So... Let me tell you the story so far....
Been a month that my system with a Pure Black 990FX MB was giving me BSOD once every 2 days, then started to do it everyday. Decided I had enough, took the AsRock 990FX Extreme 9 out of the box (finally you could say, since it was idling in that box for months), and do a fresh install-of-all-the-things.
Told myself "that shoud fix it!"
....nope
Before scrapping the old system I went with some testing with MemTest 86+
version 5.1 gave me errors, I mean, LOTS of errors, on the order of 400+ errors, but only on test #7. Multithread wasn't a thing, so it ran always on CPU #0.
Attached are some screenshots that I made with my phone's camera when I stopped the tests during the first tries believing that the ram was the issue. Then I read somewhere that test #7's errors could be related to bandwith issues between the MB and the DIMMS. So I ran more tests just to get more errors, but only on that test, #7
I ran the tests this way: tested every single DIMM alone, then all togheter. Only test #7 gave me issues. Test #10 blocked my system entirely.
Read somewhere that multithread support wasn't a thing on the 5.1 v of the program and that test #7 had some issues too, so I said "don't care, let's just go for the new MB at this point and test things there". So I did.
After installing and updating every bit of windows and drivers, the pc hangs up again.. it does that usually after hours of work without issues. Randomly. And without saying a thing, like, leaving me completely clueless. It just freezes and that's it.
So, more tests.
Downloaded the 7.1v of MemTest 86+, and starting to run tests.
First issues: the program doesn't like this MB too, because it only runs on CPU #0. Tried to run tests with Parallel, no way; with R.Robin, no way; sequential goes, but it doesn't switch CPU, it stays on #0, so.... pretty much that.
And guess what, test #7. Again. Not so many issues as with the precedent MB and the precedent version of MemTest 86+, but always different addresses, completely random. I've ran multiple passes and multiple tests and every pass of every test had some address fail here and there, random fails.
And test 13#. But only one time. Then nevermore.
All other tests went through without a hiccup.
Thing is: I've run BurnIn on Torture mode for the ram, 10 processes at 10% each, and guess what, the test passed, but the system froze 20 mins later; I've run the CPU on torture mode at 100% for 4 hours and trillions of operations and not a single error. I've run a Torture test with the GPU, the Ram and the CPU at 100%, and nothing.
But it seems to me that if I leave the system alone, after 20 mins it freezes.
I don't know what to think anymore. I thought it could be the PSU, it should crash my system under heavy load but it doesn't.
I thought it could have been my CPU, but after stress testing it for 4 hours and no errors....meh.
Thought it is the 32 GB of RAM, after all the errors with this test 7# in MemTest 86+, but....doesn't look like from BurnIn.
Anyway, since I can't attach logs, here we got with copy-paste.
Been a month that my system with a Pure Black 990FX MB was giving me BSOD once every 2 days, then started to do it everyday. Decided I had enough, took the AsRock 990FX Extreme 9 out of the box (finally you could say, since it was idling in that box for months), and do a fresh install-of-all-the-things.
Told myself "that shoud fix it!"
....nope
Before scrapping the old system I went with some testing with MemTest 86+
version 5.1 gave me errors, I mean, LOTS of errors, on the order of 400+ errors, but only on test #7. Multithread wasn't a thing, so it ran always on CPU #0.
Attached are some screenshots that I made with my phone's camera when I stopped the tests during the first tries believing that the ram was the issue. Then I read somewhere that test #7's errors could be related to bandwith issues between the MB and the DIMMS. So I ran more tests just to get more errors, but only on that test, #7
I ran the tests this way: tested every single DIMM alone, then all togheter. Only test #7 gave me issues. Test #10 blocked my system entirely.
Read somewhere that multithread support wasn't a thing on the 5.1 v of the program and that test #7 had some issues too, so I said "don't care, let's just go for the new MB at this point and test things there". So I did.
After installing and updating every bit of windows and drivers, the pc hangs up again.. it does that usually after hours of work without issues. Randomly. And without saying a thing, like, leaving me completely clueless. It just freezes and that's it.
So, more tests.
Downloaded the 7.1v of MemTest 86+, and starting to run tests.
First issues: the program doesn't like this MB too, because it only runs on CPU #0. Tried to run tests with Parallel, no way; with R.Robin, no way; sequential goes, but it doesn't switch CPU, it stays on #0, so.... pretty much that.
And guess what, test #7. Again. Not so many issues as with the precedent MB and the precedent version of MemTest 86+, but always different addresses, completely random. I've ran multiple passes and multiple tests and every pass of every test had some address fail here and there, random fails.
And test 13#. But only one time. Then nevermore.
All other tests went through without a hiccup.
Thing is: I've run BurnIn on Torture mode for the ram, 10 processes at 10% each, and guess what, the test passed, but the system froze 20 mins later; I've run the CPU on torture mode at 100% for 4 hours and trillions of operations and not a single error. I've run a Torture test with the GPU, the Ram and the CPU at 100%, and nothing.
But it seems to me that if I leave the system alone, after 20 mins it freezes.
I don't know what to think anymore. I thought it could be the PSU, it should crash my system under heavy load but it doesn't.
I thought it could have been my CPU, but after stress testing it for 4 hours and no errors....meh.
Thought it is the 32 GB of RAM, after all the errors with this test 7# in MemTest 86+, but....doesn't look like from BurnIn.
Anyway, since I can't attach logs, here we got with copy-paste.
Comment