Stable in prime95, crashes in game

Using asus p8z77-v lx, i5 2500k 4.2Ghz 1.3V, i only changed voltage and multiplier. I ran prime95 blend for about 10 hours, i also tried combination of prime95 and furmark and OCCT PSU stress test. It is all stable.

Today i played starcraft 2 and after a few minutes it crashed (first white screen then restart). I even tried upping voltage to 1.325V and it didnt help - took a while longer but still crashed. Then i put frequency to 4.1Ghz at 1.325V and it didn't crash anymore.

This doesn't really make sense since prime95 + furmark stressed my cpu a lot more then the game. Temps during stress are cpu 70C, gpu 80C, starcraft 2 my cpu is 60C, gpu is 70C.

I am thinking there must be something else beside cpu voltage. Because even 1.3V for 4.2Ghz is not good at all. I had another i5 2500k chip and it was the same.

 

Furmark doesn't work anymore for GPU stress testing. NVidia is detecting it and putting the GPU in a low-power state. :-/

You'll need to use another GPU bench/stresstest.

 

I tried also OCCT and Fire Strike (Extreme) ... And i dont think it has anything to do with gpu since i didnt overclock it ..

 

Check event viewer and see what it says after crash.

Could be unrelated to CPU

 

I am not seeing anything special in event viewer except error/critical saying pc wasn't restarted/shutdown properly.

Remember, i was playing last mission in SC2 (LOTV) on brutal with loads of enemies, 1.325V 4.2 crashed, 4.1 didn't. I played another mission with less enemies and it was stable on 1.3V 4.2, also small FTT, blend, inplace, etc. prime95 is stable over night, 3dmark, occt furmark, all kinds of combinations, everything is stable. I would even say it is the games fault, but in all the time i played it apart from this, i never had even the tiniest problem with it. And when going to 1.325V 4.1 it didnt crash.

 

I tried playing that same mission again and now even on 4.2Ghz 1.3V i couldnt produce a crash. Granted i loaded a save game from the last 10 or so minutes of the game but that is also the time when most units are on the map. I am thinking that maybe it wasn't my overclock but a bug in the mission. Anyway until i can produce this error again there is nothing else i can do. I will make sure i have some logging programs running everytime i play SC2.

 

Bugs in games cannot make your PC restart. Only driver bugs and hardware errors (due to OC or damage) can do that.

 

I've had similar experience 12 hours stable in prime95, but crash within 30 minutes gaming in PUBG.
Now I'm running multiple tests like cinebench, p95, games, handbrake, ect...
Still haven't found max stable on my cpu after a few weeks of playing.

Try to check ram & chipset voltages? Are they all on manual?

Some motherboards like mine automatically put them too high, making it hard to overclock. For me, dropping CPU VTT and PLL voltages helped a lot.

Also minimum voltages for P95 stable vs actual stable voltages are different for me as well. Prime95 stable at 1.22 volts, but had crashes in games until I went up to 1.26 volts.

My motherboard also set my RAM timings and voltages wrong, its worth double checking everything to make sure its within spec.

 

I'd try IBT (IntelBurnTest) as it is more stressing on the CPU then Prime95. I have ran good Prime95 and failed a IBT. Plus IBT will find instability it much faster then prime95 will.

 

it could be you're psu might be to weak for over clocking? I notice you have 600watts? Maybe upgrading to least 750 watts that way you can upgrade your graphics card when you want too.

 

Try prime95 custom size 1344 and 1792. I was stable on prime95/ibt and had bsod in games sometimes , then I tried these settings and I found out my overclock was unstable. If you can pass these for about an hour you are 99% stable.

 

Is ur gpu overclocked?
doen't seem is the cpu. 1.3v is more then enough for that mhz.
Try diff slot, you didn't mentioned ur ram config

EDIT: sorry you did mention ur gpu is not overclocked

 

Stress tests errors occur when we try to stabilize our o/cing.
Many times these errors can corrupt windows system files, that obviously remain corrupted even after we achieved a stable o/cing.

You can try sfc/ scannow, and see if it finds any errors.

 

Try with RealBench stresstest, best program for stability testing.

 

So OP is that crashed sorted or you still having problem?

 

As already pointed out, 600W is not a big power supply. would suggest 850W at least for your setup for overclocking and stability.

 

a quality 650~ watt is plenty for his setup.

Even OC 2500k will only use 110w~, 390x about 300 watts.

Thing is, nobody knows what OP has, i don't see anywhere he said about his specs.

 

So you can guarantee that a 650W PSU will be able to supply enough current to the 12V needed during heavy over clocking etc?

We dont know would other devices are being powered off his rig let alone just his main components.

 

If it were psu it shouldn't run prime, so I don't know what is ur point. Also 650w is more then enough for a quad core setup...

 

Your PSU is fine, it can be the mobo or GPU.. Hopefuly you can sort it out.