View Poll Results: What do you think of my PPD estimator?

Voters
26. You may not vote on this poll
  • It sucks

    3 11.54%
  • It's great

    22 84.62%
  • It could use work (please explain)

    1 3.85%
Page 1 of 3 123 LastLast
Results 1 to 25 of 66

Thread: New version of my PPD Estimator

  1. #1
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267

    New version of my PPD Estimator

    Hey guys,

    Finally a new version of my WCG PPD Estimator! And it's (hopefully) the best version yet!



    CPU support list for the most recent version; if you're looking for something older look at the hidden material below:

    • First Gen Intel Core (Nehalem)
    • Second Gen Intel Core (Sandy Bridge)
    • Third Gen Intel Core (Ivy Bridge)
    • Fourth Gen Intel Core (Haswell)
    • AMD Phenom II
    • First Gen AMD FX (Bulldozer)
    • Second Gen AMD FX (Piledriver)


    Multi-socket systems are implicitly supported; just enter the total number of cores present across all CPUs.

    To use the latest version of the WCG PPD Estimator, download the zip file and extract; then run the ppdEstimator.bat file. This estimator requires Java to be installed on your computer. While this is perhaps sub-optimal, it means that it will be much easier for me to release a Linux or OS X version of it. I'll try to work on those.

    Disclaimer: All estimates given are estimates, actual results will be different. PPD will vary by as much as 50% between two consecutive days due to the nature of the WCG validation process.


    Download Now from TechPowerUp!



    Old version and information; not relevant to most people
    Quote Originally Posted by Ion
    It's been quite a while since I was last logged in here (6 months ), but I return with good news: I've finally gotten around to updating my WCG PPD estimator, now with support for all of the following CPUs:

    Core 2 Duo (Conroe)
    Core 2 Quad (Kentsfield)
    Core 2 Duo (Wolfdale)
    Core 2 Quad (Yorkfield)
    Core i3/Core i5 (Clarkdale)
    Core i5 (Lynnfield)
    Core i7 (quad-core) Now with Sandy Bridge support
    Core i7 (hex-core)
    Athlon 64X2/Athlon X2
    Phenom I X3
    Phenom I X4
    Athlon II X2
    Athlon II X3
    Athlon II X4
    Phenom II X2
    Phenom II X3
    Phenom II X4
    Phenom II X6

    Now with support for Sandy & Ivy Bridge (and -E) as well as 1st/2nd generation AMD APUs--check out the 2013 edition!

    Any (preferably simple) feature requests? I don't promise that I'll be able to get to them right away (and, depending on the complexity, maybe not at all, my programming knowledge is limited) but I'll do my best to add new features you guys want
    Last edited by [Ion]; 11-15-2014 at 12:46 PM.
    WCG Rigs: 184 threads, mostly i7s

  2. #2
    Xtreme Cruncher
    Join Date
    Dec 2008
    Location
    Los Angeles/Hong Kong
    Posts
    3,058
    Pretty hardcore

    What did you code this with?
    Team XS: xs4s.org



  3. #3
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by lkiller123 View Post
    Pretty hardcore

    What did you code this with?
    Thanks!

    Written in a combo of C and C++ in MS Visual Studio 2010. First version (from late late last year) was C-only and had support for a much smaller range of CPUs...this one has some C++ elements, about doubles the CPU support list, and was done in VS2010 (not sure if that really matters though )
    WCG Rigs: 184 threads, mostly i7s

  4. #4
    Xtreme Cruncher
    Join Date
    Dec 2008
    Location
    Los Angeles/Hong Kong
    Posts
    3,058
    Oh by the way, I think it will look more neat if it displays only the PPD at the end.

    That would be perfect
    Team XS: xs4s.org



  5. #5
    Xtreme Cruncher
    Join Date
    Oct 2007
    Posts
    1,638
    I like it and voted so estimates seem high for the Hexa's.

    3700MHz says 7090 ppd and 4200MHz says 8050 ppd.

    free-dc stats is down but I think my machines do ~1,000 ppd less at both those speeds

    XTREMESupercomputer: Phase 2
    Live up to your name - November 1 - 8
    Crunch with us, the XS WCG team

  6. #6
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by lkiller123 View Post
    Oh by the way, I think it will look more neat if it displays only the PPD at the end.

    That would be perfect
    So you mean just the numbers and no other text?
    Quote Originally Posted by trn View Post
    I like it and voted so estimates seem high for the Hexa's.

    3700MHz says 7090 ppd and 4200MHz says 8050 ppd.

    free-dc stats is down but I think my machines do ~1,000 ppd less at both those speeds

    I based my estimates off of a slightly different method. I found someone w/ hexs, got them to go to the device statistics, and give me the points/runtime for at least 7-10 days. From there I then averaged out the numbers and computed what points the system would give if it got exactly 12 days of runtime (as it's a hex) and then worked backwards from there.

    So actual results will probably be lower, I just have found this more accurate than using BOINCStats or Free-DC.

    I'll try and gather more data for hexs, I had limited data for that, so it's probably not so accurate.

    If you wouldn't mind PM'ing me your data from some of your rigs (along w/ CPU speed & OS) I'd be much obliged
    WCG Rigs: 184 threads, mostly i7s

  7. #7
    Xtreme Cruncher
    Join Date
    Apr 2007
    Location
    Western Canada
    Posts
    1,004
    Quote Originally Posted by trn View Post
    I like it and voted so estimates seem high for the Hexa's.

    3700MHz says 7090 ppd and 4200MHz says 8050 ppd.

    free-dc stats is down but I think my machines do ~1,000 ppd less at both those speeds

    Handy, dandy little proggie there

    I'm with you TRN, the numbers for the hexa seem to high. I'm also getting about 1000 boinc pps less than the estimator?

  8. #8
    Xtreme Cruncher
    Join Date
    Dec 2008
    Location
    Los Angeles/Hong Kong
    Posts
    3,058
    Quote Originally Posted by [Ion] View Post
    So you mean just the numbers and no other text?
    Yes, that would be very nice.
    Team XS: xs4s.org



  9. #9
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by Johnmark View Post
    Handy, dandy little proggie there

    I'm with you TRN, the numbers for the hexa seem to high. I'm also getting about 1000 boinc pps less than the estimator?
    See my note about how it's calculated....I'll update it later if the general consensus is that it over-estimates for hexs

    Quote Originally Posted by lkiller123 View Post
    Yes, that would be very nice.
    I can do that later....probably later today, time permitting
    WCG Rigs: 184 threads, mostly i7s

  10. #10
    Xtreme Cruncher
    Join Date
    Oct 2007
    Posts
    1,638
    Quote Originally Posted by [Ion] View Post
    If you wouldn't mind PM'ing me your data from some of your rigs (along w/ CPU speed & OS) I'd be much obliged
    I generally rely on Free-DC stats for host averages and PPD. Right now Free-DC stats is partially out as they mess with their servers,

    All 3 of these i7 Hexa's are running at 4.2GHz and all are running windows 7 or Server 2008 R2 and x64 boinc and haven't been messed with for at least two weeks:
    http://boincstats.com/stats/host_gra...wcg&id=1349795
    http://boincstats.com/stats/host_gra...wcg&id=1185919
    http://boincstats.com/stats/host_gra...wcg&id=1177763

    That will be hard to use, my machines report erratically because I don't use immediately in my config.

    *and I have Linux data too; I saw that was on your future to-do list.
    Last edited by trn; 10-25-2010 at 08:32 PM.
    XTREMESupercomputer: Phase 2
    Live up to your name - November 1 - 8
    Crunch with us, the XS WCG team

  11. #11
    Xtreme Enthusiast
    Join Date
    Jun 2010
    Location
    Bournemouth (UK)
    Posts
    786
    Very nice little tool there [Ion]

    Much easier than scouring the forum for vapors prediction thread ;P
    Nymph - Xeon 5675 @ 4.1 w/ x2 680
    Ikaros - i7 920 w/ 275 - Build In Progress

    Long live LGA1366

  12. #12
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by trn View Post
    I generally rely on Free-DC stats for host averages and PPD. Right now Free-DC stats is partially out as they mess with their servers,

    All 3 of these i7 Hexa's are running at 4.2GHz and all are running windows 7 or Server 2008 R2 and x64 boinc and haven't been messed with for at least two weeks:
    http://boincstats.com/stats/host_gra...wcg&id=1349795
    http://boincstats.com/stats/host_gra...wcg&id=1185919
    http://boincstats.com/stats/host_gra...wcg&id=1177763

    That will be hard to use, my machines report erratically because I don't use immediately in my config.

    *and I have Linux data too; I saw that was on your future to-do list.
    I was actually hoping from your info from www.worldcommunitygrid.org

    I've found that there's little PPD difference between Windows and Linux in terms of PPD, what I meant by Linux was a version compiled for Linux
    Quote Originally Posted by Kryja View Post
    Very nice little tool there [Ion]

    Much easier than scouring the forum for vapors prediction thread ;P
    Thanks
    WCG Rigs: 184 threads, mostly i7s

  13. #13
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Germany
    Posts
    118
    Phenom X6 also seems to be too high.

  14. #14
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by soya_crack View Post
    Phenom X6 also seems to be too high.
    Thanks, I'll try to gather some more info and revise that later

    How much too high?
    WCG Rigs: 184 threads, mostly i7s

  15. #15
    Devil kept pokin'
    Join Date
    Jan 2010
    Location
    South Kakalaky
    Posts
    1,299
    PPD seem to be a bit high. AMD;s in general are 1ppd*mhz*core
    While Intels are 1.1 ppd*mhz*thread
    The non HT Intels Im not positive on.

  16. #16
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by slaveondope View Post
    PPD seem to be a bit high. AMD;s in general are 1ppd*mhz*core
    While Intels are 1.1 ppd*mhz*thread
    The non HT Intels Im not positive on.
    I'm pretty sure that's about what I'm using....the X6 may be a bit higher though...I'll check
    WCG Rigs: 184 threads, mostly i7s

  17. #17
    Xtreme Addict
    Join Date
    Jul 2005
    Posts
    1,646
    Any interest in a quick and dirty website version of this?

  18. #18
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    I could see what I could do in terms of that...it would probably involve learning PHP, so I can't promise immediate results, but it would be cool
    WCG Rigs: 184 threads, mostly i7s

  19. #19
    Xtreme Addict
    Join Date
    Jul 2005
    Posts
    1,646
    I can do the interface quickly in ASP.NET, I just don't know the math for the different CPUs vs client type.

  20. #20
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Oh, well that's easy. PM incoming
    WCG Rigs: 184 threads, mostly i7s

  21. #21
    Xtreme Addict
    Join Date
    Jul 2005
    Posts
    1,646
    Great, I'm trying to avoid work while looking like I'm busy on something.

  22. #22
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    lol

    Thanks!!
    WCG Rigs: 184 threads, mostly i7s

  23. #23
    NooB MOD
    Join Date
    Jan 2006
    Location
    South Africa
    Posts
    5,799
    Quote Originally Posted by [Ion] View Post
    I could see what I could do in terms of that...it would probably involve learning PHP, so I can't promise immediate results, but it would be cool
    I'm a full-time PHP dev, I'll gladly help out. PM me with stats to use (PPD/MHZ/core, or however you do it)

    (It's late evening, I can whip it together after work tomorrow for you.)
    Last edited by [XC] Oj101; 10-26-2010 at 09:43 AM.
    Xtreme SUPERCOMPUTER
    Nov 1 - Nov 8 Join Now!


    Quote Originally Posted by Jowy Atreides View Post
    Intel is about to get athlon'd
    Athlon64 3700+ KACAE 0605APAW @ 3455MHz 314x11 1.92v/Vapochill || Core 2 Duo E8500 Q807 @ 6060MHz 638x9.5 1.95v LN2 @ -120'c || Athlon64 FX-55 CABCE 0516WPMW @ 3916MHz 261x15 1.802v/LN2 @ -40c || DFI LP UT CFX3200-DR || DFI LP UT NF4 SLI-DR || DFI LP UT NF4 Ultra D || Sapphire X1950XT || 2x256MB Kingston HyperX BH-5 @ 290MHz 2-2-2-5 3.94v || 2x256MB G.Skill TCCD @ 350MHz 3-4-4-8 3.1v || 2x256MB Kingston HyperX BH-5 @ 294MHz 2-2-2-5 3.94v

  24. #24
    Xtreme Member
    Join Date
    Mar 2010
    Location
    Raleigh, NC
    Posts
    267
    Quote Originally Posted by [XC] Oj101 View Post
    I'm a full-time PHP dev, I'll gladly help out. PM me with stats to use (PPD/MHZ/core, or however you do it)

    (It's late evening, I can whip it together after work tomorrow for you.)
    YGPM
    WCG Rigs: 184 threads, mostly i7s

  25. #25
    NooB MOD
    Join Date
    Jan 2006
    Location
    South Africa
    Posts
    5,799
    Well I guess it can go here then...

    Xtreme SUPERCOMPUTER
    Nov 1 - Nov 8 Join Now!


    Quote Originally Posted by Jowy Atreides View Post
    Intel is about to get athlon'd
    Athlon64 3700+ KACAE 0605APAW @ 3455MHz 314x11 1.92v/Vapochill || Core 2 Duo E8500 Q807 @ 6060MHz 638x9.5 1.95v LN2 @ -120'c || Athlon64 FX-55 CABCE 0516WPMW @ 3916MHz 261x15 1.802v/LN2 @ -40c || DFI LP UT CFX3200-DR || DFI LP UT NF4 SLI-DR || DFI LP UT NF4 Ultra D || Sapphire X1950XT || 2x256MB Kingston HyperX BH-5 @ 290MHz 2-2-2-5 3.94v || 2x256MB G.Skill TCCD @ 350MHz 3-4-4-8 3.1v || 2x256MB Kingston HyperX BH-5 @ 294MHz 2-2-2-5 3.94v

Page 1 of 3 123 LastLast

Bookmarks

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •