stability testing help

Seance

Geek Trainee
AMD Athlon 64 3500+
ASUS A8N-E Nforce4 Ultra
1gb Corsair Value Ram pc3200

I'm a bit confused with some results I'm getting with Prime95...
I have an AMD Athlon 64 3500+ oc'd from 2.2 ghz to 2.52 ghz (252x10) and it appears to be stable through almost all testing I've done...
I use 252x10 because I set my memory to DDR333 instead of 400 in bios (I have corsair value ram so I could never push it past DDR428)... and with this configuration my ram runs at CPU speed divided by 12 (found that out through cpu-z :p )... so right now... 2520/12 = 210 x2 = DDR420 which the ram can handle well (also ran memtest for just under 4 hours today to make sure with no errors reported, 13 passes, when usually if I have any problems it'll fail at around 60% or before on the first pass)

for awhile I was using 255x10 however my computer froze during a video in winamp (this was after quite a few hours of using that fsb, even playing HL2, so it seems like it was JUST a tiny bit too much for it to handle), so I took it down to 252

I ran Prime95 for 8 hours over last night while I was asleep (Blend torture test) and woke up to find 0 errors... however I then started the In-place FFT torture test and it lasted 7 minutes until getting...

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

I used to get this error during the blend test when I had pushed my memory a bit too far, and memtest would also fail after the first 5 minutes in that case... but now I'm getting it during this test... and I have no idea what it means...
I've read a few times before people saying that Prime95 isn't very reliable in testing AMD 64, though I'm not sure the reasoning, but I thought maybe that could be the issue

however... as I said
I have completed 8 hours of Prime95's blend test
4 hours of memtest
and I even completed the 32M calculation in SuperPI perfectly fine in 32 minutes

should I worry about that error?
 
This isn't really my area (what is?!) but judging from your blend test and the fact that Prime95 isn't very reliable in testing AMD 64 I don't think you should worry about the error message. Post again if a problem occurs which you think may be due to this and maybe then we could establish why it comes up with the error message but I think it should be fine :good:
 
For best results people stress their systems for 24 hours, so only doing it for a few hours may not reveal any apparant errors. Try doing the test again to see if the same thing happens. If it does, then I would recommend lowering your FSB a tad, since overclocking too far can slowly eat away your system. But if its ok or your PC runs smoothy in everything else than your fine.
 
While I am big on Prime 95, I too have been hearing about it not being a good stress test for Athlon 64's. Another thing that'll put it under load is Folding @ Home. Besides a blatant recruitment for our team, I have noticed that it pounds the CPU really hard. I've also heard of people running PiFast as well, but haven't run that myself.
 
Big B said:
...I've also heard of people running PiFast as well, but haven't run that myself.
I have, and it doesn't seem to hammer your resources quite like F@H does. Besides this, it typically runs in the foreground, making your system unusable until the tests are complete.
 
Back
Top