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

OC Benchmark BS???

Discussion in 'Overclocking & Cooling' started by frankenchrist, May 1, 2007.

  1. frankenchrist

    frankenchrist New Member

    Apr 14, 2007
    249 (0.06/day)
    Thanks Received:
    I have gotten my system stable in all benches....:pimp:
    yesterday I changed my drivers and all benches fail eventually...*Even Everest*:banghead:

    All but memtest which roxx without soxx!!:rockout:

    Now I am stunned:shadedshu
    And pissed:cry:

    Are all these benchies neccesary and are there any I can truely trust as a good bench???

    My system is running Clam cold under water and I am under the ussumption that my system is great but my processor is just wigging... BUT! I have been told by quite a few that this MF..er will do min of 500 bonus...

    How high can I go with my Voltage on 130nm under water.?

    I love asking questions because I have learned A hella lot in the last few fron this site!

    Hope I can help you someday...

    10 Year Member at TPU
  2. Kursah

    Kursah Moderator Staff Member

    Oct 15, 2006
    10,834 (2.69/day)
    Thanks Received:
    Missoula, MT, USA
    Your experience will help a lot of people, as everyone has slightly different set-ups...I am pretty sure noone here has the same EXACT setup, which is great. I have learned much, and I have recently been able to help those in need. Can you list the benchmarks you use? I recommend Orthos at Priority 9 w/small fft's for cpu only, if you can pass this for at least 8 hours, you should be good. Then I use Orthos Priority 1 w/small fft's, ATI Tool w/fuzzy cube open and memtest for windows (to test all but about 70-100mb of system memory) to really stress my system. You may have to modulate, the more CPU power you use, the slower the fuzzy cube goes, I learned using artifact scanner in these instances can help work out that GPU and that's also why I use Priority 1... I'd go from about 600fps in fuzzy cube to less than 60 if I ran orthos in priority 9.

    If you are failing benches, roll back to when you were using older drivers. For some setups, certain drivers for certain peripherals will cause issues with others. Also depends on OS. Also I'm sure you do, but make sure you have the best rated version of Bios for your MB...I say best rated because the newest here is not always the best just like drivers. Do one at a time, and bench. Make a notepad or document and write down what scores you get with what settings...a demo of what I do:

    Kursah's 3dMark Scores:

    FSB 266, CPU 1.86, Memory strapped to 800 w/4-4-4-12, 7.2 cats, bios 807
    FSB 480, CPU 3.36, Memory 1:1 @ 960 w/4-4-4-12, 7.4 cats, bios 1004
    05- 12,000
    06- 5500

    And I've done one for memory tests in Everest (all 4 RAM tests) at different speeds, that way I have a definate reference to see improvements for what settings, and decide if it's worth pushing that way or not. As I learned with my p965, going from 4-4-4-12 to 4-4-4-5 wasn't worth much until I messed more with subtimings, then I went from an average of 20-50mbs and .01ms improvement to 50-100mbs and 0.5ms improvement.

    Remember though, synthetic benches can only emulate slight real world performance, and we use a lot of them here for either stability or bragging rights. I like the challenge of squeezing every ms, mbs and point out of my system, but the down side, it probably helped to kill my MB...of course I wasn't pushing nearly as hard on my P5B as a lot of folks so maybe I got a lemon....

    If I were you though, I'd start back at where you were, and then move forward at a slower pace, verify the updated drivers are ones you need, have features that help you and not hinder you, etc. Document what you get, how long you were able to pass orthos. Keep it updated, and you will not only learn a lot, you will be a lot more knowlegable and successful!
    10 Year Member at TPU

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