Intel's heroic v11.1 compiler, and the impact on AMD

By: slacker (s.delete@this.lack.er), March 7, 2010 3:46 pm
Room: Moderated Discussions
I am trying to buy a new system for a specific, integer-oriented task. After spending well over an hour trying to understand Intel's arcane product segmentation between i7-900, i7-800, i5-700, i5-600, and i3-500 processors, I have given-up and decided to go with a cheaper Phenom (where AMD's attempt at punishing consumers is restricted to their many different sockets)

Since there are no readily-available benchmarks for my application, I decided to look at SPEC CPU2006 and perform judicious subsetting to see what sort of performance loss I would experience with a Deneb vs. Nehalem processor. Looking at gcc and h264ref, it looks like a Core i7-920 would be ~25% faster than a Phenom X4-945 on my application.

There was something very strange about the SPEC results, however. I knew that Intel's v11.1 compiler was producing 2.5x higher results on 456.hmmer compared to their v11.0 compiler, which suggested something naughty was going on. However, what shocked me is how poorly Phenom was doing on the same benchmark. Specifically, a 3.0GHz Phenom scores 5.90 on the benchmark, while a 2.93GHz Core-i3 scores 40.6.

Here is tabulation of relevant data:

i) Phenom II X4 940 - 3GHz, 6MB L3, DDR2-800 CL5 (quad-core) -- http://tinyurl.com/ya46zcg
ii) Opteron 2222 SE - 3GHz, 1MB L2, DDR2-667 CL4 (dual-core) -- http://tinyurl.com/y8hqoov
iii) Xeon X3450 - 2.67GHz (3.2GHz boost), 8MB L3, DDR3-1333 (quad-core) -- http://tinyurl.com/ydcddhu
iv) i3-530 - 2.93GHz, 4MB L3, DDR3-1333 (dual-core) -- http://tinyurl.com/yafbuo8

Results are listed as base/peak. Intel Xeon & i3 results were submitted using icc v11.1, Phenom results using icc v11.0, and Opteron results using PathScale 3.0

              Phenom         Opteron         Xeon           i3           Xeon/Phenom (base)
perlbench 18.0/22.4 13.3/16.2 22.2/27.3 19.9/23.7 1.23
bzip2 14.2/13.8 11.0/11.0 16.7/16.8 15.2/15.2 1.18
gcc 15.5/16.4 10.7/12.2 22.7/25.6 18.0/20.5 1.46
mcf 12.2/12.2 12.0/15.4 39.5/44.0 25.6/31.8 3.24
gobmk 19.0/20.5 15.3/17.5 20.7/23.9 20.7/21.5 1.09
hmmer 5.90/6.00 14.9/20.3 42.4/46.2 40.6/42.4 7.19
sjeng 16.0/17.2 13.8/14.6 24.4/25.4 21.5/22.2 1.53
libquantum 57.1/ - 16.4/16.4 228/232 109/167 4.00
h264ref 26.9/27.8 22.0/23.3 30.2/36.8 32.4/33.8 1.12
omnetpp 12.8/14.4 10.4/10.4 21.1/26.2 13.8/17.6 1.65
astar 12.7/13.3 10.5/ - 17.5/18.1 14.5/14.7 1.38
xalancbmk 17.5/ - 16.0/16.2 35.5/ - 27.7/ - 2.03

Average 16.4/17.3 13.5/14.9 30.4/33.6 24.6/27.6 1.85
Subset1 14.64/ - 13.28/ - 24.63/ - 21.49/ - 1.68
Subset2 16.53/ - 13.26/ - 22.0/ - 19.64/ - 1.33

Subset1: All benchmarks, minus libquantum
Subset2: All benchmarks, minus mcf, hmmer, & libquantum


While everyone is aware that libquantum is a broken benchmark, Intel's new-found 2.5x speed-up on 456.hmmer suggests that they have broken another CPU2006 test. I thought it would be illuminating to re-calculate the INT2006 averages while removing these questionable benchmarks from the results. These calculations are listed as "Subset1" and "Subset2" in the table above.

By removing mcf, hmmer, and libquantum from the test suite, the Nehalem Xeon's performance advantage falls from 1.85x down to 1.33x. This is because these three subtests run ~3 - 7x faster on the Xeon, and heavily distort the typical performance gap between the chips. Why do they run so much faster on Xeon? In the case of libquantum, it appears to have been broken in to an embarrassingly-parallel problem which Intel has been exploiting for quite some time. For some reason, their parallel code doesn't run well on Phenom, or isn't running at all (by means of processor ID checks).

A different story applies to hmmer. While Phenom scores an abysmal 5.90 using icc 11.0, the 3 year old Opteron 2222 scores 14.9 using the PathScale compiler, over 2.5x better. For reference, a Core i7-920 scores 15.8 using icc v11.0 on the same benchmark. In other words, AMD is getting hammered on hmmer because icc is producing awful code for Phenom. I suspect the same thing is happening with mcf.

The real question, however, is whether or not the SPEC scores have a strong correlation with real-world applications. On average, does one see an 85%, 68%, or 33% performance advantage when using a Core i7-920 vs. a Phenom X4-945? I found a set of desktop application benchmarks which help answer this question:

http://www.anandtech.com/bench/default.aspx?p=80&p2=47&c=1

The Nehalem chip is faster on every benchmark. It appears to be 10-25% faster on most benchmarks, and nearly 50% faster on a few (3DSMax Underwater, Pov-Ray, PAR2, WinRAR). There is one benchmark where Nehalem is nearly 100% faster (monte carlo simulation in Excel). Regardless, Nehalem looks like it achieves closer to a 33% performance advantage over Deneb than the 85% speed-up that the SPEC INT scores would suggest.

It is not Intel's job to optimize their compilers for AMD chips. However, Intel goes out of it's way to de-optimize* their compilers for AMD chips, which is miserable behaviour. This is doubly bad for AMD when it's Intel who has submitted every single SPEC CPU2006 score for K10-architecture processors! AMD is not without fault, though, because they started this "SPEC-submissions-on-the-other's-behalf" battle when they submitted SPECpower results** for an 8-socket Intel system.

Does anyone here have integer workloads that run 1.75 - 2.0x faster on equivalently-clocked Nehalem processors compared to Deneb?


* http://www.agner.org/optimize/blog/read.php?i=49
* http://www.swallowtail.org/naughty-intel.shtml

** http://www.formortals.com/amd-submits-suboptimal-specpower-benchmarks-for-intel/
 Next Post in Thread >
TopicPosted ByDate
Intel's heroic v11.1 compiler, and the impact on AMDslacker2010/03/07 03:46 PM
  You may want to see this result tooEduardoS2010/03/07 05:29 PM
    You may want to see this result tooslacker2010/03/07 09:39 PM
  Intel's heroic v11.1 compiler, and the impact on AMDanon2010/03/07 08:34 PM
    Intel's heroic v11.1 compiler, and the impact on AMDslacker2010/03/07 09:29 PM
      Intel's heroic v11.1 compiler, and the impact on AMDStupidDog2010/03/07 11:45 PM
      Intel's heroic v11.1 compiler, and the impact on AMDUngo2010/03/08 12:18 AM
        Intel's heroic v11.1 compiler, and the impact on AMDslacker2010/03/08 07:48 AM
          Intel's heroic v11.1 compiler, and the impact on AMDnone2010/03/08 03:01 PM
            Intel's heroic v11.1 compiler, and the impact on AMDGian-Carlo Pascutto2010/03/09 01:18 AM
  SPEC should disbandDoug Siebert2010/03/07 11:15 PM
    SPEC should disbandGian-Carlo Pascutto2010/03/08 01:35 AM
      SPEC should disbandGabriele Svelto2010/03/08 01:44 AM
        SPEC should disbandGian-Carlo Pascutto2010/03/08 06:41 AM
          SPEC should disbandGabriele Svelto2010/03/08 10:54 PM
            SPEC should disbandGian-Carlo Pascutto2010/03/09 01:16 AM
              SPEC should disbandnone2010/03/09 09:44 AM
                SPEC should disbandsomeone2010/03/10 08:59 AM
                  SPEC should disbandnone2010/03/10 10:03 AM
            SPEC should disbandHenrik S2010/03/09 07:15 AM
      SPEC should disbandsomeone2010/03/08 06:11 AM
        SPEC should disbandGian-Carlo Pascutto2010/03/08 06:29 AM
        liquid coolinganon2010/03/08 08:01 PM
          liquid coolingslacker2010/03/09 12:55 AM
          liquid coolingsomeone2010/03/09 08:45 AM
            liquid coolingRichard Cownie2010/03/09 10:05 AM
              liquid coolingMark Roulo2010/03/09 02:03 PM
                liquid coolingRichard Cownie2010/03/09 03:19 PM
              liquid coolingslacker2010/03/09 04:20 PM
                liquid coolingRichard Cownie2010/03/09 05:43 PM
                  liquid coolingslacker2010/03/09 08:10 PM
                    liquid coolingRichard Cownie2010/03/10 03:57 AM
                      liquid coolingslacker2010/03/10 02:11 PM
                        liquid coolingRichard Cownie2010/03/10 03:43 PM
                          liquid coolingslacker2010/03/11 02:18 PM
              liquid coolingGabriele Svelto2010/03/10 01:39 AM
                liquid coolingRichard Cownie2010/03/10 06:53 AM
                  liquid coolingSalvatore De Dominicis2010/03/10 07:23 AM
                    liquid coolingUngo2010/03/10 01:16 PM
                liquid coolingJesper Frimann2010/03/11 05:12 AM
            liquid coolingmpx2010/03/10 11:10 PM
              liquid coolingRichard Cownie2010/03/11 07:47 AM
                liquid coolingMichael S2010/03/12 04:56 AM
                  liquid coolingRichard Cownie2010/03/12 08:03 AM
                    GPU-only blades and racks? (NT)EduardoS2010/03/12 05:07 PM
                      GPU-only blades and racks? - no wayRichard Cownie2010/03/13 06:02 AM
                        GPU-only blades and racks? - no wayEduardoS2010/03/13 09:22 AM
                          GPU-only blades and racks? - no wayMark Roulo2010/03/13 01:05 PM
                            GPU-only blades and racks? - no wayEduardoS2010/03/13 05:26 PM
                            GPU-only blades and racks? - no wayTvar'2010/03/13 09:21 PM
                            GPU-only blades and racks? - no wayPaul2010/03/14 04:45 AM
                              GPU-only blades and racks? - no wayVincent Diepeveen2010/03/14 11:03 AM
                                GPU-only blades and racks? - no wayAnon2010/03/14 03:42 PM
                                  GPU-only blades and racks? - no wayVincent Diepeveen2010/03/15 11:58 AM
                                    nVidia Tesla vs. non-Tesla BandwidthMark Roulo2010/03/15 12:37 PM
                                    GPU-only blades and racks? - no wayAnon2010/03/15 08:47 PM
                                    GPU-only blades and racks? - no wayPotatoswatter2010/03/16 03:12 PM
                                  GPU-only blades and racks? - no wayUngo2010/03/16 12:04 AM
                    liquid coolingMichael S2010/03/13 10:24 AM
        Computer TimingTerry A. Gray2010/03/10 09:52 AM
    SPEC should disbandDavid Kanter2010/03/09 10:36 AM
  Intel's heroic v11.1 compiler, and the impact on AMDmpx2010/03/08 12:29 AM
    Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/08 12:57 PM
      Intel's heroic v11.1 compiler, and the impact on AMDGabriele Svelto2010/03/08 11:04 PM
        Intel's heroic v11.1 compiler, and the impact on AMDariveira2010/03/09 03:01 AM
        New compression technologyRob Thorpe2010/03/09 11:27 AM
          New compression technologyVincent Diepeveen2010/03/09 02:00 PM
            New compression technologyRob Thorpe2010/03/09 03:31 PM
            New compression technologyS. Rao2010/03/09 03:33 PM
              New compression technologyJesper Frimann2010/03/10 04:00 AM
                New compression technologyGabriele Svelto2010/03/10 04:49 AM
                  New compression technologyJesper Frimann2010/03/10 05:07 AM
                    New compression technologyGabriele Svelto2010/03/10 06:04 AM
                      DNS in DenmarkJesper Frimann2010/03/10 06:28 AM
                New compression technologyVincent Diepeveen2010/03/10 04:59 AM
                  New compression technologySalvatore De Dominicis2010/03/10 06:30 AM
                  New compression technologynemlis2010/03/10 01:17 PM
                  BroadbandRob Thorpe2010/03/10 01:53 PM
                  New compression technology?2010/03/12 12:09 AM
                    New compression technologyPaul2010/03/15 12:36 PM
                      New compression technologyJouni Osmala2010/03/15 09:23 PM
                        New compression technologynemlis2010/03/16 10:54 AM
                  New compression technologyhobel2010/03/28 08:23 AM
                internet speeds in netherlandsVincent Diepeveen2010/03/10 05:10 AM
                  internet speeds in netherlandsJesper Frimann2010/03/10 06:12 AM
                    great tax regime in netherlandsVincent Diepeveen2010/03/10 01:45 PM
                      great tax regime in netherlandsJesper Frimann2010/03/11 01:31 AM
                        great tax regime in netherlandsTerry A. Gray2010/03/12 08:02 AM
                          great tax regime in netherlandsJesper Frimann2010/03/12 10:00 AM
                            great tax regime in netherlandsEduardoS2010/03/12 05:38 PM
                            Constitutions and survelliance statesRob Thorpe2010/03/13 09:10 AM
                              Constitutions and survelliance statesJesper Frimann2010/03/13 04:10 PM
                              Constitutions and survelliance statesslacker2010/03/14 02:31 PM
                                Culture and survelliance statesRob Thorpe2010/03/15 02:07 PM
                                  Culture and survelliance statesJesper Frimann2010/03/15 10:41 PM
                                  Culture and survelliance statesslacker2010/03/16 02:00 AM
                  internet speeds in netherlandsWouter Tinus2010/03/10 08:46 AM
                  internet speeds in netherlandshobel2010/03/28 08:10 AM
                bandwidth available to consumershobold2010/03/10 05:36 AM
              New compression technologyRob Thorpe2010/03/10 04:57 AM
                New compression technologyPaul2010/03/11 03:25 PM
                  New compression technologyEduardoS2010/03/12 05:41 PM
                  New compression technologyRob Thorpe2010/03/13 06:14 AM
                    New compression technologyMichael S2010/03/13 11:45 AM
                      New compression technologyRob Thorpe2010/03/13 12:45 PM
                New compression technology?2010/03/12 12:18 AM
                New compression technologysigh...2010/03/13 11:55 AM
                  New compression technologyRob Thorpe2010/03/14 07:12 AM
                    New compression technologyS. Rao2010/03/14 07:19 PM
                      New compression technologyRob Thorpe2010/03/15 12:40 PM
      Intel's heroic v11.1 compiler, and the impact on AMDmpx2010/03/08 11:29 PM
      7-Zip compess faster than bzip2 ????Michael S2010/03/09 07:43 AM
        7-Zip compess faster than bzip2 ????Gian-Carlo Pascutto2010/03/09 07:48 AM
          indeed, 7-Zip compesses faster than bzip2 Michael S2010/03/09 09:58 AM
            indeed, 7-Zip compesses faster than bzip2 Paul2010/03/09 01:24 PM
              indeed, 7-Zip compesses faster than bzip2 Rob Thorpe2010/03/09 03:34 PM
              indeed, 7-Zip compesses faster than bzip2 Gian-Carlo Pascutto2010/03/09 10:27 PM
                indeed, 7-Zip compesses faster than bzip2 Michael S2010/03/10 02:14 AM
                  indeed, 7-Zip compesses faster than bzip2 Gian-Carlo Pascutto2010/03/10 05:25 AM
                   hardware and compressionVincent Diepeveen2010/03/10 05:26 AM
                     hardware and compressionnone2010/03/10 11:41 AM
                       hardware and compressionVincent Diepeveen2010/03/10 01:33 PM
            indeed, 7-Zip compesses faster than bzip2 Vincent Diepeveen2010/03/09 02:07 PM
              indeed, 7-Zip compesses faster than bzip2 Michael S2010/03/10 06:04 AM
            indeed, 7-Zip compesses faster than bzip2 hobold2010/03/10 05:14 AM
      Intel's heroic v11.1 compiler, and the impact on AMDRob Thorpe2010/03/09 11:18 AM
  Intel's heroic v11.1 compiler, and the impact on AMDmpx2010/03/08 04:56 AM
    Desktop Xeon ECC SupportDavid Hess2010/03/08 08:20 AM
      Desktop Xeon ECC Supportanon2010/03/08 09:56 AM
        Desktop Xeon ECC SupportTomas2010/03/08 10:22 AM
          Desktop Xeon ECC SupportDavid Hess2010/03/08 11:31 AM
            Desktop Xeon ECC SupportMichael S2010/03/08 12:03 PM
              Desktop Xeon ECC SupportTomas2010/03/09 02:19 AM
                Desktop Xeon ECC SupportMichael S2010/03/09 04:53 AM
                  OK, thanks!Tomas2010/03/09 05:19 AM
                    Fujitsu Celsiusslacker2010/03/09 04:10 PM
                Damn it's nice to see....Jesper Frimann2010/03/10 12:33 AM
                  Damn it's nice to see....Richard Cownie2010/03/10 05:39 PM
                    Damn it's nice to see....Jesper Frimann2010/03/11 12:01 AM
                  Damn it's nice to see....JasonB2010/03/11 05:31 PM
                    Damn it's nice to see....Richard Cownie2010/03/11 07:14 PM
                    Damn it's nice to see....Michael S2010/03/12 03:54 AM
                      Damn it's nice to see....JasonB2010/03/12 10:33 AM
                        Damn it's nice to see....EduardoS2010/03/12 06:26 PM
                          Damn it's nice to see....JasonB2010/03/14 07:33 AM
                        Damn it's nice to see....Michael S2010/03/13 10:19 AM
                          Damn it's nice to see....JasonB2010/03/14 08:10 AM
        Desktop Xeon ECC SupportDavid Hess2010/03/08 11:41 AM
          Which motherboard did each of you buy? (NT)slacker2010/03/08 03:02 PM
            Which motherboard did each of you buy? David Hess2010/03/08 07:40 PM
              Which motherboard did each of you buy? slacker2010/03/09 04:12 PM
                Which motherboard did each of you buy? David Hess2010/03/09 06:27 PM
      Desktop Xeon ECC SupportGabriele Svelto2010/03/09 05:03 AM
        Desktop Xeon ECC SupportDavid Hess2010/03/09 08:33 AM
    Intel and ECCEduardoS2010/03/12 08:08 PM
      Intel and ECCMichael S2010/03/13 09:45 AM
  Intel's heroic v11.1 compiler, and the impact on AMDGroo2010/03/08 05:15 AM
    Intel's heroic v11.1 compiler, and the impact on AMDMichael S2010/03/08 06:15 AM
      Intel's heroic v11.1 compiler, and the impact on AMDmpx2010/03/08 07:40 AM
      Intel's heroic v11.1 compiler, and the impact on AMDSimno342010/03/08 08:55 AM
    Intel's heroic v11.1 compiler, and the impact on AMDanon2010/03/08 08:08 PM
    Surprise, surprise(or maybe not)NIKOLAS2010/03/10 09:40 AM
  Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/08 07:03 AM
    Intel's heroic v11.1 compiler, and the impact on AMDslacker2010/03/08 08:38 AM
      Intel's heroic v11.1 compiler, and the impact on AMDPaul2010/03/08 11:39 AM
        Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/08 12:20 PM
          Intel's heroic v11.1 compiler, and the impact on AMDPaul2010/03/08 01:29 PM
            Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/08 02:19 PM
              Intel's heroic v11.1 compiler, and the impact on AMDnemlis2010/03/09 06:39 AM
          Intel's heroic v11.1 compiler, and the impact on AMDSimno342010/03/09 01:05 AM
            Intel's heroic v11.1 compiler, and the impact on AMD?2010/03/09 03:55 AM
              Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/09 04:16 AM
                Intel's heroic v11.1 compiler, and the impact on AMD?2010/03/09 07:34 AM
              Intel's heroic v11.1 compiler, and the impact on AMDnemlis2010/03/09 05:24 AM
                Intel's heroic v11.1 compiler, and the impact on AMD?2010/03/09 07:09 AM
                  Intel's heroic v11.1 compiler, and the impact on AMDnemlis2010/03/09 08:21 AM
            Intel's heroic v11.1 compiler, and the impact on AMDSimno342010/03/09 08:57 AM
              Intel's heroic v11.1 compiler, and the impact on AMDGabriele Svelto2010/03/10 02:29 AM
          Wot, no SSE2 on Apple?James2010/03/09 04:04 AM
            Wot, no SSE2 on Apple??2010/03/09 04:13 AM
              Wot, no SSE2 on Apple?EduardoS2010/03/09 06:25 PM
              Wot, no SSE2 on Apple?James2010/03/10 04:16 AM
                Wot, no SSE2 on Apple??2010/03/10 09:11 AM
                  Wot, no SSE2 on Apple?Vincent Diepeveen2010/03/10 02:01 PM
                    of course xcode produces sse2 codePotatoswatter2010/03/10 02:27 PM
                    Wot, no SSE2 on Apple?nemlis2010/03/11 02:06 AM
                      Wot, no SSE2 on Apple??2010/03/11 02:50 AM
                        Wot, no SSE2 on Apple?hobold2010/03/11 05:38 AM
                          Wot, no SSE2 on Apple??2010/03/11 07:51 AM
                            a theory of communicationhobold2010/03/11 03:38 PM
                              a theory of communication?2010/03/11 11:44 PM
                                a theory of communicationhobold2010/03/12 04:35 AM
                        Wot, no SSE2 on Apple?nemlis2010/03/11 01:21 PM
                          Wot, no SSE2 on Apple??2010/03/12 12:06 AM
                            Wot, no SSE2 on Apple?nemlis2010/03/12 02:01 AM
                            MPEG2nemlis2010/03/12 02:42 AM
                    Wot, no SSE2 on Apple??2010/03/11 04:04 AM
          Intel's heroic v11.1 compiler, and the impact on AMDlongtimelurker2010/03/16 07:27 AM
            GCC auto-vectorizationS. Rao2010/03/16 10:39 AM
              GCC auto-vectorizationPotatoswatter2010/03/16 02:35 PM
              GCC auto-vectorizationAndi Kleen2010/03/16 07:43 PM
              GCC auto-vectorizationVincent Diepeveen2010/03/17 05:22 AM
              GCC auto-vectorization?2010/03/17 06:54 AM
                GCC auto-vectorizationhobold2010/03/17 09:26 AM
                  GCC auto-vectorizationRichard Cownie2010/03/17 10:05 AM
                    GCC auto-vectorizationGian-Carlo Pascutto2010/03/17 10:14 AM
                      GCC auto-vectorizationRichard Cownie2010/03/17 10:43 AM
                        GCC auto-vectorizationGian-Carlo Pascutto2010/03/17 10:05 PM
                          GCC auto-vectorizationRichard Cownie2010/03/18 07:40 AM
                          GCC auto-vectorizationVincent Diepeveen2010/03/22 03:27 AM
                      GCC auto-vectorization?2010/03/17 10:45 AM
                        GCC auto-vectorizationRichard Cownie2010/03/17 11:06 AM
                          GCC auto-vectorizationPaul2010/03/17 03:25 PM
                            GCC auto-vectorizationRichard Cownie2010/03/17 05:48 PM
                    GCC auto-vectorizationSalvatore De Dominicis2010/03/18 01:14 AM
                      GCC auto-vectorizationRichard Cownie2010/03/18 01:29 AM
                        GCC auto-vectorizationGabriele Svelto2010/03/18 01:56 AM
                          GCC auto-vectorizationRichard Cownie2010/03/18 04:14 AM
                            GCC auto-vectorizationGabriele Svelto2010/03/18 04:33 AM
                              Correction: I forgot the sizeof(float) in memcpy(). (NT)Gabriele Svelto2010/03/18 04:36 AM
                              GCC auto-vectorization?2010/03/18 05:02 AM
                                GCC auto-vectorizationGabriele Svelto2010/03/18 05:42 AM
                                  GCC auto-vectorizationRichard Cownie2010/03/18 06:35 AM
                                    GCC auto-vectorizationS. Rao2010/03/18 07:29 AM
                                      GCC auto-vectorizationRichard Cownie2010/03/18 07:52 AM
                                        GCC auto-vectorizationS. Rao2010/03/18 07:58 AM
                                        GCC auto-vectorizationRob Thorpe2010/03/18 10:08 AM
                                          GCC auto-vectorizationRichard Cownie2010/03/18 06:46 PM
                                            hmm, maybe it's not that badRichard Cownie2010/03/18 07:08 PM
                                              hmm, maybe it's not that badRob Thorpe2010/03/19 08:37 AM
                                        GCC auto-vectorizationanon2010/03/18 05:16 PM
                                          GCC auto-vectorizationRichard Cownie2010/03/18 06:27 PM
                                            GCC auto-vectorizationanon2010/03/18 06:54 PM
                                              GCC auto-vectorizationMichael S2010/03/19 06:02 AM
                                                GCC auto-vectorizationanon2010/03/19 02:29 PM
                                            GCC auto-vectorization?2010/03/18 11:03 PM
                                              GCC auto-vectorizationGabriele Svelto2010/03/18 11:47 PM
                                          GCC auto-vectorizationMichael S2010/03/19 05:23 AM
                                    GCC auto-vectorizationGabriele Svelto2010/03/19 12:32 AM
                                      Terminology!?2010/03/19 02:46 AM
                                        Terminology!JasonB2010/03/19 04:29 AM
                                          Terminology!Richard Cownie2010/03/19 04:44 AM
                                          Terminology!?2010/03/19 08:57 AM
                                            Terminology!anon2010/03/19 03:02 PM
                                            Terminology!JasonB2010/03/19 05:13 PM
                                              Terminology!?2010/03/20 01:47 AM
                                                Terminology!JasonB2010/03/20 09:43 PM
                                                  Terminology!Michael S2010/03/21 02:35 AM
                                                    Terminology!JasonB2010/03/21 06:32 AM
                                      GCC auto-vectorizationgallier22010/03/19 04:46 AM
                                        GCC auto-vectorizationAndi Kleen2010/03/19 11:30 AM
                                          GCC auto-vectorizationgallier22010/03/20 07:34 AM
                                            GCC auto-vectorizationAndi Kleen2010/03/20 07:45 PM
                                              GCC auto-vectorizationPaul2010/03/20 09:29 PM
                                                GCC auto-vectorizationgallier22010/03/22 12:12 AM
                                              GCC auto-vectorizationgallier22010/03/22 12:02 AM
                                                GCC auto-vectorizationPaul2010/03/22 12:05 PM
                                                  GCC auto-vectorizationgallier22010/03/22 11:40 PM
                                      GCC auto-vectorizationMichael S2010/03/19 05:06 AM
                                        GCC auto-vectorizationGabriele Svelto2010/03/19 06:00 AM
                                      GCC auto-vectorizationPaul2010/03/19 06:21 PM
                                  GCC auto-vectorization?2010/03/18 11:35 PM
                            GCC auto-vectorizationRob Thorpe2010/03/18 07:45 AM
                      GCC auto-vectorization?2010/03/18 02:07 AM
                        GCC auto-vectorizationSalvatore De Dominicis2010/03/18 02:48 AM
                        GCC auto-vectorizationGabriele Svelto2010/03/18 02:48 AM
                          GCC auto-vectorization?2010/03/18 04:45 AM
                            GCC auto-vectorizationGabriele Svelto2010/03/18 06:08 AM
                            GCC auto-vectorizationRob Thorpe2010/03/18 07:26 AM
                            GCC auto-vectorizationJasonB2010/03/18 09:01 AM
                              GCC auto-vectorizationRichard Cownie2010/03/18 09:23 AM
                                definition of "conforming"Richard Cownie2010/03/18 10:53 AM
                                  definition of "conforming"JasonB2010/03/18 07:41 PM
                                    definition of "conforming"Richard Cownie2010/03/19 04:59 AM
                                      anecdotal performanceRichard Cownie2010/03/19 05:38 AM
                                        anecdotal performanceGabriele Svelto2010/03/19 06:02 AM
                                          anecdotal performanceRichard Cownie2010/03/19 06:37 AM
                                      definition of "conforming"JasonB2010/03/19 08:13 AM
                                        definition of "conforming"Richard Cownie2010/03/19 09:44 AM
                                          definition of "conforming"JasonB2010/03/19 05:06 PM
                                            definition of "conforming"Richard Cownie2010/03/19 06:10 PM
                                              definition of "conforming"JasonB2010/03/20 10:46 PM
                                                definition of "conforming"Richard Cownie2010/03/21 05:39 AM
                                                  definition of "conforming"JasonB2010/03/21 07:50 AM
                                            definition of "conforming"?2010/03/21 10:25 PM
                                GCC auto-vectorizationJasonB2010/03/18 06:34 PM
                                  GCC auto-vectorization?2010/03/19 12:15 AM
                                    GCC auto-vectorizationGabriele Svelto2010/03/19 12:40 AM
                                    GCC auto-vectorizationJasonB2010/03/19 12:44 AM
                                      GCC auto-vectorizationanon2010/03/20 02:41 PM
                                        GCC auto-vectorizationJasonB2010/03/20 08:45 PM
                        GCC auto-vectorizationJasonB2010/03/18 08:35 AM
                    GCC auto-vectorizationhobold2010/03/18 03:50 AM
                      GCC auto-vectorizationAnon2010/03/18 07:55 PM
                GCC auto-vectorizationdmsc2010/03/17 04:55 PM
                  misaligned accessesRichard Cownie2010/03/17 05:54 PM
                    misaligned accesses?2010/03/19 07:24 AM
            Intel's heroic v11.1 compiler, and the impact on AMDVincent Diepeveen2010/03/17 03:40 PM
              Intel's heroic v11.1 compiler, and the impact on AMD?2010/03/17 10:51 PM
              Intel's heroic v11.1 compiler, and the impact on AMDnemlis2010/03/18 07:16 AM
              Intel's heroic v11.1 compiler, and the impact on AMD?2010/03/19 06:52 AM
                Intel's heroic v11.1 compiler, and the impact on AMDnemlis2010/03/19 09:06 AM
  Intel's heroic v11.1 compiler, and the impact on AMD (NT)KiusteR2010/03/08 07:12 AM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell avocado?