I've been fighting system instability for a couple of months now and have tried a lot of different things, so I'll try to be as specific as possible without going into unnecessary details. If I've left something out, I apologize. I finished this build in March after several months of off-and-on tinkering. I purchased an EVGA SuperNOVA 1200P2 since it was what EVGA's Power Meter tool recommended (not very scientific, I know, but whatever... it seemed like enough).
My system is as follows:
- 64GB (8x8) DDR4-3000
- 2x 980ti flashed to MaxAir BIOS (1.25v at load; 350W at 100%; 425W at 121%)
- Rampage V Extreme
- 512GB 950 Pro PCIEx4 SSD
- 2x 1TB 980 EVO SATA SSD
- 2x 3TB WD Red 5400rpm HDD
- Asus Essence STX ii Soundcard
- 2x D5 Pumps
- 11x Fans (connected to/controlled by an Aquaero 6)
- EVGA SuperNOVA 1200P2
The first signs of weirdness came when I was filling/bleeding the loop and the power supply would occasionally turn off while jumpered. I didn't think much of it at the time since I was moving the case around a lot to get rid of air bubbles and figured I must have jostled the plug enough to pull it loose.
Once the bleeding/filling process was finished, I struggled mightily to get the system fully stable. I never had any issues getting it to power up, but I couldn't ever escape reliability issues. I could get to a "stable" clock that would run Aida64's stress test overnight without fault, only to have it randomly restart (no blue screen or crash message) while watching Netflix, for example, three days later. The next day it would restart while playing a game, then yet again overnight. Sometimes I could narrow a crash down to a particular process from looking through Windows' reliability history, but most of the time the only thing it showed me was the generic "Windows restarted unexpectedly."
In the past couple of weeks I've noticed the motherboard "dropping" memory. There are no errors, it just will only detect 4, 6, or 7 of the 8 DIMMS I have installed. I spent the better part of the past week tinkering with voltages and settings to try to get all 8 working, but there doesn't seem to be much rhyme or reason to it. All the sticks work individually and in any combination up to 6, but once I go from 6 to 8, sometimes it detects all 8; other times it only detects 6 or 7. The strangest part (to me, anyway) is that the DIMM(s) that it drops can change from one boot to the next. Right now it may not be detecting B1 and B2; I'll reboot and it won't detect B2 and D2; I'll reboot again and it will pick up all 8.
This weekend I finally decided that I needed to suck it up, drain the system, and pull out the CPU to check for bent pens/socket damage. Everything was fine; there was no damage. During the fill/bleed process after re-seating the CPU and putting the loop back together, the PSU shut off a couple of times, just as it had done in March when I filled it the first time. It struck me as more odd this time since I wasn't moving the case around most of the times when it shut down and I'm just more cognizant in general of crashes/power issues with the system. Sometimes I could turn it right back on without any trouble; other times it would either turn right back off or not do anything at all. In those cases it eventually came back on, but only after I let it sit powered down for a few minutes.
After an afternoon of leak testing (and fighting with my PSU), I removed the jumper on the 24-pin connector, plugged it into the mobo, and powered up the system (with everything at stock speeds/voltages). Everything seemed fine at first glance: all the memory was detected, temperatures were notably lower (I screwed up the TIM spread when I first built the system, which gave me about a 10C lateral ΔT across the chip), and it passed a quick 15min Aida64 stress test without any problems. Then it restarted. I got back into Windows, opened up the reliability history, and it just showed "Windows did not properly shut down." 5 minutes later, it restarted again. Got back into Windows, opened Chrome, and it did it again. At that point, I just turned everything off, went to bed, and resolved to find out more today.TLDR - Could all my issues come down to a faulty power supply? I plugged my components into EVGA's PowerMeter this morning and it suggests either a 1050 or 1200W PSU. However when I plugged all my info into Outer Vision's power supply calculator, it suggests that I need 1548W. The fact that it won't stay on when jumpered tells me there's something not right with it regardless, but is a 1200W Platinum unit simply not strong enough to do what I'm asking it to do?