I built a new computer this weekend, and it was going great for a while. Installed Windows 7 64-bit, installed some apps. Ran prime95 for about an hour, system stayed cool. A couple hours later it just started dying with no warning. It can happen in the BIOS setup screen, so I don't think it is a software issue. System now usually only stays up for something less than five minutes.
Bought some components at Newegg, some at the local Fry's. Everything is brand-new, except the CPU which was a customer return at Fry's, and a USB wireless fob. Not doing any overclocking.
Sapphire Vapor-X 5770 graphics
G.Skill DDR3 PC3-12800 ram 2x2GB. (running at 1333MHz, not the rated 1600MHz)
Cooler Master Hyper 212 Plus
Windows 7 Ultimate 64-bit
WD 1GB Caviar Black SATA
Samsung SH-S243 SATA DVD burner
I noticed a couple of anomalies during the multiple crashes. CPU Hardware monitor and SpeedFan sometimes show the +5V rail as 5.8V and the +12V rail as 10.03V. Could that be causing the crashes?
I noticed during one reboot that there was a bad CMOS checksum so the BIOS settings had been reset.
The wireless (Netgear WPN111) sometimes has connection problems, but I suspect that is not the cause since the crashes still occur with it removed.
The system was running fine for a few hours, so I don't think it is an obviously wrong connection somewhere. WhoCrashed reports no crashes, and they suggest it could be either malfunctioning hardware, power failure or a thermal issue. Well, the system is not hot and the power supply should be well within load limits.
My natural inclination is to blame the Fry's previously returned CPU. Does anybody have any other ideas? I have about a week before I can't return the cpu/mobo combo (the combo were on sale together, but they claimed they didn't have any factory sealed cpu's). Is there some unusual interaction between components?
If I can figure out how to attach a screen shot of the system status, I'll do it.
Unfortunately, I don' t have another modern psu. I started dis-assembling the computer last night, and the shutdown problem continued with only the case, mobo and cpu powered up. I'd have to add the system has deteriorated to the point where it doesn't even like to power up reliably. I did the paper clip test of the power supply with fans attached and that is fine, although that doesn't really say too much.
I haven't been able to figure out how to attach a picture, but the cpu core temps varied from a max of 26-33C, gpu max of 35C, cpu power maxed at 110W. The highest "other" temperature was 29C. The voltages all seemed rock solid (as reported by CPUID Hardware Monitor, min V = max V for everything but CPU VCORE, and VIN2), with the exception of +5V and +12V seeming being out of spec. I did have the system shutdown once when I was in the BIOS screen and saw those voltages reported as being right on 5V and 12V, so those voltages are not causing an immediate problem. Could they damage components in the longer run though?? I wouldn't consider a few hours the long run though.