Prime 95 - inconsistent results!

lookin4dlz

Senior member
May 19, 2001
688
0
0
After upgrading my bios, I started playing around again with getting more speed. In running the obligatory P95 torture test, I've noticed that P95 doesn't give the same results at the same settings. Several times. I've run P95 solid for 5 minutes, stopped it, waited for the computer to cool down & run P95 before it has errors after one minute. Last night, I ran P95 w/o error for 30 minutes before stopping it, played Civ3 for a couple of hours & everything was fine. I re-booted & was going to run P95 all night, but went to bed instead. This morning, I got up & was going to run P95 all day while at work, but it only ran i minute before it had errors.

Does P95 run the same tests in the same order every time you use it? If not, perhaps that's why I'm getting inconsistent results.
 

rogue1979

Diamond Member
Mar 14, 2001
3,062
0
0
Sounds like some instability related to a heat issue. When your computer is cooler it runs Prime95 longer without errors. If it fails after one minute you definately have something overclocked too far, either your memory timings or your cpu speed.
 

Shimmishim

Elite Member
Feb 19, 2001
7,504
0
76
we put too much faith in prime.. :)

there's a thread about it...

someone even ran it for days and days before it finally errored...

inconsistent it is...
 

lookin4dlz

Senior member
May 19, 2001
688
0
0
Originally posted by: rogue1979
Sounds like some instability related to a heat issue. When your computer is cooler it runs Prime95 longer without errors. If it fails after one minute you definately have something overclocked too far, either your memory timings or your cpu speed.

I'll test it some more to see if it's heat-related, but I'm not going over 47C or so during testing.

 

gururu

Platinum Member
Jul 16, 2002
2,402
0
0
Overclocked components sometimes give unrealized errors upon reboot. It is definitely an indication of instability.
I wouldn't put the blame on Prime95 just yet. See how consistent the program is at default settings.
I don't know why it is, but I would imagine that every boot results in slight voltage variances across the board.

I've seen it big time with video cards in particular.
 

Jojo1971

Platinum Member
Apr 18, 2002
2,200
0
0
one thing for sure your system isnt stable, you probably have to lower your fsb and see what happens when you run prime95.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I am one of those ones who used to have more faith in prime95...Don't get me wrong I think errors in prime95 need to be taken seriously, but definitely take into considerations other stress programs you can run....

With my chip and similar to other high clocked p4 c1 chips I actually get opposite results to you.....


I can run 5-12 hours with no errors and yet could fail a Divx encode in 10 minutes...OK you say Divx is a buggy codec, right??? I then ran Besweet mp3 encoder and errored out....Then ran autocadd 2002 and ran render on a large 3d file and it errored out...Ran memtest for 10passes all test with cache on and no errors....Ultimately I raised vcore .025v and the error went away. I ran into later again and once again prime95 says stable as can be for 12 hours yet I error in Divx and autocadd....This time I needed to raise vcore .05v to get stability but I achieved it....


My case is why couldn't prime95 find this instability?? I like many in the past could run all these programs on an xp or even a p4 non c1 stepping chip oc'd to unstable errors per prime95 yet never error in these programs....Makes you wonder....


I think prime95 needs an update to handle these newer chips with advanced hardware prefetch, and other advancements that may be giving us an unrealistic view.....


Many p4 c1 stepping users had very similar results to mine....We do not trust prime95 as the definite answer to stability no more....
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
By the way run memtest86 as sometimes bios updates have an unfortunate ability to fix a bug they intended to to only create 2-3 new problems...maybe timings associated with the memory have been tweaked and thus is related to your new inconsistent results....


If all fails go back to previous bios and see if the problem remains...
 

lookin4dlz

Senior member
May 19, 2001
688
0
0
Thanks for the corrobration Duvie. Bottom line is that if it works for the apps you run w/o error, then all is well.