1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Processor GFlops Compilation

Discussion in 'General Hardware' started by newtekie1, May 19, 2009.

  1. hoss331

    hoss331 New Member

    Joined:
    Nov 1, 2008
    Messages:
    119 (0.06/day)
    Thanks Received:
    24
    Location:
    Florida
    If you want you can delete my other runs and just use this one.

    hoss331 | Q9650 @ 4.32ghz | 60.63655 | 14.03623 | 64-bit | max memory 3250mb

    [​IMG]
  2. lemonadesoda

    lemonadesoda

    Joined:
    Aug 30, 2006
    Messages:
    6,242 (2.17/day)
    Thanks Received:
    963
    We can all play the increase-memory-for-a-better-score game :shadedshu :shadedshu

    lemonadesoda | 2x E5420 @ 2.50 Ghz | 62.21 | 24.88 | 32-bit | max memory 2047 mb

    [​IMG]

    Let's keep the benchmark table simple: stick with 1024 MB RAM. While most users could do 2048MB runs, and get better scores, it means recreating the table. Mixing and matching memory in the table is clearly not a good idea.

    Hoss, try OC your 1024MB run again, fiddling with memory timings... if you want to get to the top of the scoreboard... keep trying. ;)

    Then I'll OC my sytem rather than running it at stock. ROFL :roll:
    Last edited: May 28, 2009
    Naekuh says thanks.
  3. lemonadesoda

    lemonadesoda

    Joined:
    Aug 30, 2006
    Messages:
    6,242 (2.17/day)
    Thanks Received:
    963
    Nope. Because a Quad is TWO Duals on one die. Do you suggesting cutting those scores in two also?

    What you could do however is to have a THIRD calculation that is the Score per Ghz per virtual core. That means divinding by 1 for Pentium 3. By 2 for P4 on HT, by 2 for Core 2, by 4 for Core 2 Quad, by 8 for Nehalem on HTT, by 8 for Dual Core Quad Xeons.

    Getting a bit confusing though. Anyway, that's not the point of benchmarking. You dont want to keep dividing like that. What you want to do is assess the TOTAL CAPABILITY of the rig. The only reason we have the GHz divider is to see who much the results are from OC, and whether we want to OC our existing setups further, or move over to a newer faster architecture.
  4. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Table "Update", and I put that in quotes for a reason. I added nothing new, but I did go through and delete any score that was done with the memory setting different than 1024MB. No score will be accepted unless all the rules in the original thread are followed. The only exception will be for computer that physically don't have enough RAM to allow 1024MB runs.

    Also, I've fixed the column problems with the 2 scores, I was in a hurry last night trying to finish the updating and missed adding in the manufacturer.
    lemonadesoda says thanks.
    Crunching for Team TPU
  5. lemonadesoda

    lemonadesoda

    Joined:
    Aug 30, 2006
    Messages:
    6,242 (2.17/day)
    Thanks Received:
    963
    newtekie, please update your OP, to make it clear what the rules are. The OP doesnt state, MUST RUN AT 1024MB... etc. etc.
  6. Naekuh

    Naekuh New Member

    Joined:
    Apr 2, 2009
    Messages:
    503 (0.26/day)
    Thanks Received:
    108
    Location:
    Miko Fox? Fire Miko? Its FireFox-Tan!
    u think 4.4 is the best i can do on a 975 + classified? :laugh:

    Im waiting for someone to beat me, so i can put pressure again.
    That brings fun in competition, not when one person just slaughters the board via landslide.
  7. lemonadesoda

    lemonadesoda

    Joined:
    Aug 30, 2006
    Messages:
    6,242 (2.17/day)
    Thanks Received:
    963
    I just knew you were holding something back. What a teaser. ;)

    Just to whet your appetite, at a moments notice I can BSEL mod my Xeons to get another 10%. So start practising at pulling in 70 Gflops... or you will be in trouble. :p But the drama is when I pull out from 5th place to regain the crown. LOL
  8. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    The original post says to run it at the standard stress level, this is 1024MB.
    Crunching for Team TPU
  9. Naekuh

    Naekuh New Member

    Joined:
    Apr 2, 2009
    Messages:
    503 (0.26/day)
    Thanks Received:
    108
    Location:
    Miko Fox? Fire Miko? Its FireFox-Tan!
    LOL well.. im only pushing 1.35vcore with a dedicated cpu loop.

    My load temps for that 4.4 didnt push into 70's.

    So i figure i have about another 300-400mhz (linpack stable) as backup incase someone decides to bring a nuke into this thread.
  10. hoss331

    hoss331 New Member

    Joined:
    Nov 1, 2008
    Messages:
    119 (0.06/day)
    Thanks Received:
    24
    Location:
    Florida
    Are we taking this a little to seriously? I didnt recall seeing a limit on how much ram we could use, and its not like I was trying to pass if off as a 1gb run. Yea im a cheater because I used a harder test. :rolleyes: I dont see what the big deal is using max as long as it was notated.
  11. lemonadesoda

    lemonadesoda

    Joined:
    Aug 30, 2006
    Messages:
    6,242 (2.17/day)
    Thanks Received:
    963
    Read the rest of my post and dont be so uptight. Perhaps also read the next post I made to newtekie to point out that perhaps the memory limit wasnt clear to everyone. Then read your post above, sit down, have a drink, and start OC'ing.
  12. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    It has been in the first post since I created the thread:

    Standard = 1024MB =/= Custom
    Crunching for Team TPU
  13. DOM

    DOM

    Joined:
    May 30, 2006
    Messages:
    7,551 (2.54/day)
    Thanks Received:
    828
    Location:
    TX, USA
    Add Me

    DOM | Intel Core 2 Duo E8500 @ 5.1GHz | 34.7517 | 6.8141 | 64-bit |

    [​IMG]
  14. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    41,942 (11.76/day)
    Thanks Received:
    9,364
    DOM says thanks.
  15. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Table Updated
    Crunching for Team TPU
  16. mudkip

    mudkip

    Joined:
    Dec 14, 2008
    Messages:
    1,226 (0.60/day)
    Thanks Received:
    148
    Location:
    The Netherlands
    maby it would be nice if you'd order the table on Gflops.. now they're all just random
  17. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Its too much work to sort the table manually.
    Crunching for Team TPU
  18. dcf-joe

    dcf-joe

    Joined:
    Feb 11, 2008
    Messages:
    372 (0.16/day)
    Thanks Received:
    20
    Location:
    Nebraska, USA
    Add Me

    dcf-joe | Intel | Core i7 920 @ 3.8 GHz | 48.04096 | 12.64235789 | 64-Bit | Program set to run on 8 threads

    [​IMG]
  19. DanTheBanjoman Señor Moderator

    Joined:
    May 20, 2004
    Messages:
    10,553 (2.85/day)
    Thanks Received:
    1,383
  20. DanTheBanjoman Señor Moderator

    Joined:
    May 20, 2004
    Messages:
    10,553 (2.85/day)
    Thanks Received:
    1,383
    And then? The DSEB/DG doesn't have any options to change vcore in the first place. So there is no gain.

    As for Z7S's being troublesome, this is only my second. Some people at XS had more :D

    Strange though, changing vcore worked fine before with this BIOS version. Not sure what changed, the thread on XS mentions it as a common issue though. Either way, my new board is ordered. The Z7S will be degraded to secondary machine, will sell the S5370 and X7DAL-E.

    Also, BSEL mods are needed either way since there is no control over the FSB/memory divider.
  21. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Table updated

    That doesn't work properly, I tried it. For some reason it sorts them, but not in the correct order. It gets them close to sorted into the right oder, but it puts some of the scores in the 40s higher than the 50s and 60s. Sorting by clicking the table headers works perfectly though.
    Crunching for Team TPU
  22. DanTheBanjoman Señor Moderator

    Joined:
    May 20, 2004
    Messages:
    10,553 (2.85/day)
    Thanks Received:
    1,383
    That some reason was a space missing in those entries, I'm not sure why you have all those spaces in there in the first place anyway. I added the space and it's sorted fine now.
    newtekie1 says thanks.
  23. entropy13

    entropy13

    Joined:
    Mar 2, 2009
    Messages:
    4,892 (2.50/day)
    Thanks Received:
    1,182
    entropy13 | Intel| Core i7 920 @ stock w/ HT and Turbo on | 28.0738 | 10.5541 | 64-bit
    [​IMG]
  24. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,664 (6.23/day)
    Thanks Received:
    5,838
    Table Updated

    Thanks, I figured it was something stupid on my part like that, but when I tried it it didn't work correctly right away and I said screw it and just added the ability to sort by column headers and haven't had the time or patience to mess around with figuring it out(though it was on my list of things to do).:toast:

    I put the spaces there simply to make things easier to read if I have to edit the table, like when I had to go back and add the Chip Maker manually for most of the entries.
    Crunching for Team TPU
  25. Bot

    Bot

    Joined:
    Apr 14, 2009
    Messages:
    584 (0.30/day)
    Thanks Received:
    105
    BoT | Intel | Q9450 @ 3.6 | 47.2457 | 13.1238 | 64 bit

    burnin_06.jpg

    really nice scalling with the cpu and fsb speed combination in this bench

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page