PC Freezing on High CPU Load Only

mayest

Senior member
Jun 30, 2006
306
0
0
[*]Overview of the problem
3-year old home-built PC freezing when under heavy load.


[*]Full description of the problem and symptoms
I built this PC three years ago just for fun and to test the Vista beta. It runs fine until I use some software that really taxes the CPU (e.g., HandBrake, Prime95, SuperPi, etc). It seems that anything that continuously runs the CPU at 90% to 100% load will eventually cause the PC to freeze up. Almost always it is a hard freeze with no error message (no BSOD). The screen shows everything as it was when it froze. Sometimes, the PC just shuts off with no message.

Running the current version of SpeedFan, I found that the CPU (Core 2 Duo E6600) was hitting about 60 degrees C at the time of the freeze. Increasing the heatsink fan speed seemed to help, but just made it run longer before freezing. Gigabyte's EasyTune utility (from 3 years ago) showed the CPU temp at about 30 degrees C, but I think that isn't measuring the core temps as the newer SpeedFan does (I recall this being an issue when the Core 2 Duo's first came out).

I want to think that this is a temp problem, but it seems like it shouldn't shut down at 60 degrees C (even though Intel says that is the max acceptable temp).

[*]Did it work normally at one time, or has the problem always existed?
The problem has always existed since I built the PC.

[*]Is the problem consistent and repeatable, or entirely random, or semi-random?
It is repeatable. I just run HandBrake for awhile (time varies from a couple of minutes to maybe 50 minutes or so). Again, the PC seems to be fine if the CPU isn't heavily taxed.

[*]I already tried these steps:
Replaced the stock heatsink/fan with an Arctic Cooling Freezer 7 Pro today. Ran Prime95 for awhile (10 minutes?) and temps were about 50 degrees C when the PC froze. This didn't seem to affect the idle temps, but under load they were about 10 degrees C cooler. That seems to suggest that CPU temp may not be the problem.

Updated BIOS to the most recent (rev 14, IIRC). No change.

Ran Windows memory diagnostic with no known errors.

Ran MemTest86+ for over an hour with no errors yet (still running).

[*]My software:
  • Operating System: Originally XP (up to date)/Vista (not sure of SP level) dual boot. Now, Win 7 RC/Vista dual boot.
  • Antivirus: None on Win 7, ESET Nod 32 on Vista - this is not a virus problem.
  • Firewall: Hardware (router) and Windows firewall (all versions at default settings).
  • The various O/S's seem to work fine as I noted, until I run something that really works the CPU.



[*]My hardware
  • Home Bulit by me :)
  • Motherboard: GIGABYTE GA-965P-DS3 LGA 775 Intel P965 Express ATX
  • CPU: Core 2 Duo E6600 2.40 Ghz
  • Video card: GIGABYTE GV-NX76T256D-RH GeForce 7600GT 256MB 128-bit GDDR3 PCI Express x16 (Fanless, runs around 40 degrees C)
  • RAM: CORSAIR 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 667 (PC2 5300) (voltage is BIOS default)
  • Power supply: Antec 450 watt SmartPower 2.0 (came in Antec Sonata II case)
  • Hard Drives: 2 Seagate Barracuda 7200.10 ST3320620AS (SMART reports them being healthy)
  • Case Fans: 2 one pulling in from front, one pushing out back. Plenty of ventilation.

No other hardware. No over-clocking. No funny settings. Everything is standard.

[*]Other information that might be relevant
I can't think of anything else to add. I thought that it was heat related, or bad RAM, but I would have thought that the Freezer 7 Pro would have solved the CPU temp problem and I haven't found any RAM errors yet.

Any ideas?

Thanks in advance.

Tim
 

RadiclDreamer

Diamond Member
Aug 8, 2004
8,622
40
91
If temps are showing good and it only freezes during load, my next guess would be a power supply issue, have you chcked your voltages?
 

mayest

Senior member
Jun 30, 2006
306
0
0
Thanks. The power supply is something that I have suspected since the beginning, but I don't have any obvious (to me) evidence that it is failing. Very well could be, though.

Is there any way that I can diagnose a bad PSU in the PC? Or, should I just buy a new one?

Many years ago, I had to replace power supplies a couple of times in purchased PCs. IIRC, though, that was because they simply refused to power up.
 

itrfuzzylogic

Senior member
Feb 13, 2002
491
0
0
Honestly I would just buy a new one, I've seen multiple unexplained issue and the problem cause was the PSU. Don't waste your money on a cheapie $20 psu from newegg either, you know at least get something respectable like a 500w antec for like $30-40 on sale at like a local store. I have had great success with them repairing hundreds of computers!
 

mayest

Senior member
Jun 30, 2006
306
0
0
Thanks, that's exactly what I did. I got a PC Power & Cooling Silencer 610. Overkill, but I should be able to use it in the future and I've had good luck with them in the past. Plus it was only $100 shipped. Hopefully that will solve the problem. If so, then I can turn this from a test rig/backup device into an HTPC.
 

mayest

Senior member
Jun 30, 2006
306
0
0
Ok, if anybody is still paying attention. :)

I got the new PSU installed. No change, still having the same problem. So it wasn't the PSU, but I'm still happy to have the new one.

I ran Prime95 again, but on a different setting. There are three choices (Small, In Place, and Blend). Blend is the default, and it says that it uses lots of RAM. Small uses very little RAM and primarily exercises the CPU. I had been running it on Blend when it was crashing. This time I tried Small and it ran for an hour with the CPU pegged at 100% and did NOT crash.

So that seems to point to a RAM problem. I removed one stick and ran Prime95 on Blend. It ran for nearly an hour before the PC rebooted. So, that is significant progress, but in the end it still did the same thing.

I now think that this is either a RAM problem or a motherboard problem, and I'm leaning toward the motherboard. The reason is that the memory has tested good several times using MemTest and the Windows memory testing utility. Also, the motherboard (GIGABYTE GA-965P-DS3) was a brand new design when I bought it. It was Rev1, and they came out with a couple of new revisions later. In fact, I got the board shortly before the first Core2Duo chips shipped, so it wouldn't be surprising if the board has some obscure problems.

Does that make sense?