MMM
Results 1 to 25 of 291

Thread: More results with ASUS GTX580 DirectCU II

Threaded View

  1. #11
    Nice Hardware!!!
    Join Date
    Jan 2006
    Location
    Finland/Tampere
    Posts
    1,846
    Today I did some testing with my second Direct CU II card. This card is very impressive, but I did not manage to get satisfying results out today. I thought to share my findings.

    The way I test the card is following:

    Voltages I use are:
    (Antec HCP PSU or corsair AX 1200W)
    VGPU 1.55V-1.6V
    VPLL stock -1.35V
    VMem 1.88-1.9V
    (Full pot temperature from sensor is -174, so values are not 100% correct. They will just give the idea about it)

    1.) Use OCZ freeze thermal paste. It is clearly the best for this use. Use a lot of it.
    2.) Cool down to -60C and see if 1200MHz works. If no artifacts and issues, go to step 3.
    3.) Cool down to -80C and see if 1300MHz works. If no artifacts and issues, go to step 4.
    4.) Cool down to -120C and see if 1400MHz works. If no artifacts and issues, go to step 5.
    5.) Cool down to around -130C and try to run Hwbot Unigine heaven at 1450MHz. Go colder after the start of the benchmark.

    The issue I have is, that when I go over -135C without 3D load, I hear that crack or snap and card wont run at all. The contact between GPU pot and core just get loose and the bondage between them is lost.
    Then i just back down to 1200MHz and let Hwbot heaven benchmark run 30 seconds. After that you are usually able to continue

    The load these cards produce is so huge, that contact between GPU and pot must be optimal. I think my main issue have been too tight and too firm contact. I need to learn how to install pot "badly" to get it working better..locigal haha.

    So the ones who struggle with their cards, start slowly with temperature and check that pattern I wrote with the clocks your card can do for sure.
    Most of them should be able to hit those same values with same voltages and temperatures. If you have direct lockup in the beginning, your contact have gone bad by too cold temperature. Then you need to back down and try again after short warm up.
    If you have colorful "stars" on the screen it is most likely bad contact and too high clocks for it. The same frequency can usually be done without any "stars" on the screen, when contact is ok.

    I don't have too much time to bench before end of the upcoming summer, so this will be bugging me for sure.
    I was 8 seconds away from finishing the benchmark today and I think that is it from this spring

    Have fun with your oc adventures! We will see later this year!
    Last edited by SF3D; 03-27-2011 at 01:26 PM.
    You are as good as your samples are!

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
  •