GPU keeps crashing on CPU load

[antonrr90]

~System Information~

MOBO: ASUS P8Z77V-Pro

CPU: 3770K

PSU: EVGA 1000G1

GPU: Gigabyte GTX 1070

RAM: Corsair Vengence 4x8GB (32GB total)

OS: Windows 10 Pro (Pre Creators Update)

 

If I am rendering a scene in Blender3D(using GPU) my screens flash a bit like they are refreshing, then the application freezes, leaving Windows log with GPU TDR error, yet I have TdrLimit sec to 10 seconds. When I am using IntelBurnTest (High), my temps are staying great, but my GPU keeps crashing with the same Tdr error I was getting during a GPU render cycle. So I am at a loss on which component is actually causing the problem. The GPU is is less then a year old, and the rest of the PC is just over 3 years old, never had any heating problems on anything. Anyone have any clues of what to check first, because I have less then a $100 to even put into debugging/parts. I Initially thought is was the CPU, which I had OC'd to 4.5Ghz but my temps even under load never went above 72C, and even then it was never there very long. It just seems like whenever the CPU is taking on a load, the GPU gets starved out, but the error it is throwing is only for communication sequencing, and I am not even getting flagged with a problem by Windows like I was during my use with Blender3D. So, I am at a loss on this one. I am giving it a cleaning today, and re-seating a few things, but hope someone will have an idea if it continues, because it is getting worse everyday.

[Enderman]

Have you tried turning all overclocks off?

[Freezoide]

Exactly what Enderman said.
Not sure how thoroughly you tested the stability of the overclock but even when several tests run stable, another benchmark or application may come into conflict with the OC.

The same way I discovered that my CPU clock was too much when using Handbreak for transcoding a vid.

[antonrr90]

The CPU testing I did with IntelBurnTest ran for 12 hours, no errors on the CPU side, but the GPU did crash 14 times according to the Windows Error Log, but no warnings were ever displayed, all 3 of my screens just kept going off and on, and even GPU-Z didn't show those crashes either, as it's results were clean and steady. I am starting to wonder if Windows 10 is causing this mess. I heard a couple other people have 3 display problems with Windows 10, but it only started after the previous Windows 10 update, (Not the creator update, I am still in defer mode on that one) I just finished vacuuming out the system, it needed it, and re-seated the GPU and all the power connections from the PSU, just getting ready to run a few test now. I will drop the OC back to default, and see if it cleans it up a bit. That OC was 100% stable for 4 years with temps below 72C, I hope that isn't the problem, as I do a lot of rendering, and I was getting about 15-20% improvement with the OC. Thanks for the responses, let me know if you have any other thoughts, I will try to update after I do a few more tests this evening. This is my Primary system, so I am in kind of a pinch.

 

[antonrr90]

FREAKING WINDOWS 10 - Last update deleted my Power Options, so instead of being on High Performance it was at balanced, which was cutting power to my graphics card on demand.

 

So thanks for the responses guys, this crap is all on Microsoft and their highly NEGLIGENT update practices.