Help - determining damaged hardware?

[Sinistra03]

Thanks in advance for your time.

Built my own computer with some help/some new/some used parts - all good for 4 months. Had a power outage (was protected by surge - but the surge ability was ruined on the protector afterwards), that night started experiencing regular "freezes" where after so much time certain programs would start to permanently not respond. However, if I was running a voicechat program at the same time, that was still functional and operating for a while. After a while the whole computer would freeze up, and I would either manually restart, or - if I left my computer alone for long enough - it would blue screen with STOP error messages. Similarly, if I had no programs running and let it idle, it would usually react when I sat down but almost immediately programs would become non-responsive and blue screen soon enough. Now, the system is typically stable for 20-40 minutes then it has the same "slow freeze up" followed by a blue screen if I don't just restart. Same exact problem for almost 2 weeks now.

Steps I took:
1) Considering the randomness of the problem, thought I had bad RAM - memtest86'd old RAM (2x 4GB Ripjaws sticks) alone (1hr) and together (2 hrs) in various slots - no errors. Bought new RAM anyway - memtest86'd only new RAM together 60 min no errors.
Decided it wasn't RAM! Thought maybe RAM slot, CPU, MOBO or PSU damage.
2) Stress tested PSU with OCCT - blue screen after so much time. Stress tested CPU with Prime95 - blend = crashes, small FFTs = crashes (within 60 minutes - I would typically go afk) felt like I couldn't decide if it was PSU or CPU.
3) Eliminated overheating concerns - core temp's were normal, no dust build up or bad fans (5 of them).
5) Eliminated overclocking - reset BIOS to optimized defaults, then to factory defaults - same issues
4) Stress tested GPU with Furmark - 60 minutes no crash, just did again today though and blue screen.
6) Eliminated software - clean reinstall windows (twice) - same crashes
7) Visually inspected MOBO - didn't see any blown or leaking parts
8) Decided that considering the power outage, it was probably the PSU getting intermittent voltage or something. Ordered new PSU - same problems after installing today.

It doesn't matter how much stress the computer is under - IE I can run graphics intensive video games and browser windows for 20 minutes no problem then freeze, the crash just seems rather intermittent and random. I just feel like I can't narrow it down to the bad hardware. After installing a new PSU today I thought mechanically everything looks connected correctly.
Two weeks later at this point I'm at a loss - I feel like I've eliminated BIOS mistakes, software problems, virus problems, PSU, RAM - I either have a damaged CPU, damaged MOBO, or damaged graphics card. Any ideas on what to try that can narrow it down before I really lose my cool and order brand new everything?

Specs:
W7 Ultimate
SSD 100GB + old HD 1000GB
Intel Core i5 CPU 3.30GHZ
Gigabyte Z68X-UD3 Intel MOBO
2 x 4GB DDR3 RAM
GeForce GTX 580 GPU
850 Corsair PSU

 

[Harvey]

It sounds like you've tested everything but the motherboard, which doesn't show any visible signs of damage. Unfortunately, not all damage to IC's, resistors, capacitors, etc. are visible so you're down to the old maxim from Sherlock Holmes:

"When you have eliminated all which is impossible, then whatever remains, however improbable, must be the truth."

I'd say you're down to a damaged motherboard. It may not show any visible damage, but all it would take to make it thermally sensitive would be one or more slightly damaged internal semiconductor junctions.

It may be more expedient to replace the motherboard and test the current one again with other components, later.

Good luck. 🙂

 

[westom]

2) Stress tested PSU with OCCT - blue screen after so much time. Stress tested CPU with Prime95 - blend = crashes, small FFTs = crashes (within 60 minutes - I would typically go afk) felt like I couldn't decide if it was PSU or CPU.

What you have called stress tests really are not stress. The computer works only slightly more with more programs running. Stress tests are peformed by competely different methods. And not by overnight testing.

Best stress test is to operate a computer in a 100 degree F room. An ideal computer temperature. If any part is defective (intermittent), then diagnostic tests are more likely to see it. If any computer fails at 100 degree F (and then work normally again at 70 degrees), then the computer is defective and will probably fail more often after many months.

Now, a blue screen contain many numbers. Those numbers isolate the defect to a much smaller list of parts. Never ignore any number. Even if you don't know it, the fewer who actually know computers can then provide a useful answer. Your symptoms devoid of hard facts (numbers) means most every reply can only be wild speculation. Means the few who know this stuff are mostly silenced.

For example, you had as much reason to suspect bad RAM as you did a defective power switch, problematic wall receptacle, or bad memory location in the video controller. Until a hard fact or number (ie from the BSOD) point to a specific part, then all parts remains suspect.

Best solution means to not change any part; to not even disconnect one wire. Instead, collect all facts. One best fact are diagnostics that execute only when the system is in a 100 degree room. And that includes CPU temperature numbers, voltages measures by a digital voltmeter (not by the motherboard), and failure information stored in the system event logs.