PDA

View Full Version : Weird Prime95 message after OCing


superklye
08-01-04, 09:27 AM
I have an Athlon 2600+ and have wanted to use my DDR400 RAM at that speed instead of underclocking to 333. So last night I decided to again try OCing and see where it got me.

First thing I did was bump the core from 1.65 to 1.675 and put the FSB at 200, CPU multiplier at 10 and then save/exit. It posted fine. I tried to do this the other day without upping the vcore and the system wouldn't post, resulting in me having to reset the freakin CMOS (which is a pain).

Get to Windows, start up AIM and Far Cry and played for like 30 minutes with no problems at all.

I restarted again put the multiplier at 11 and get back to Windows, but the system crashed before ever even getting in. I set the vcore to 1.7 and got into windows just fine. Did the same AIM/Far Cry test and the computer was working just fine.

I decide to fire up Prime95 finally and see what it's all about. It went for 52 minutes before getting a fatal error and stopped. So I reset and put the vcore at 1.725. By now it was almost 1am and I was tired, so I decided to just let it go while I slept. I woke up around 3:45 and the computer was at the Windows login screen, so I'm assuming the computer crashed.

I reset again and put the vcore at 1.75 and did the same thing. I woke up at 5:40 and the test had again failed after like an hour. Reset again, put the vcore at 1.775.

By now, the computer was at like 53c idle. I fired up Prime95 and it was almost immediately at 72c. That isn't good. But I let it go. I woke up again around 8:30 and the test had, again, stopped. But now this is where it was kinda weird and the results are posted:

[Sun Jul 31 00:34:31 2005]
Self-test 1024K passed!
[Sun Jul 31 00:50:10 2005]
Self-test 8K passed!
[Sun Jul 31 01:05:56 2005]
Self-test 10K passed!
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Sun Aug 01 01:39:28 2004]
FATAL ERROR: Resulting sum was 3854852623140008, expected: 3854852623107246
Hardware failure detected, consult stress.txt file.
[Sun Aug 01 05:56:13 2004]
Self-test 1024K passed!
[Sun Aug 01 06:11:35 2004]
Self-test 8K passed!
[Sun Aug 01 06:27:05 2004]
Self-test 10K passed!
[Sun Aug 01 06:43:30 2004]
Self-test 896K passed!
[Sun Aug 01 06:59:36 2004]
Self-test 768K passed!
[Sun Aug 01 07:18:39 2004]
Self-test 12K passed!
[Sun Aug 01 07:37:24 2004]
Self-test 14K passed!
FATAL ERROR: Resulting sum was 3.847555211654861e+016, expected: 3.847555211660104e+016
Hardware failure detected, consult stress.txt file.

Those are all the results for all of the times I ran the program but the part in question is the final one. It doesn't have the message on there but it was something like:

"2h 1m time, -1 errors, 0 warnings" or somethign...I don't remember exactly, but I remember it said -1 errors. What does that mean?

I'm kinda scared to OC my machine now...it was getting VERY hot and I fear that I may have damaged the CPU by having it running so hot for so long.

SigmaOrionis
08-01-04, 02:09 PM
Hi Superkyle,

Does your machine work fine at stock speeds with Prime95? If it does you didnt damage your hardware. If your machine gets too hot with Prime95 it will usually crash the program with an error and then cool right down.

Those messages you were receiving are general messages. The way Prime95 works is it does mathematical calculations that it already knows the answer to. It then checks the answer your computer gave against the known right answer and if its wrong will give you a hardware failure error.

I just tried overclocking my P42.4B to 3.0 ghz and my computer gave me the post beep. I got to see p4 2.98ghz 18x166 on the screen and then the memory count. At this point I was feeling pretty good. Then just before it initializes the cards and drives it froze. :( I also had to reset the CMOS to even be able to get the thing to repost. I think I've had my fill of overclockin for now. :)

superklye
08-01-04, 04:19 PM
Hi Superkyle,

Does your machine work fine at stock speeds with Prime95? If it does you didnt damage your hardware. If your machine gets too hot with Prime95 it will usually crash the program with an error and then cool right down.

Those messages you were receiving are general messages. The way Prime95 works is it does mathematical calculations that it already knows the answer to. It then checks the answer your computer gave against the known right answer and if its wrong will give you a hardware failure error.

I just tried overclocking my P42.4B to 3.0 ghz and my computer gave me the post beep. I got to see p4 2.98ghz 18x166 on the screen and then the memory count. At this point I was feeling pretty good. Then just before it initializes the cards and drives it froze. :( I also had to reset the CMOS to even be able to get the thing to repost. I think I've had my fill of overclockin for now. :)

I dunno how it performs at stock speeds...I'll give that a try tonight when I go out for a few hours and see what it says. But thanks for your help, and like you, I've kinda had my fill of OCing too. :)

npras42
08-01-04, 04:49 PM
By my reckoning you're just OCing too far. As said check without the OC and if its ok, then start going up at 5MHz increments on the FSB.

Oh and don't bother with FarCry. Only use Prime95 to test stbility. If it passes that, it will/should pass anything. The same cannot be said of FarCry.

Riptide
08-01-04, 05:38 PM
Heh, my system would pass Prime fine @ 208FSB but when I tried to run 3dmark03 or 2001 it would crash. Sometimes hard lock. Upping AGP voltage didn't help, either.

superklye
08-02-04, 07:07 AM
sweet.

i ran it over night for 7h 1m and there wasn't a single error. I think my CPU was fine and I think it was also fine at 3200+, just didn't have enough juice.