BSOD's with Prime95 ver. 27.4

[Chike]

Re: BSOD's with Prime95 ver. 27.4

Try to run the FFT size that caused the BSOD, you might not need to wait 5 hours for it.
It's not the time it was running it's pecific pattern that cause it.

[flyfisher]

Re: BSOD's with Prime95 ver. 27.4

Good idea, you can see what FFT it was on when it BSODed (is that a verb?) in the results.txt file.

[Chike]

Re: BSOD's with Prime95 ver. 27.4

in case of BSOD it would be diffcult to see what was the FFT size but you can see what was last successful FFT run and assume that the one that comes after is the one that cause it.
I am not sure if all computers run the same sequence but you can set time to run each test to one and run at default settings for an hour or two and you will have the list of FFT size sequnce you will ever run. By default Prime95 run each test 15 minutes that is 4 sizes per hour.
The successful runs are listed in results.txt in Prime95 directory.

[wevsspot63]

Re: BSOD's with Prime95 ver. 27.4

Yep, did all of that but last successful run prior to failures were different. I was a little too over confident because I didn't even do short runs of 1344 and 1792ftts..... in the beginning. I incorrectly assumed that hitting 4.5Ghz was going to be a walk in the park. After a few hours/then days of frustration I settled in and started doing it "right". Anyway, long story short - if I could pass 15 minute Prime95 runs on the 1344fft and 1792fft I was good to go for long term Prime95 stability test.

[Chike]

Re: BSOD's with Prime95 ver. 27.4

Those large FFts so maybe higher VCCIO and possibly higher VCCSA would solve it.

[wevsspot63]

Re: BSOD's with Prime95 ver. 27.4

It did :)

[Lsdmeasap]

Re: BSOD's with Prime95 ver. 27.4

Try to run the FFT size that caused the BSOD, you might not need to wait 5 hours for it.
It's not the time it was running it's pecific pattern that cause it.

Ya, but in the end, after many hours of sessions you spend making it stable for xx FFT, then you have spent many hours on it anyway. Plus that last final run you think will be stable, you know you'll want to let it run for 8+ hours to be sure you are still passing the xx FFT trouble size.

So it seems like a huge hassle still. I run 6-8 instances (Depending on core count) of HCI memtest using all ram split between then, and that will easily crash out settings that pass LinX, so I bet once you get HCI stable you're probably Prime95 stable as well.

[Chike]

Re: BSOD's with Prime95 ver. 27.4

Prime95 run may instances automatically and also et you choose number of threads manually if you with so I doubt it would be more hassle than HCI memset.
When you run known failing FFTs sizes they useually fail after a minute or two.
You may as well pass HCI memtest and fail some Prime95 FFT so there's no realy way to tell.

[wevsspot63]

Re: BSOD's with Prime95 ver. 27.4

To each our own :)

Or maybe to each both of them :)

[Chike]

Re: BSOD's with Prime95 ver. 27.4

The more the merrier, and still after all the testing you cannot be sure it's 100% stable, but it does not really matter, as long as you are happy with it and don't get BSODs too often

[Lsdmeasap]

Re: BSOD's with Prime95 ver. 27.4

Prime95 run may instances automatically and also et you choose number of threads manually if you with so I doubt it would be more hassle than HCI memset.
When you run known failing FFTs sizes they useually fail after a minute or two.
You may as well pass HCI memtest and fail some Prime95 FFT so there's no realy way to tell.

Ya, I know what you mean about Prime95 settings. But I didn't know when you find the failing FFT size, then it will always fail in a minute or two, so I guess knowing that then testing new settings against that FFT size might not take as long as I thought.

But still, in the end, once you have everything where you think it's stable, don't you still have to test for 8+ hours? I can do that, and used to do it that way too, just seems like such a long time now after using other shorter testing methods.

I agree about passing HCI and failing prime, same would be true about passing prime and failing HCI too So, ya, I guess we all have our own methods and they work for us.

You're right, stable is when you don't have BSOD's doing whatever you do with your computer, as long as you pass whatever method you like and don't get BSOD's often you're good to go