Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Results 1 to 3 of 3

Thread: "Funny" problem with Q6600 on GA-P35-S3G

  1. #1
    Join Date
    Jan 2009

    Default "Funny" problem with Q6600 on GA-P35-S3G


    I seem to have made a bad pick on my Q6600 cause I can only reach 3,2GHz on 1,5V Vcore - so I stick with 3Ghz @ 1,3875V for now.

    This 3GHz setting runs stable in regards of crashes, but something mad happens sometimes which I can't explain.
    I am using EIST - as most of the time I don't need the 3GHz and the 2GHz (334 x 6) is still much more than needed. Transition from 6x to 9x works and back works nicely (in regard to crashes).

    My problem is, that, even when running prime95 torture test (blend) with 4 Threads and a 100% CPU-load on all cores at start EIST changes the multi to 9x, so cpu runs as expected on 3GHz, but quite reproducable (also not always) when Test1 (4000 Lucas-Lehmer iterations using FFT 1024k) finishes on one core and starts Test2 th multi drops to 6x and stays there.

    If I restart the torture test I am often (not always) again @ 9x, but probably will drop as soon as Test2 starts to 6x again.
    This problem also happens when running at stock speed (6x/9x 266) of the cpu!

    I have this problem on XP 32-Bit and Vista 64-bit.
    cpu-z and realtemp are used to check the actual core speed.
    Core temps are on 50-55 C and quite constant on load with either 6x or 9x.

    Any ideas why this happens and how to solve this are welcome as I guess this is not just a prime problem and it might happen when I really want the 3GHz ;)
    Last edited by egalus; 01-17-2009 at 10:53 PM.

  2. #2
    Join Date
    Jan 2009

    Default Re: "Funny" problem with Q6600 on GA-P35-S3G

    I did some more testing. First I upgraded the bios from 5a to the last beta 5b and disabled C1E (Eist still enabled), than I set everything to spec values regarding cpu,ram,pcie...

    Reran the test and boom, same "fall-back to 6x" problem again.

    By accident I found something new, when changing the displayed core in cpu-z from core#0 to core#2 the multi goes up from 6x to 9x.
    Checking this further it seems that core#0 and #1 run at 6x, core#2 and #3 run at 9x.

    To rule out that just cpu-z has wrong reading on multi for the first of the two dualcores I let the test run for some time and already at Test4 the #0 and #1 are one test behind #2 and #3, which indicates that the multi reading is right.

    Now I am really puzzled.

  3. #3
    Join Date
    Jan 2009

    Default Re: "Funny" problem with Q6600 on GA-P35-S3G

    Further tests suggest that TM2 is the problem.

    As soon as coretemp of a core (the hottest on my quad is #1) hits 58+ C it seems like TM2 drops the multi to 6x. I have no idea why it does that cause I just hit 58-59 on this one core (when overclocked to 3GHz) on full-load after 15 minutes and do not go up from that (with disabled TM2, so still at 9x). 2 Seconds after I stop the torture test temps drop to 40-43 C.
    So I guess I will keep TM2 off.

    But another funny investigation:
    On Vista x64 it's always (even when running on 9x) worker#1 (core #0) and worker#2 (core#1) that are slower in doing the tests (by about 30-40%!) while on XP it's woker#1 (cpu#0) and worker#4 (cpu#3).

    Is there any way to test if all cores perform equally? As this seems strange for me too.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

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