeMone2004/05/03 10:02 AM
      New Silicon Insider Article Onlinemas2004/05/03 10:15 AM
        New Silicon Insider Article OnlinePaul DeMone2004/05/03 10:24 AM
      New Silicon Insider Article Onlinemas2004/05/03 11:31 AM
        New Silicon Insider Article OnlineDavid Kanter2004/05/03 10:16 PM
      New Silicon Insider Article Onlinemas2004/05/03 01:53 PM
        New Silicon Insider Article OnlinePaul DeMone2004/05/03 02:03 PM
          New Silicon Insider Article OnlineMarc M.2004/05/03 09:04 PM
  Opteron CMP performance estimation?Wouter Tinus2004/05/03 11:21 AM
    Opteron CMP performance estimation?Paul DeMone2004/05/03 11:59 AM
      But your CMP scaling is signficantly worseIlleglWpns2004/05/03 01:21 PM
        But your CMP scaling is signficantly worsePaul DeMone2004/05/03 01:56 PM
          Check which clock rate?IlleglWpns2004/05/03 02:41 PM
            Check which clock rate?Paul DeMone2004/05/03 02:49 PM
              You are right, my badIlleglWpns2004/05/03 02:58 PM
  Opterons with more cacheDresdenboy2004/05/04 12:44 AM
    Opterons with more cachePaul DeMone2004/05/04 07:09 AM
      Opterons with more cacheDoug Siebert2004/05/05 03:59 PM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell purple?