Prime-stable overclocking

[patrik]

I just want to share my experience of overclocking one of my latest computers. It is an ASUS P4P800S. I started the normal way of running the torture test overnight without errors.

Then I went to double-checks (DC's). The client doesn't detect all errors, so it's recommended to run a number of DC's and watch if they end up in lucas_v.txt (verified) of hrf3.txt (non-matching residue; either your test or the first one was bad (or both)).

Many people in these forums seem to think that around five DC's are enough, but my example below shows that the errors can occur quite irregularly. I will settle for 20 matching double-checks, but then I will still lower the FSB somewhat more just to be safe.

Below, I lowered the FSB by 1MHz whenever I got a non-matching result (triple-check needed). Note that I had 16 matching tests in April, but then I got a few bad ones.

C=     Cleared
Bad=   Verified bad
Triple=Triple-check needed
-=     Not in status files

BusMHz Exponent  Status   Returned
-----------------------------------------
FSB200 11392253  C        17-Feb-04 06:05
FSB264 11392259  Bad      22-Feb-04 23:43
   264 11466197  C        25-Feb-04 07:49
FSB263 11511427  Triple   27-Feb-04 22:45
   263 10694647  C        01-Mar-04 06:51
   263 10590257  C        03-Mar-04 19:07
FSB262 10653361  -        07-Mar-04 06:35
   262 10653989  C        09-Mar-04 19:33
   262 10744463  C        12-Mar-04 17:24
   262 11599409  C        14-Mar-04 17:57
   262 11132479  C        17-Mar-04 20:59
   262 11618623  C        19-Mar-04 21:31
   262 11638777  C        22-Mar-04 16:32
   262 11638819  C        25-Mar-04 06:14
   262 11673083  C        27-Mar-04 18:49
   262 11674613  C        30-Mar-04 20:02
   262 11708531  C        02-Apr-04 21:10
   262 11725981  C        05-Apr-04 04:34
   262 11726047  C        07-Apr-04 22:42
   262 11751463  C        14-Apr-04 16:47
   262 11792689  C        14-Apr-04 16:47
   262 11792701  C        15-Apr-04 19:57
   262 11846039  C        18-Apr-04 06:01
   262 11846053  Bad      21-Apr-04 05:22
   262 11878609  Bad      24-Apr-04 05:55
   262 11903431  Triple   26-Apr-04 07:26
   262 11933143  C        29-Apr-04 05:26
   262 11933213  C        01-May-04 16:49
FSB261 11976787  Bad      04-May-04 18:41
   261 11391823  C        07-May-04 05:24
   261 12030209  C        09-May-04 20:00
   261 11358587  C        12-May-04 16:40
   261 12070727  Triple   15-May-04 12:03
FSB260 12096209  C        17-May-04 20:03
   260 12124991  C        20-May-04 12:40
   260 12145319  C        23-May-04 07:14
   260 12168551  C        26-May-04 18:11
   260 11478787  C        29-May-04 08:26
   260 12225757  C        31-May-04 10:14
   260 12225761  C        03-Jun-04 05:50
   260 11566199  C        05-Jun-04 21:50
   260 12315461  C        08-Jun-04 15:56
   260 11251081  C        11-Jun-04 06:11
   260 11751851  C        13-Jun-04 23:05
   260 10765331  C        16-Jun-04 14:10
   260 11627801  C        19-Jun-04 06:02
   260 11853241  C        21-Jun-04 17:40
   260 10841683  C        24-Jun-04 05:25
   260 10848631  C        27-Jun-04 09:02
   260 12399109  C        30-Jun-04 15:13
   260 12446237  C        03-Jul-04 07:40
   260 11839669  C        06-Jul-04 21:24

[paulunderwood]

Below, I lowered the FSB by 1MHz whenever I got a non-matching result (triple-check needed). Note that I had 16 matching tests in April, but then I got a few bad ones.

Northern hemisphere warming up? Keep your machine clean! Be conservative about overclocking: If looks good then drop the FSB to allow for temperature variations!