New build keeps crashing, likely CPU overtemp, help!

Toggle sidebar Toggle sidebar
A

antilogic1

Commendable
Sep 25, 2016
8
0
1,510
I just built a new PC and everything is great...except it crashes when playing games. I have been searching the forum and internet for a few days looking at similar cases, but I have yet to crack my problem yet completely. I would really like some advice!

What happens:
After ~15-20 minutes (pretty consistently) I get a black screen and the fans go to max. Sound (from game) seems to continue to Play. I have to hard shutdown and then re-turn on PC.

CPU: Intel Core i7-6700K
Cooler: Noctua NH-D15
Mobo: GABYTE Ultra Durable GA-Z170X-Gaming 7
GPU: GIGABYTE GeForce GTX 1070 8GB
RAM: CORSAIR Vengeance LPX 32GB (2 x 16GB) DDR4 3000
SSD: Samsung 950 Pro - 512 GB - M.2
HDD: Samsung 850 EVO 1 TB & Some older WD HDDs
PSU: SeaSonic G Series SSR-550RM 550W
Tower: Fractal Design Define R5

Things I've tried:
-Enabling XMP in bios ("disabled"-> "profile 1")
-Trying to set DRAM voltage manually to 1.34 (it was 1.2V, but my RAM specs say 1.35V..weird?)
-Checking WhoCrashed (no dumps found?). Checking Window Reliability Monitor (it says Application, Windows and Misc Failures)
-Updating Nvidea drivers using GeforceExperience app (372.90)
-Running FurMark, Memtest86 and Games monitoring temperatures using HwInfo.

Screenshot of HW Info is here:
https://postimg.org/image/3jmfr4f6f/
Picture of my setup here:
https://postimg.org/image/50qxtcujh/
Graph of temps over 7 minutes of Prime95:
http://postimg.org/image/kshrssb0h/

I think I've finally made some progress... I'm now running Prime95 and my CPU seems to be climb from 35C to 70C as soon as I start Prime95. When running it for a while, it appears to hit 92C so far as a peak. My 2x CPU on the D15 seem to top out at 1200RPM although they are still barely audible at this speed.

Could this be causing the restarts? I doubt a game (e.g. Xcom 2) would peg my CPU at 100% like prime95 can though...so I'm not sure if this is it or not.
 
Solution
Multipack
Possibly, but not necessarily. It's a laborious process, but try testing one stick of RAM at a time in each slot. It could be that there is a fault with your MB/CPU etc instead. If both sticks fail on their own then the chances of having two new faulty RAM modules is quite unlikely. Maybe you can narrow it down to one RAM/one board slot. Like I say, its a pain but the last thing you want is to replace the memory then find you still have the same problem. Incidentally, the RAM isn't on the supported list for that board, but I would say its highly Unlikely thats the problem as they never test all makes and models. What would help is to make sure your BIOS are up to date and you've cleared the CMOS too.
Sort by date Sort by votes
Multipack

Multipack

Honorable
Sep 5, 2016
4,283
3
9,960
Prime95 with memory stress included will take temps v high v quickly but 95 is too high so you have a cooling problem for sure. But the 1.35v is the OC voltage. The default MHz frequency only needs 1.2v. You shouldnt up the voltage that high at 2133MHz.
 
Upvote 0 Downvote
A

antilogic1

Commendable
Sep 25, 2016
8
0
1,510
Thanks for the response!

OK, on the memory, thanks will lower back to auto (1.2V).

My cooler should be pretty overpowered, since I'm not overclocking, which leaves me confused. Should I try removing and re-installing? Should the heatsink itself be hot? (mine isn't hot to the touch).
 
Upvote 0 Downvote
Multipack

Multipack

Honorable
Sep 5, 2016
4,283
3
9,960
I'd be surprised if your cooler wasnt good enough. Although it isn't the best, temps that high (92, not 95 like I said. Had Prime in my head lol) shouldnt really be getting hit. The heatsink should be warm yes but it dissipates heat very quickly and if it wasn't working at all your temps would skyrocket as soon as you hit the power button. 1200rpm is the max for the fans. Did you attach it tightly enough ? So the thermal paste is properly sealed? What does a non-memory Prime95 test peak at?
 
Upvote 0 Downvote
A

antilogic1

Commendable
Sep 25, 2016
8
0
1,510
Non-memory Prime95 seems to peak around 90C so far. It fluctuates a lot less than the blend test.

I assumed I had attached it tight enough, but I'll look into that or try re-seating it as a next step.

What kind of temp would be considered normal with prime95 pegged at 100%?
 
Upvote 0 Downvote
Multipack

Multipack

Honorable
Sep 5, 2016
4,283
3
9,960
I can get to 75-80 with a 6700k personally but have seen others get higher. You are basically damaging the CPU internals at 92 if done for any length of time tbh. It could well be throttling to stop it going higher too as that's about the max.
 
Upvote 0 Downvote
A

antilogic1

Commendable
Sep 25, 2016
8
0
1,510
:(

I'm still having problems, so maybe I had more than one issue.

I'm getting some strange crashes (witcher3 -> game quits suddenly to desktop with no error, xcom->"game has stopped responding").

And I just got a black screen, but sound stayed on and the game seemed even playable still. The fans went to max speed and I checked and it is the GPU fans. The case reset button doesn't work.

I just made sure I had the latest drivers, and steam overlay is turned off.

Any more ideas? I guess I should monitor temps again closely (both GPU and CPU)?
 
Upvote 0 Downvote
Multipack

Multipack

Honorable
Sep 5, 2016
4,283
3
9,960
Do some software scans/physical checks with your own eyes on your hardware and see what you can find. Also, check Windows security and maintenance for any reported problems. It could be CPU/PSU or MB. Hard to say without ruling things out one by one. Try software checks first, like chkdsk/memtest86 and processor diagnostics. If you pass all of those then you may have to turn to the hardware itself.
 
Upvote 0 Downvote
Multipack

Multipack

Honorable
Sep 5, 2016
4,283
3
9,960
Possibly, but not necessarily. It's a laborious process, but try testing one stick of RAM at a time in each slot. It could be that there is a fault with your MB/CPU etc instead. If both sticks fail on their own then the chances of having two new faulty RAM modules is quite unlikely. Maybe you can narrow it down to one RAM/one board slot. Like I say, its a pain but the last thing you want is to replace the memory then find you still have the same problem. Incidentally, the RAM isn't on the supported list for that board, but I would say its highly Unlikely thats the problem as they never test all makes and models. What would help is to make sure your BIOS are up to date and you've cleared the CMOS too.
 
Upvote 0 Downvote
Solution
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom