just upgraded my son's gaming PC and am experiencing an odd problem that hopefully someone can offer some insight/advice on best approach for troubleshooting.
I just installed a 2.6ghz quad-core CPU (replacing a Core 2 Duo) and a Radeon 6870, as well as 8GB of gskill DDR2 1600 on a Gigabyte GA-EP43-UD3L motherboard. It has a single DVD drive and a 1GB 7200 RPM Seagate sata HD. 5 case fans. SB Audigy 2 audio card (disabled onboard sound). Also powering a mouse, a goofy mouse pad with flashing lights (don't ask) and a couple of 8GB USB flash drives he needs for school work.
I had some trouble getting the system to boot on first try after putting the new CPU in but eventually got it running after completely dismantling and putting back to together with 2 sticks of RAM instead of four and re-seating the CPU twice with newer thermal paste.
The weirdness started after he began gaming. After 5-20 minute sessions of BF3 or MW3 the system will shut down completely. On checking a HW monitor within W7 I noted the CPU temp is jumping from around 45 C idle to 64-65 on load within the first 2 minutes of gaming, then climbing to the current cut-off temp of 75 as set in the BIOS. I understand the HW monitor, nor BIOS' thermal measurements are terribly precise but I assume this to be a safe pointer that it's leaping too far, too fast and certainly hitting the BIOS shut-down temp.
The system runs without any issues when simply browsing the web, watching movies, etc. It's only under the load of gaming that it shuts down.
The GPU's temp never gets above 65 C with a custom fan/heatsink. The case's ambient temp is maxing out at 78 C and as stated above, the CPU idles at around 45 C. I know this is a little high for an idle temp but even with a stock cooler my though was it shouldn't jump more than 10-15 degrees. I've also reseated the CPU after a very careful clean and applying brand new A5 paste, having used some from a year old+ syringe on first try. I'm 99.9999% sure the CPU is seated perfectly.
Using the same HW monitor I've noticed the CPU fan speed appears to drop sometimes from 2200RPM (rough avg max) to 500 RPM. Also, perhaps more significantly, the voltage graph is a little wobbly. However, I'm certainly no expert so I'm not sure what level of fluctuation is considered acceptable/safe. The 12v rail for example is running at 11.8-12.32 with some notable quick drops. I have a PSU tester that seems to suggest the PSU is okay, but I also read these aren't terribly reliable? I can't find definitive numbers on what the different rails' allowable fluctuation might be. I can post the graphs here if useful, but all the CPU, DRAM, 3.3V, 5V and 12V all show what I'm technically calling 'wobbles'.
Here's where I found a possible clue as to cause. The 12v It will flat-line for minutes on end then will suddenly wobble with a .1 - .3 drop and then pull back up. The PC in question is on the same circuit as a Window AC unit and another relatively high end PC as well as an electric piano, lamps, etc. While I can't directly correlate the overheating shutdowns to the AC unit's compressor kicking in, it does seem to shut down more when the PC next to it is under heavier load and when the window AC unit is running for a while. Nothing else appears to be having trouble, however, and the breaker never blows with everything on.
The system has a 480W PSU. While online calculators seem to suggest this is just about adequate, I'm leaning heavily toward thinking either the PSU is failing slowly, or it's just not big enough to hold the system stable. I sense some might read this and suggest I'm stating the bloody obvious ( ) but I wanted to get some feedback before I go lay down 160 bucks on a new 850W PSU.
I've run the 4 RAM sticks through more than 24 hrs of Memtest in another system and they come out without errors.
I ran Prime95 on the troubled system all night last night and while the voltages showed the same fluctuations noted above, the CPU temp didn't top 65 C and there were no shut-downs? I don't know how great a strain Prime95 puts on the CPU and PSU versus a high end game, however. I then ran Prime95 on the PC next to the trouble system at the same time as well as the Window AC unit and found that the system overheated after about 30 minutes, seemingly suddenly, however, after holding at 65 C for a while.
If not the PSU, what would be the logical steps in trying to isolate the cause here? Would you try replacing the PSU before anything else? What's the most appropriate max CPU temp to set for cutoff in BIOS? I read different limits for this CPU from different users. I realize this got long-winded! Any advice is appreciated.
Just to confirm that after a week with a new 850W PSU in the system it hasn't topped 49 degrees on load; now idling at 32. Seems either the PSU was dying or couldn't manage the load and keep fans/voltage on par. Posting only in case someone has a similar issue in the future.
well, it was showing sudden shutdowns but only under load. I guess the CPU was holding up well enough to idle or perform lower demand tasks like browsing/watching movies, etc, but under the strain of a game like Battlefield 3 or MW3 it experienced the issue as you describe?