R7 1700x CPU

>R7 1700x CPU
>3.8GHz at 1.35v
>rendering video, CPU utilization ~80% and CPU temps ~70c due to rather warm ambient temps in low-mid 20s
>at any point during the render, lose video signal to monitor
>have to reset PC to get signal back, lose my rendering progress
What the hell? What am I doing wrong? Do I need to increase the voltage some more? Using a Hyper 212 LED cooler by the way.
Pic related is what I've got it set to right now; 3.7GHz @ 1.35v. This is the first time I've overclocked anything before, if that wasn't already obvious.

Attached: 3.7ghz 2.png (633x794, 44K)

>Poozen
Found the issue

buy zen2

You have to run a prime95 stress test with small FFTs and AVX enabled for at least an hour to verify a CPU overclock and/or undervolt you mong.

I ran Intel Burn Test and it passed

According to XFR specs you are in fact undervolting because 3.8GHz is max stock single threaded performance with 1.45v as long as temps and power delivery is stable (ie can't do this on all cores on a full load).

How long? Also if it has "intel" in the title it doesn't sound like it would be 100% reliable for AMD CPUs.

You're right, apparently the version of IBT isn't very reliable for AMD CPUs. I'd start prime95 right now but I'm recording some audio in my line in right now and I don't want it to crash my system while that's going. I'll restart, raise it back to 3.8 and then start prime95 in 30-40 minutes.

>HWMonitor
don't use that. It is not reliable. Use HWiNFO64 instead.

Gotcha

Attached: hwinfo64.png (1133x628, 90K)

>ryzen CPU
>loses signal out from GPU
>blames the CPU

Hmm... what do?

Your overclock is unstable son. More voltage or less clock. Make sure your memory is stable too.

Revert to stock? Or tune the OC properly. Motherboard may not be able to push out voltage to all the parts.

That can happen, when undervolt fails the OS may not bsod at all and just freeze or cut video signal like in OPs case.

It's an UV, he's going -100mV on all cores at 3.8GHz (max stock single thread).

Your oc ain't stable. Use prime95 3+ ours, asus realbench stresstest 2+ ours, and blender benchmarks.

>small FFTs
Never use small FFT for stability testing, it just hammers the cpu with the same instruction, it's shit at detecting instabilities and needlessly hard on the cpu life span. It's only useful to test extreme thermal scenarios.
Use custom blend with 80% or more ram, ftt from 1024 to 1344, avx depending if you have an offset and what you want to test.

>and CPU temps ~70c
it's actually 50C, install HWinfo retard

Maybe the problem is your PSU...

Possible. It's a Corsair TX650W, bought used.

Sounds like a gpu problem.

70c isn't even hot. Stop being a pussy. Crank it up to 1.4v.

>Stop being a pussy
Again, new to this. What should I do regarding the GPU?

drop the volts down to 1.26v and keep it near or bellow 62°C