Post All Your BarsWF x64 / BarsWF x64 CUDA CLient Test Results Here Please!

  • After I did some brainstorming today on what to post about I thought about a topic so people can post print screens of their BarsWF x64 CUDA Client test runs. This is pure for research for me, so I can see how various setups perform with the number of GPGPU's they have, For my test runs with the GTX 480 & 680's I used Forceware 310.70 WHQL, since 310.90 WHQL gave me issues, as where the GTX 295 RE tests were done with 285.62 WHQL.

    So after I made that GTX 660 & GTX 660 Ti Research topic at the EVGA forums for Slash! which is a good friend of mine from 2cpu.com, which you can read here:
    http://www.evga.com/forums/tm.aspx?m=1829726

    I thought about starting a similar for VA as well , but more all round thread where people can post results alike :)
    Thus this topic! :)

    All you need is a System that has 1 CPU or more and has CUDA 1.xx support or greater.
    CUDA 5.xx also works great btw! :topmodel:

    The more CPU's, Cores & Threads you have, the higher your CPU score will be. The CPU score counts per thread and not only per core.

    The only GPGPU's that are supported are from NVIDIA G80 and up :spitze:
    The more GPGPU's & the more CUDA Cores per GPGPU, the higher your GPU Score will be.

    You can download BarsWF x64 CUDA Client from here:
    http://3.14.by/en/md5

    Just remember to disable SLI and then run the application during it's function press print screen then spacebar, then upload your picture here in this topic :)

    Here are my results of my two main setups:

    Firefox SMP 2006
    2x Dual Core AMD OpteronDP 285 E6 HE, 4x 2616Mhz, 2x 2MB L2
    4x 2048MB PC-3200 ECC Reg Dual Rank HP/Micron
    ASUS K8N-DL, NVIDIA Professional 2200 Chipset
    EVGA GeForce GTX 480 PCI-E 1536MB 384Bit GDDR5 Rev.A3 2110
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/55521-1/BarsWF+x64+CUDA+with+FW+310_70+WHQL.jpg]

    That for a system from 2006 and a GTX 480 seems to be very good [sm=001_smile.gif] This was and still is my main LAN setup, it runs every game just fine too, the main hardware is from 2006, but thanks to it's PCI-E 1.1 x16 slot you can use a very powerful GPU to keep this one up to date too, that GTX 480 does a tremendous job with it also [sm=001_smile.gif]


    Blue-Leader SMP 2009
    2x Six-Core AMD OpteronDP 2435 D0, 12x 2613Mhz, 12x 512KB L2 & 2x 6144KB L3
    4x 4096MB PC2-6400 ECC Reg Kingston Server Ram
    Supermicro H8DAE-2 NVIDIA Professional 3600 Chipset + MCP 55 Pro & IO55
    2x EVGA GeForce GTX 680 SuperClocked Signature Edition PCI-E 2048MB 256Bit GDDR5 Rev.A2
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/54977-2/BarsWF+x64+with+Dual+GTX+680+SC+SE_s.jpg]

    This is my main system and it was built in 2009 with two Quad Core Diopter 2380 C2's and it was later upgraded in 2012 with a pair of Six-Core OpteronDP 2435 D0's which gave me an average boost of 61% and that for just 235 Euro's for a sweet 12 core upgrade from the 8 I had before them, this made me stay with a Dual AMD Opteron system, no need for anything else.
    Here a printscreen that shows the results my Dual OpteronDP 2380 C2 had with it's EVGA GeForce GTX 295 Red Edition's:
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/49087-2/BarsWF+x64+on+GTX+295+RE_s.jpg]

    Here the same test but with my Opty 2435's:
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/54585-2/BarsWF+x64+CUDA+Client+on+GTX+295+RE_s.jpg]

    Which shows you a clear CPU performance increase of about 61% :thumbup:
    with Dual Quad 2380's it did 270.70 Mhashes per sec and with Dual Hexa 2435's it does 442.70 Mhashes per second [sm=001_tt1.gif]

    Now everyone has seen my scores, I'd love to see yours too!

    @ GrandAdmiralThrawn
    Could you please also run this with your Mac Pro Dual G5 Quad 2.5Ghz? I'd be very interested to see how that system does BarsWF x64 compared to my Dual Dual Core AMD Opteron 285 E6 HE ! :spitze:

    Einmal editiert, zuletzt von Gold Leader (13. Januar 2013 um 16:47)

  • @ GrandAdmiralThrawn
    Could you please also run this with your Mac Pro Dual G5 Quad 2.5Ghz? I'd be very interested to see how that system does BarsWF x64 compared to my Dual Dual Core AMD Opteron 285 E6 HE ! :spitze:


    I'm unsure how. I can neither see any MacOS X PPC binaries, nor any Linux version that I could probably compile by myself for the PPC microarchitecture. On top of that there is no PowerPC CUDA...

    Edit: It's not exactly "good" code right there it seems, optimizations or not. It appears to be unportable to UNIX (at least not without some developer skills), so it looks like Windows only. And on top of that it seems that the code is very much focused on x86. No way I can build this code on PPC, neither on MacOS X, nor Linux.

    1-6000-banner-88x31-jpg

    Stolzer Besitzer eines 3dfx Voodoo5 6000 AGP Prototypen:

    • 3dfx Voodoo5 6000 AGP HiNT Rev.A-3700

    [//wp.xin.at] - No RISC, no fun!

    QotY: Girls Love, BEST Love; 2018 - Lo and behold, for it is the third Coming; The third great Year of Yuri, citric as it may be! Edit: 2019 wasn't too bad either... Edit: 2020... holy crap, we're on a roll here~♡!

    Quote Bier.jpg@IRC 2020: "Je schlimmer der Fetisch, desto besser!"

    3 Mal editiert, zuletzt von GrandAdmiralThrawn (13. Januar 2013 um 23:39)

  • True, here (this is just a single card though), I don't dare switching off SLI remotely, not sure what the VNC kernel driver'll do in that case, so single card only, this is an EVGA GeForce GTX 580 Classified Ultra:

    [Blockierte Grafik: http://www.xin.at/thrawn/pics/hw/gtx580/barswf.png]

    And here are my previous cards, the GTX 680s on stock speed using an older Barswf version:

    [Blockierte Grafik: http://www.xin.at/thrawn/pics/hw/gtx680/barswf.png]

    Edit: Ah, did it after all. I can now confirm that the UltraVNC hook driver survives an SLI-noSLI-SLI switch without any problems, not even a disconnect or anything. ;)

    Here, both EVGA GTX 580 Classified Ultras, no surprises though:

    [Blockierte Grafik: http://www.xin.at/thrawn/pics/hw/gtx580/barswf-2x.png]

    1-6000-banner-88x31-jpg

    Stolzer Besitzer eines 3dfx Voodoo5 6000 AGP Prototypen:

    • 3dfx Voodoo5 6000 AGP HiNT Rev.A-3700

    [//wp.xin.at] - No RISC, no fun!

    QotY: Girls Love, BEST Love; 2018 - Lo and behold, for it is the third Coming; The third great Year of Yuri, citric as it may be! Edit: 2019 wasn't too bad either... Edit: 2020... holy crap, we're on a roll here~♡!

    Quote Bier.jpg@IRC 2020: "Je schlimmer der Fetisch, desto besser!"

    3 Mal editiert, zuletzt von GrandAdmiralThrawn (14. Januar 2013 um 09:09)

  • Ooooo many thanks for the Dual GTX 580 CU test run there GAT haha very nice, very close to my score hah! :respekt::respekt::respekt::respekt: Your CPU is slightly higher than mine, okay I have 12 cores/12 threads @ 2.6Ghz you have 6 cores/12 threads @ 4.0 Gzh right?
    And my GPGPU's are my savior I guess, still 2K+ for a GF110 is insane though, I must admit that Fermi is much more like G200 than GK104 as it's CUDA Cores go, very interesting this :)

  • Correct, 6 Cores, 12 Threads, Gulftown core, at 4GHz.

    The 580s are running at a shader core speed of 1.8GHz.

    1-6000-banner-88x31-jpg

    Stolzer Besitzer eines 3dfx Voodoo5 6000 AGP Prototypen:

    • 3dfx Voodoo5 6000 AGP HiNT Rev.A-3700

    [//wp.xin.at] - No RISC, no fun!

    QotY: Girls Love, BEST Love; 2018 - Lo and behold, for it is the third Coming; The third great Year of Yuri, citric as it may be! Edit: 2019 wasn't too bad either... Edit: 2020... holy crap, we're on a roll here~♡!

    Quote Bier.jpg@IRC 2020: "Je schlimmer der Fetisch, desto besser!"

  • Many thanks bud :respekt:

    3.8K very nice for a GK110-300-A1 hehe :thumbup:

    Here the score of my NVIDIA GTX 480 Engineering Sample Rev.A2 5009
    @ GPU 527Mhz , 480 Fermi CUDA Cores @ 1053Mhz & 1280MB 320Bit GDDR5 @ 700Mhz GDDR5 / 2800Mhz QDR
    CPU's 2x Dual Core AMD OpteronDP 285 E6 HE @ 4x 2.6Ghz
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/60230-1/BarsWF+x64+CUDA+Client+test+run+on+GTX+480+ES.jpg]

    And here the scores of the same system but with the EVGA GTX 480 Production Sample Rev.A3 2110:
    @ GPU 701Mhz, 480 Fermi CUDA Cores @ 1401Mhz & 1536MB 384Bit GDDR5 @ 924Mhz GDDR5 / 3696Mhz QDR
    CPU's 2x Dual Core AMD OpteronDP 285 E6 HE @ 4x 2.6Ghz
    [Blockierte Grafik: http://www.3dfx.ch/gallery/d/60232-1/BarsWF+x64+CUDA+with+FW+310_70+WHQL_001.jpg]

    3 Mal editiert, zuletzt von Gold Leader (20. November 2013 um 22:55)

  • hah nice scores guys, I gotta admit that a GTX 580 is much better with CUDA per CUDA core than Kepler is, quite impressed to see that FS03's GF110-375-A1 is almost doing the same as CryptonNite's GK104-400-A2, very impressive :thumbup:

  • hah nice scores guys, I gotta admit that a GTX 580 is much better with CUDA per CUDA core than Kepler is, quite impressed to see that FS03's GF110-375-A1 is almost doing the same as CryptonNite's GK104-400-A2, very impressive :thumbup:


    Indeed. Kepler wasn't really optimized for CUDA, afaik.

    "Du bist und bleibst a Mensch und du kannst eben net deine menschlichkeit überwinden."

    Dennis_50300

  • hah nice scores guys, I gotta admit that a GTX 580 is much better with CUDA per CUDA core than Kepler is, quite impressed to see that FS03's GF110-375-A1 is almost doing the same as CryptonNite's GK104-400-A2, very impressive :thumbup:


    Indeed. Kepler wasn't really optimized for CUDA, afaik.


    Yeah, unless you use a GTX Titan, then the numbers will really change, as that these do about 4500 to 4600 Mhashes per sec hehe 8) as GK110-400-A1 goes, with it's 2688 Kepler CUDA Cores , this card is half GeForce half Tesla K20X, it's best of both worlds in a single package.
    a GTX 780 Ti may have 192 more cores, but it won't do more as Double precision rendering goes, the Titan is about 6.7 tiems faster 1.5 teraflops per sec versus 234 Gigaflops per sec xD, Just nuts. Although I do wonder how 2688 Fermi cores would do in comparison, I think they will come very close or even be faster :)

    Einmal editiert, zuletzt von Gold Leader (20. November 2013 um 22:45)

  • Zitat

    Achtung! Die letzte Antwort auf dieses Thema liegt mehr als 1 152 Tage zurück. Das Thema ist womöglich bereits veraltet. Bitte erstellen Sie ggf. ein neues Thema.


    Foxtrot, Uniform, Charlie, Kilo :spitze:

    [Blockierte Grafik: https://abload.de/img/barsopteronnox6u.png]

    AMD Opteron 185 @ 3,12 GHz
    XFX nVidia GeForce GTX295

    "Du bist und bleibst a Mensch und du kannst eben net deine menschlichkeit überwinden."

    Dennis_50300

    Einmal editiert, zuletzt von CryptonNite (17. Januar 2017 um 06:38)