Overclock.net banner
1 - 10 of 10 Posts

·
Registered
Joined
·
922 Posts
Discussion Starter · #1 · (Edited)
Was running a SOTR bench, when everything went black, system shut down and restarted, and the event log shows a 101 bugcheck.

I had my 12700K clocked pretty aggressively (p-cores 52-52-52-51-51-51-51-51, e-cores 40-40-40-40, 42 ring bus, +2 TVB). Previously I had been 52-51-51-50-50-50-50-50 on the p-cores, but I had about 20C to tjmax and thought that I could get a little closer.

But I didn't adjust core voltage higher (e.g. remove offset in V/f), just bumped the multipliers. The core voltages did adjust themselves higher, though, up to about 1.34V.

I rolled back to the earlier p-core settings, was able to get through the bench okay, and everything seems fine now.

Could the bugcheck just be due to the unstable overclock, and nothing more? Some googling online suggested that it could be indicative of a hardware failure.

Edit:

Here are the events thrown in the reliability history:

Rectangle Font Number Parallel Screenshot

Rectangle Font Parallel Number Screenshot
 

·
Registered
Joined
·
2,967 Posts
What PSU and graphics card are u running? Is there much else attached to the PSU, might have just drawn too much power, and the PSU tripped.

So far when OCing my 12700k, I've only had a few freezes, in P95 and once in CB23. The screen gets smeared, and werid, wrapped around. Each time win10 tried to dump the memory or whatever, but froze, so I rebooted.


I can't say what would happen if I ran something that taxes my GPU at the same time. I'll try some games.
 

·
Registered
Joined
·
922 Posts
Discussion Starter · #3 ·
After some more reflection after I posted, I think that that's exactly what happened.

The PSU is a Corsair HX1200, and the GPU is an EVGA 3080Ti FTW3 with hybrid cooling. The CPU overclock was stable for CPU-focused tests -- I scored 24,728 with R23, and also cut a second off of y-cruncher 2.5b down to about 68.8 seconds -- but when I ran SOTR and hit both the CPU and the GPU, that was too much. I had the GPU at a 112% overclock, which I had never had a problem with until I bumped the CPU cores. Everything went back to stable when I dropped the cores back to where they were, and all of the benches are running fine again.

So, for now, I think I'm okay. That just never happened before, and it was startling -- have had TDRs, blue screens, freezes and blackouts over the years, but never a complete power supply crowbar like that.
 

·
Registered
Joined
·
2,656 Posts
Yes, I do constantly. The GPU is stock, with no BIOS or hardware mods.

Is a 1200W supply not enough for that and a 12700K?
It depends on PSU design, really. Or whether or not the PSU is faulty.
If you're never touching on 1200W, you might want to contact Corsair and ask why you're blacking out even with less wattage being used.
Or RMA it while you can. Corsair has a send-first policy, so you can keep using yours for now.
 

·
Registered
Joined
·
201 Posts
I had a similar problem when I had my 3080 Ti FTW3 and Corsair HX1050.

In TimeSpy, Cyberpunk, and some other games, my system would shut down (but not restart, probably because I don't have it set to restart after power failure). The only way I was able to complete benchmarks/stress tests and play games was to undervolt my GPU. I did .900 @ 1920 MHZ and didn't have a problem. This also lowered GPU wattage considerably at load.

I had a hunch it was my PSU. Ended up getting a Seasonic TX-1000 (only decent power supply that wasn't marked up at the time). No problems since.

However, your power supply is newer, so I'm not sure you're running into the same issue. Also I don't see any posts related to the HX1200 running into OCP issues.

If you're willing to eat the cost of an RMA shipping label, at least the PSU is easy to swap out since you don't need to re-run the cables. May be worth a shot to rule out the PSU.
 
  • Rep+
Reactions: stahlhart

·
Premium Member
Joined
·
2,663 Posts
A good test for the PSU I found it OCCT they have a test called POWER which will load both cpu and gpu at the same time. You could probably do the same with something like p95/y cruncher and furmark etc at the same time to load cpu and gpu.

I have a couple Seasonic PSU Titanium 850/1000w that will shutdown even running CPU and GPU at stock when I try OCCT Power test. They can work fine for months but sometimes when playing a game I would have random shut downs and finally figured out it was the PSU and confirmed it using OCCT power test.

Also try running it and then stop and running again as it the intial "shock/spike" that seems to cause the PSU to shutdown.

Edit CPUs and GPUs i have used to test range from 11900k to 12900k and 2080 ti, 3070 and 3090 and all will shut down the "defective" PSUs.

 

·
Registered
Joined
·
1,757 Posts
I seriously doubt its your PSU. An HX1200 is more than enough for a 3080 ti and a 12700k. Even if both were clocked to the moon. You more than likely didn't give the cpu enough vcore for the OC you requested. I run mine at 51 on all p cores and 40 on all ecores at 1.36. My 6900xt is also clocked fairly high and pulls upwards of 500w. I'm using an RMx 1000w which is gold rated. Your HX is platinum rated.
 
  • Rep+
Reactions: stahlhart
1 - 10 of 10 Posts
Top