So, I'm using prime95 to test if the cores are stable, but now that I restarted the program itself doesn't work lol.

Prime is too heavy of a stresser and a lot of versions have issues with the vishera architecture.

Use aida 64

if your system can't handle any potential workload regardless of intensity without crapping out or overheating it's not truly stable. That's a fact of life.

I wonder why people disowned p95 so hard, I guess because it's easier to consider it bullshit than acknowledge how hard they're being jewed. It all began with Ivy Bridge and its TIM 'solution' came out and people were saying how p95 could damage it and blaming it instead of the real culprit which was the corner cutting on intel's part. It got so bad that after Haswell came out they had to release a dumbed down version of p95 that didn't push it as hard as it could. Now the issues are even worse with kaby lake maybe they'll dumb it down again to cater to people that keep giving corner cutting hardware vendors their money for chips that can't even adequately dissipate their heat.

For god's sake they even got rid of the little copper slug on the locked chips' coolers with the release of Skylake. Now it's just a piece of aluminum junk with a fan strapped to it. But it's all p95's fault I guess.

This post was edited by DCSS on May 20 2017 04:39pm

So, I just tried to overclock my fx 8350 with fan cooling and it was about to hit 80C so I had to shut it off at 78C

Time to get a Evo 212... can't game on undervolted fx 8350 1.325

- set to 4.1ghz, getting to 75C

had it originally underclocked at 1.325 @ 3.8GHZ and it hit 62C on PUBG

testing with PUBG on~now

This post was edited by d2jspNA on May 20 2017 04:39pm

If you spend a few minutes googling vishera and prime 95 you'll see it's a common problem. That's a fact of life.

Aida, OCCT and other less intensive stressers are perfectly fine for simulating maximum workload.



62C is your max core, 70C is your max socket.

Again, the issue seems to be with the chip or architectures shortcomings to not be able to perform small FFT's without crapping the bed. I wouldn't blame p95 just because bulldozer is a housefire poo in the loo arch.

You can't define a maximum workload as anything less than crunching very efficient factoring algorithms on every single thread simultaneously, if you do, you concede the right to use the word 'maximum'. Prime95 isn't forcing the cpu to operate out of spec or crash, it's simply asking it to do math, the thing it is designed to do. A lot of very difficult math very fast, sure, but the fault is with the chip if it fails that task.

Basically every real world scenario that maxes out the cpu to 100% is going to give the transistors valuable microseconds to dissipate heat and such as the cpu waits on i/o requests and instructions, so it's going to be much less taxing than something like prime, but how do you define what a maximum workload is besides worst case utilization?

This post was edited by DCSS on May 20 2017 04:58pm

Let me put it this way

What task would call for a 24 hour prime stable that a 24 hour aida would not be sufficient for?

You can go up to 4.9 with a hyper 212

If you're lucky.

indeed, the 9590 (same chip) pulls like 220w at 4.7ghz, gl cooling that with air lol.