Shutdown During High CPU Usage (CPU or Motherboard?)

[roy5000x2]

Okay, this has been a fun one. My computer is crashing during certain higher/prolonged CPU usage sessions. So far I've seen full shutdown during:

• Assassin's Creed Odyssey (15-20 min in)
• Deus Ex: Mankind Divided (1 hour + in)
• OCCT:Linpack stress test (5-10 min in) *this one exhibits odd behavior. Around 5 or 10 minutes into the test, CPU temperatures will completely spike and go from a steady ~upper 60s to touching 90C before either the test stops itself from overheating or a shutdown occurs.
• Intel Processor Diagnostic Tool (during CPU Load test)

 

Hardware/Software most items are 5 years old now unless called out:

 

Spoiler

 

• Windows 10 64 bit
• Intel 4790k running at stock clock
• Asus Maximus Gene VII running BIOS 3003 (I had updated to 3053, though that's also when I first experience shutdown issues and rolled back. Roll back didn't help unfortunately)
• Corsair H100i
• EVGA GTX 1080 Ti (was overclocked, have since brought it back to stock clock speeds)(around 1 year old)
  • 436.02
• EVGA Supernova G2 850W (brand new)
  • Replacing a Corsair HX850 that was failing the OCCT PSU test
• 16GB DDR3 Corsair Dominator Platinum 1866
• Corsair 350D Case

 

 

Things I've checked so far

Spoiler

 

Thermals: After trying out several different pieces of monitoring software individually (HWinfo, RealTemp, CorsairLink, AI Suite), CPU usually sits in the upper 60s, jumping up to the low 70s during gaming / stress tests (note about OCCT:Linpack above with the exception). H100i Pump will get into the upper 40s. GPU sits at a comfy ~70 under synthetic load.

Event Viewer Logs: Just your regular Event ID 41 stating an improper shutdown occurred. Some DCOM errors have been floating around as well, though a lot of what I've read up on doesn't point to them as being an issue.

Memory Check: RAM passed Window's memory test (though I haven't run MemTest86 yet)

Replacing the PSU: My Corsair HX850 didn't pass OCCT's PSU test, so I swapped it out with a brand new EVGA Supernova G2 850W. While I can now pass said test, does not help prevent shutdown.

BIOS rollback: Just in case, I rolled back from ASUS's 3503 to 3003 BIOS, as I started experiencing the shutdowns around the time I flashed the newer BIOS. No change in behavior however.

Other Stress Tests

• CPU: I can run FurMark + OCCT:CPU / AIDA64 for well over an hour with no issues (haven't tried Prime95 since that was notorious for trying to grab more voltage that it should with Haswell). Other much less stressful games (Borderlands 2, CS:S) don't pose any issues either.
• GPU: FurMark for an hour + no problems, OCCT:GPU doesn't pose any challenges. Heaven keeps on going no problems

 

 

 

Other oddities I've noticed recently

Spoiler

 

First was my motherboard having trouble booting and throwing the A2 error code. Sometimes unplugging my keyboard during boot would resolve, though eventually it refused to move past the code. Resolved by doing a BIOS flashback to 3503.

Nvidia driver / crashing issues: Some games (Everspace, DX:MD) will sometimes crash and throw memory allocation errors or the D3D device removed error. This seems don't seem to be too uncommon for people.

Computer also shutdown when attempting to install the latest version of Nvidia's drivers (wiped out all of my old drivers and install the new ones fresh to resolve).

 

 

 

I'm running out of ideas. Really starting to lean towards either the CPU or Motherboard dying, though other insight could be nice. Might just be time to build a new computer.

I do still have an old GPU as well if people think I'm off track and this is a GPU issue at the end of the day.

 

[astranger200]

23 minutes ago, roy5000x2 said:

Okay, this has been a fun one. My computer is crashing during certain higher/prolonged CPU usage sessions. So far I've seen full shutdown during:

• Assassin's Creed Odyssey (15-20 min in)
• Deus Ex: Mankind Divided (1 hour + in)
• OCCT:Linpack stress test (5-10 min in) *this one exhibits odd behavior. Around 5 or 10 minutes into the test, CPU temperatures will completely spike and go from a steady ~upper 60s to touching 90C before either the test stops itself from overheating or a shutdown occurs.
• Intel Processor Diagnostic Tool (during CPU Load test)

 

Hardware/Software most items are 5 years old now unless called out:

 

  Hide contents

 

• Windows 10 64 bit
• Intel 4790k running at stock clock
• Asus Maximus Gene VII running BIOS 3003 (I had updated to 3053, though that's also when I first experience shutdown issues and rolled back. Roll back didn't help unfortunately)
• Corsair H100i
• EVGA GTX 1080 Ti (was overclocked, have since brought it back to stock clock speeds)(around 1 year old)
  • 436.02
• EVGA Supernova G2 850W (brand new)
  • Replacing a Corsair HX850 that was failing the OCCT PSU test
• 16GB DDR3 Corsair Dominator Platinum 1866
• Corsair 350D Case

 

 

Things I've checked so far

  Hide contents

 

Thermals: After trying out several different pieces of monitoring software individually (HWinfo, RealTemp, CorsairLink, AI Suite), CPU usually sits in the upper 60s, jumping up to the low 70s during gaming / stress tests (note about OCCT:Linpack above with the exception). H100i Pump will get into the upper 40s. GPU sits at a comfy ~70 under synthetic load.

Event Viewer Logs: Just your regular Event ID 41 stating an improper shutdown occurred. Some DCOM errors have been floating around as well, though a lot of what I've read up on doesn't point to them as being an issue.

Memory Check: RAM passed Window's memory test (though I haven't run MemTest86 yet)

Replacing the PSU: My Corsair HX850 didn't pass OCCT's PSU test, so I swapped it out with a brand new EVGA Supernova G2 850W. While I can now pass said test, does not help prevent shutdown.

BIOS rollback: Just in case, I rolled back from ASUS's 3503 to 3003 BIOS, as I started experiencing the shutdowns around the time I flashed the newer BIOS. No change in behavior however.

Other Stress Tests

• CPU: I can run FurMark + OCCT:CPU / AIDA64 for well over an hour with no issues (haven't tried Prime95 since that was notorious for trying to grab more voltage that it should with Haswell). Other much less stressful games (Borderlands 2, CS:S) don't pose any issues either.
• GPU: FurMark for an hour + no problems, OCCT:GPU doesn't pose any challenges. Heaven keeps on going no problems

 

 

 

Other oddities I've noticed recently

  Hide contents

 

First was my motherboard having trouble booting and throwing the A2 error code. Sometimes unplugging my keyboard during boot would resolve, though eventually it refused to move past the code. Resolved by doing a BIOS flashback to 3503.

Nvidia driver / crashing issues: Some games (Everspace, DX:MD) will sometimes crash and throw memory allocation errors or the D3D device removed error. This seems don't seem to be too uncommon for people.

Computer also shutdown when attempting to install the latest version of Nvidia's drivers (wiped out all of my old drivers and install the new ones fresh to resolve).

 

 

 

I'm running out of ideas. Really starting to lean towards either the CPU or Motherboard dying, though other insight could be nice. Might just be time to build a new computer.

I do still have an old GPU as well if people think I'm off track and this is a GPU issue at the end of the day.

 

I would be pointing fingers at ram. run tests on the ram to see if its having issues

[roy5000x2]

12 hours ago, astranger200 said:

I would be pointing fingers at ram. run tests on the ram to see if its having issues

I'll definitely give running more RAM tests a go

 

12 hours ago, SnowWolf370 said:

I really hate to be the person who... Says this, because it's always kinda douchy. But I did not see it mentioned in the things you've tried.

 

At first I was thinking about things, but then I read what you tried.. Then I was thinking, maybe it could be software based? Have you tried running a clean install of Windows?

Hey all good, solid suggestion! Haven't done a fresh Windows install since I first put 8.1 on the boot drive. Guess it's high time I sorted out all of the steps needed / backups so I can wipe the drive and start fresh. Sounds like a good long weekend activity 

 

 

Okay, think I have a good path forward between doing some RAM validation first just in case, then reading up and doing a fresh Windows 10 install. Appreciate the input!

[roy5000x2]

Wanted to circle back to try and not be that guy if anyone ever is trying to solve for the same issue and then doesn't see an answer in a somewhat useful looking thread 

 

Did a full Reset which did wonders. Helped clear out a lot of clutter on the boot drive, PC is a little snappier, and thankfully shuts down less often yay!!!

(Also did a mem86 test which passed, and swapped power cables from one cable running from my PSU with 8 + 6 pin to the GPU to two cables just in case)

Less often though, still ran into an abrupt system shutdown when running OCCT: Linpack and once when playing Deus Ex Mankind Divided. I'm wondering if the OCCT: Linpack is causing the system to try and get more voltage similar to how Prime95 interacts with the Haswell architecture if anyone remembers that. I'm not going to worry about it too much.

As for a shutdown during gaming though, I decided to take another approach that seems to be stable: undervolting my GTX 1080 Ti. I'm getting near the same speeds on the card at 1v compared to it at 1.065v, so this is already a power and thermal win.

 

This also illuminates that the issues I was running into were most likely several things happening at the same time across the PSU, CPU, and GPU. I mean most components (sans PSU and GPU) are 5 years old now, so the wear and tear could be showing for everything else.

 

Most likely if I wanted to have the ability to crank everything up to the limits I'd need to buy a whole bunch of parts to test with / build a new computer. In the meantime though, I'm happy where things are at. Thanks for the suggestions!