I am at my absolute wits end with this. I have built many computers but never run into anything so persistent or destructive to my sanity:
Almost exactly 1 year ago I built a computer, here is the original configuration:
5-7year old case (exact name/manufacturer unknown, is gone now)
Apevia ATX-AS680W power supply (680w) - carried over from old build
Gigabyte GA-990FXA-UD3 motherboard
AMD FX-8350 Black Edition Vishera 8-Core 4.0 GHz
G.SKILL Ares F3-1866C10D-16GAB (2x 8gb ddr3 1866)
GIGABYTE GeForce GTX 970 4GB WINDFORCE gfx card
WD D50 01AALS-00L3B2 (500gb standard HDD) - also carried over from old build
Running windows 7 ultimate 64bit
Computer is built, runs fine, no issues for (if I recall) at least the first few months. One day it shuts down randomly. No power down, no BSOD, no power surge, no extreme load, no extreme heat, no freeze, screen artifacts, nothing that would give a clue, it simply instantaneously (almost faster than pulling the plug) ceases to be on. I search for error logs, check wiring, check PSU, seating of cards, etc, install CPUID HWMonitor to watch heat, etc. Issue does not repeat itself for days, then happens randomly again. Over the course of a couple of months it recurs constantly, but never for a "reason", there's no power issues, no loose wires, not based on heat, not based on load, not based on specific software or anything I can find.
So assuming it's a PSU issue, I replace the PSU, it now has a Corsair GS700 700w power supply.
Issue does not recur for a while, seems fixed. Few weeks later, happens again. Sometimes it happens once in a week, sometimes it happens 16 times in a night and is barely functional at best. Of note, it often would require shutting off the PSU switch or unplugging it before it could be rebooted (assuming I needed to discharge capacitors). Another note is that it would sometimes require to repeat this process many times before it would power on (repeatedly it on pressing power button it would power on for <1second before instantly shutting off.)
Additional testing is done:
* memtest 86 is ran for ~8 hours on each stick of ram individually. They do not fail at all.
* Graphics card is swapped for a much older but still compatible ATI Radeon 6870. Thing sounds like a wind tunnel and is like 5 years old, but it works. Failure continues to happen occasionally.
* Motherboard is RMA'd - unable to find a reason for failure that makes any sense at this point, I RMA the motherboard. It takes Gigabyte a rather long time (3 weeks if I recall?) to process and send back, but they do so in the end, sending me... the exact same motherboard with no information. I call them and find out that they had sent it back saying "no failure."
* The hard drive is replaced with a 480gb Sandisk SDSSDHII-480G-G25 SSD. Windows is reinstalled for probably the 4th time. After having a brand new issue with PCI-e power save setting messing with my hard drive - which gets solved, the old roblem persists.
This has been going on now for nearly a year, though occasionally I'll have several weeks of daily use without failure. One day while pulling my hair out and wondering what gods I have offended to deserve this, I notice that just *barely* touching the motherboard power connector is enough to trigger the failure. Eureka! I've found it. I'm not going through that RMA again, so I decide I'm going to buy a new motherboard. Having noticed high (but not within auto-shutoff range) heat, I also get a new CPU cooler - a cooler master Hyper 212 EVO, and my ratty old case is replaced by a Corsair CC-9011038-WW (I thought maybe faulty wiring or something?).
New components arrived last week, were assembled together Monday morning 2/15/16 - with a surprisingly similar motherboard, and the installation of some new drivers for it, I am able to avoid having to re-install windows. Computer works great now, temps are down a solid 20 degrees with better airflow, better wiring, and a better cpu cooler, but are still monitored. Computer is gamed on all day for ~8 hours. No problem. Deliberate stressing is done (within just daily usage stuff) by just running lots of programs and multitasking, etc. Flawless.
Then I have a new issue. Suddenly, after being fine all day while playing high-end games, the screen freezes, there's a screeching in the headphones for a moment, and everything is non-responsive. After a few moments the computer automatically restarts itself. It freezes again whenever anything is done in windows. This occurs again and again, though I am eventually able to look at windows error logs in safe mode, nothing ever stands out. Pretty much everything but in tears at this point, I straight re-install windows again, but end up with the same freezing problem after (still no shutdown issue, that's totally solved now, right? RIGHT??!) After swapping graphics card out thinking that might be the problem again (I had originally thought the main issue was gfx card related) this current new freezing persists. The problem ends up being solved by replacing the ram with some older ddr3 1600 - Kingston KHX1600C9D3K2/8GX which is 2x 8gb sticks. The issue is instantly gone and never recurs. (Well, hasn't recurred in 2 days.)
So having replaced PSU, case, motherboard, hdd, and ram (and potentially eliminated the gfx card as the issue by getting it to happen with it not in the computer)... I seem to have a functional computer finally (yes!) ... until today. Playing a game.... and it shuts off again. No warning, no error code, no freeze, hang, BSOD, no warning, no heat issue, etc... just gone. I cannot describe how crestfallen I am at this point. I have done literally everything I can think of to fix this. I am quite poor, this computer has drained every penny I can scrape together just to keep running and ended up costing me 1.5x its initial cost by this point... literally the only original components left in this computer are the processor and the graphics card, and I do not have the money to replace either, especially without knowing which is at fault.
Does ANYONE have any ideas? Any thoughts? Is there anyone who can help me end this living hell? I cannot seem to find any sort of resource that has been able to tell me how I could test for this, how I could diagnose what this is, and every single thing that makes sense has already been replaced.