28 thread Haswell Xeon issue

[10esseetony]

Some of you remember I posted somewhere a few thousand pages ago/a few days ago, that I had a failure of my AIO liquid cooler, and had temporarily allowed gravity to hold an AMD Wraith cooler in place until I got a replacement. The temporary solution, and the $20 delivered permanent solution both brought the CPU down from low 90's C to mid 60's C, which is just perfect for this chip.

The board is X99, the chip is Xeon-E5-2683?v3....low end 14 core Haswell. This system has been renamed many times, but it's original name came from Rudy Toody, WUSS, short for Work Unit Sucking/Stealing System. You can always count of Fred for a good laugh.

Since the cooler replacement, I have updated to Mint 20.*, and added a 2nd GTX1070. The CMOS battery has been dead for years, I finally replaced that.

Under Mint 20, it would run, then pause, no response from the mouse/keyboard, for minutes at a time. Psensor eventually would update and show that indeed, no system activity during this 'pause'. Hundreds of tasks were corrupted, and the system was blacklisted, eventually. So, instead of trusting the update from Mint 19.3 to 20, I did a fresh install. It ran great for hours, then did the same thing.

Currently I have aborted all tasks and am trying Win10.

At first, QuChem seemed to be the problem. Switching to NumberFields helped, for many many hours, then the problem came back. I fear the PSU may be faulty, if I flip the power switch on the unit, it will NOT restart for several minutes. I love my 1200w Corsair Gold unit....

I have JUST started Win10 crunching, and will update if it happens again. I would rather have a faulty MB then to lose that PSU....That PSU has some history to it.

[crashtech]

Looks like AT frowns on hot linking! But interestingly enough, I have a system that would spontaneously reboot once in a while only when running Quchem. It turned out to be a low hours Seasonic Gold PSU that was the problem. At least try some other known good PSU, I'd think. Looks like that Corsair has had an illustrious life driving some fairly challenging loads.

[10esseetony]

So far the system is 100% stable under Windows.

I do have a lingering problem though, same as Howdy mentioned in another thread (probably the PSU). If I power down the system, I have to let it sit for 10 minutes or so (PSU switched off), before it will power back up.

I suspect (just a hunch), that it has something to do with the PSU signal that tells the MB it is ready. +5vsb perhaps? I have 3 cheap PSU testers, and one of them has an LCD display that tests for that, whatever it is called. The WORST I have seen is .7 seconds from the time you flip the power switch on the PSU, until it is 'ready' to power the computer. Most are less than .5 seconds. Naturally I can't find that particular tester right now, nor the 3rd one. I did find 1 out of 3 though!

I sure hope my beloved Cosair 1200w Gold unit isn't on it's last leg.