mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2006-08-28, 11:32   #1
Chocolinx
 
Aug 2006

5 Posts
Exclamation Prime95 Error With Stock P43.2E

Lately my I've been having several crashes happening very randomly but they happen frequent enough that it's been bothering me. So I decided to run Prime95 to see if any errors would pop up. And here are my results after my first run:
[Sun Aug 27 01:30:33 2006]
Self-test 1024K passed!
[Sun Aug 27 01:48:33 2006]
Self-test 8K passed!
[Sun Aug 27 02:04:19 2006]
Self-test 10K passed!
[Sun Aug 27 02:19:25 2006]
Self-test 896K passed!
[Sun Aug 27 02:36:48 2006]
Self-test 768K passed!
[Sun Aug 27 02:52:08 2006]
Self-test 12K passed!
[Sun Aug 27 03:09:48 2006]
Self-test 14K passed!
[Sun Aug 27 03:27:23 2006]
Self-test 640K passed!
[Sun Aug 27 03:43:10 2006]
Self-test 512K passed!
[Sun Aug 27 04:00:14 2006]
Self-test 16K passed!
[Sun Aug 27 04:17:44 2006]
Self-test 20K passed!
[Sun Aug 27 04:34:02 2006]
Self-test 448K passed!
[Sun Aug 27 04:50:37 2006]
Self-test 384K passed!
[Sun Aug 27 05:06:43 2006]
Self-test 24K passed!
[Sun Aug 27 05:23:25 2006]
Self-test 28K passed!
[Sun Aug 27 05:39:39 2006]
Self-test 320K passed!
[Sun Aug 27 05:56:30 2006]
Self-test 256K passed!
[Sun Aug 27 06:12:44 2006]
Self-test 32K passed!
[Sun Aug 27 06:29:43 2006]
Self-test 40K passed!
[Sun Aug 27 06:46:40 2006]
Self-test 224K passed!
FATAL ERROR: Rounding was 0.49609375, expected less than 0.4
Hardware failure detected, consult stress.txt file.

Right now I'm doing a second run but it's been 12 hours into the run and so far no errors. Continuing to run it through the night but do you think there's a problem with my CPU by any chance? I've tested my memory before too and there hasn't been any problems with those. If the CPU is the problem I'm sending it in for Warranty immediately.

Biggest problem for me was when I was playing Final Fantasy XI and in the middle of a HUGE fight it crashed suddenly >.< Got really annoyed because it happend a few times while playing my game.
Chocolinx is offline   Reply With Quote
Old 2006-08-28, 12:30   #2
Cruelty
 
Cruelty's Avatar
 
May 2005

31308 Posts
Default

Check the temperature of your CPU while stress-testing, maybe it is overheating and the existing cooler barely manages to keep CPU within operating limits.

BTW: which stress test do you run?
Cruelty is offline   Reply With Quote
Old 2006-08-28, 13:57   #3
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

7×1,051 Posts
Default

With one error in 8 hours it will be very tough to find the source of the problem. It could be bad memory, bad CPU, weak power supply, heat, etc. I think that chip is hyperthreaded, if so run two copies of prime95 to increase the stress.

Check your temperatures and voltages with freely available programs. Then you can try the following (one at a time until the problem goes away) if your BIOS allows it:

1) Increase voltage to ram slightly
2) Increase voltage top CPU slightly
3) Try relaxing your memory timings.
4) Swap memory with a friend's computer to see if the problem goes away
5) If temps are high install more case fans or replace the heat sink
6) If the computer is under warranty, return it for a replacement.

Last fiddled with by Prime95 on 2006-08-28 at 15:03
Prime95 is offline   Reply With Quote
Old 2006-08-28, 17:09   #4
Chocolinx
 
Aug 2006

5 Posts
Default

So far I've tested my memory with Memtest and after 24 hours of testing memory, memory seems to be perfectly fine.

In terms of CPU heat, my current MAX temp is 61ºC idle is around 38-40ºC. I have a Scythe Mime on my CPU so the cooling is sufficient I think. All the temps were recorded with SpeedFan4.29. I'll try running Prime95 with two open but i was told that Hyperthreading isn't the same as Dual Core and I shouldn't have to run two for HT. I'll post again if I run into more errors.
Chocolinx is offline   Reply With Quote
Old 2006-08-28, 17:25   #5
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

7×1,051 Posts
Default

Quote:
Originally Posted by Chocolinx View Post
So far I've tested my memory with Memtest and after 24 hours of testing memory, memory seems to be perfectly fine.
This does not prove the memory is OK. There have been numerous instances where Memtest passed and the memory turned out to be the culprit.

Quote:
In terms of CPU heat, my current MAX temp is 61ºC idle is around 38-40ºC. I have a Scythe Mime on my CPU so the cooling is sufficient I think. All the temps were recorded with SpeedFan4.29. I'll try running Prime95 with two open but i was told that Hyperthreading isn't the same as Dual Core and I shouldn't have to run two for HT. I'll post again if I run into more errors.
Temps are OK. Running two prime95's (each using half as much memory) does put a little bit more stress on the CPU.

You've got a real tough case as you are just on the edge of being stable.
Prime95 is offline   Reply With Quote
Old 2006-08-29, 03:56   #6
Chocolinx
 
Aug 2006

5 Posts
Default

I tried running two copies of Prime95 and I don't think it worked. I must've done something wrong. Because it didn't move past the first rounds of test on either after 8 hours.

I read the readme file but I think I still did it wrong. Could you please try explaining to me how I should do it?

Oh another small update tibit. So my second round of Prime 95 that I had running before trying the double Prime, it ran through it without an error after 48 hours. Now does seem a little odd? I didn't reset my computer at all in between trials either. If it was just a RAM error too wouldn't the computer just crash my programs and not my whole system? The resets while playing games are very very random.

One more thing probably pretty major, now don't laugh: Recently my Arctic Freezer4 heatsink had it's fan die on it! So the computer was running for about a week without a CPU fan but the system fans seem to keep it cool enough with full load only being about 70, so well below throtle. Replaced it with Scythe Mime. Now during the process of replacing the heatsink I had to remove the CPU to clean off the old grease and place new grease on it. But somehow when taking it out one pin got bent in the process but I bent it back and it seem to slide back in the slot just fine. Would this cause my CPU to crash my system too?
Chocolinx is offline   Reply With Quote
Old 2006-08-29, 12:24   #7
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

7·1,051 Posts
Default

Quote:
Originally Posted by Chocolinx View Post
I tried running two copies of Prime95 and I don't think it worked. I must've done something wrong. Because it didn't move past the first rounds of test on either after 8 hours.
In both prime95s, use the custom torture test with half as much memory or run the in-place torture test.

Quote:
Oh another small update tibit. So my second round of Prime 95 that I had running before trying the double Prime, it ran through it without an error after 48 hours. Now does seem a little odd?
No.

Quote:
If it was just a RAM error too wouldn't the computer just crash my programs and not my whole system?
It could do either.

Quote:
One more thing probably pretty major, now don't laugh: Recently my Arctic Freezer4 heatsink had it's fan die on it! So the computer was running for about a week without a CPU fan but the system fans seem to keep it cool enough with full load only being about 70, so well below throtle. Replaced it with Scythe Mime. Now during the process of replacing the heatsink I had to remove the CPU to clean off the old grease and place new grease on it. But somehow when taking it out one pin got bent in the process but I bent it back and it seem to slide back in the slot just fine. Would this cause my CPU to crash my system too?
I think you are OK with the pin repair. IMO, a broken / damaged pin would lead total failure not a rare random failure.
Prime95 is offline   Reply With Quote
Old 2006-08-29, 18:48   #8
Ethan Hansen
 
Ethan Hansen's Avatar
 
Oct 2005

23×5 Posts
Default

A few pointers:
  1. A broken pin may or may not be immediately apparent. A large portion of the pins are either Vcc (supply voltage) or Vss (ground). The effects of a missing pin will not be felt until the portion of a circuit that pin feeds draws high current. Prime95 heavily exercises much of the CPU, so this could well cause an intermittent failure. That said, I would not focus on the pin for now. I have bent and straightened far too many pins to count on processors over the years and only rarely did it lead to a broken connection.
  2. Did you run the original memtest program or the more recent memtest86+? If you have not run the new version, download it and let it churn through a pass or two. The newer version does a far more thorough job at exercising today's CPU and memory configurations.
  3. Fire up ThrottleWatch while you are crunching primes and blasting bad guys. Enable logging in ThrottleWatch so you do not need to monitor the graphs. This will flag any instances where the CPU throttled itself due to overheating. If any clock slowdowns are seen, this points to a less than ideal thermal interface between CPU and heatsink. Take the assembly apart, thoroughly clean both surfaces with a solvent, and reassemble.
  4. If your motherboard supports it, try boosting the CPU voltage slightly. If Prime95 and games work without crashes, your CPU does not have the margin necessary for intensive workouts. If it is still under warranty, you should be able to get a replacement.
  5. If none of the above does the trick, try slightly boosting the voltage to your memory (assuming your MB supports this). If all functions correctly, there is a memory marginality. As most memory modules these days come with lifetime warranties, you should be able to get a replacement.
Ethan Hansen is offline   Reply With Quote
Old 2006-09-04, 22:11   #9
Chocolinx
 
Aug 2006

5 Posts
Default

Quote:
Originally Posted by Ethan Hansen View Post
A few pointers:
  1. A broken pin may or may not be immediately apparent. A large portion of the pins are either Vcc (supply voltage) or Vss (ground). The effects of a missing pin will not be felt until the portion of a circuit that pin feeds draws high current. Prime95 heavily exercises much of the CPU, so this could well cause an intermittent failure. That said, I would not focus on the pin for now. I have bent and straightened far too many pins to count on processors over the years and only rarely did it lead to a broken connection.
  2. Did you run the original memtest program or the more recent memtest86+? If you have not run the new version, download it and let it churn through a pass or two. The newer version does a far more thorough job at exercising today's CPU and memory configurations.
  3. Fire up ThrottleWatch while you are crunching primes and blasting bad guys. Enable logging in ThrottleWatch so you do not need to monitor the graphs. This will flag any instances where the CPU throttled itself due to overheating. If any clock slowdowns are seen, this points to a less than ideal thermal interface between CPU and heatsink. Take the assembly apart, thoroughly clean both surfaces with a solvent, and reassemble.
  4. If your motherboard supports it, try boosting the CPU voltage slightly. If Prime95 and games work without crashes, your CPU does not have the margin necessary for intensive workouts. If it is still under warranty, you should be able to get a replacement.
  5. If none of the above does the trick, try slightly boosting the voltage to your memory (assuming your MB supports this). If all functions correctly, there is a memory marginality. As most memory modules these days come with lifetime warranties, you should be able to get a replacement.
Okay I've done most of the above lol. So far no change in Voltage changes. Still resets under very heavy CPU load but may not happen over 24 hours but it does eventually happen. Sometimes even happens within 5 hours. Very random shut downs.

I tried disassembling it and reapplying paste and placing it back on still same problem, no change. I've used both Memtest86 and Memtest86+ both come back to me with no problems everytime! Even after 48 hours! Weird? Prime 95 hasn't given me an error lately either. I do remember when I pulled off the Arctic Cooling HeatSink though the CPU came out in a very bad way! The heatsink took the CPU with it without having the Socket Lock undone. Could that have damaged the CPU?
Chocolinx is offline   Reply With Quote
Old 2006-09-06, 16:48   #10
Ethan Hansen
 
Ethan Hansen's Avatar
 
Oct 2005

23×5 Posts
Default

Having the processor come out along with the heatsink is common. Thermal grease can create a strong seal. In the future, release the socket clamp before removing the heatsink and processor. I doubt this caused any damage.

Did you try running ThrottleWatch? If not, do so. This has the marks of a CPU problem. If there is a thermal problem, you may see throttling.

-> So far no change in Voltage changes.

I'm not sure what you mean by this. Did you try increasing the CPU core voltage? If not, and your BIOS supports it, do so. Run ThrottleWatch constantly -- if there is thermal throttling going on, increased Vcc will make it more apparent.
Ethan Hansen is offline   Reply With Quote
Old 2006-09-06, 17:55   #11
nomadicus
 
nomadicus's Avatar
 
Jan 2003
North Carolina

2×3×41 Posts
Default

I had the same problem. Was running just fine for many months when it went unstable similar to you.
I made the small adjustments in voltage mentioned in the previous post and became suspicious of my memory.
I double checked my CPU and memory settings with cpuz. Then looked at the mobo. I found that the motherboard set the voltage to 2.6V when the memory manufacturer's website said 2.8V. The default voltage bit me on this ASUS motherboard.
fwiw - I know some DFI motherboard undervolt the CPU.
Moral of the story: assume nothing is the way it is.
nomadicus is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
prime95 error sam1 PrimeNet 2 2016-09-04 11:40
Prime95 version 28.9 Safe for 5960x on stock speed? smaltese Information & Answers 6 2016-05-01 18:02
Prime95 - ERROR pls help! Capio Hardware 3 2012-02-11 20:04
Torture test rounding error on stock 7750 buildup jwh Information & Answers 8 2009-06-30 20:51
Prime95 v.25.5 FATAL ERROR Keepcase PrimeNet 1 2008-03-07 11:05

All times are UTC. The time now is 23:45.

Fri Mar 5 23:45:44 UTC 2021 up 92 days, 19:57, 0 users, load averages: 1.15, 1.07, 1.18

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.