8350 crashing under load at stock

Everything is at stock - voltages, freq, bios settings

 This usually happenes:

 I run prime95 - no errors for 10-20 min and then system crashes. Just black screen and restart. However, if i start prime95 and launch some application while its running (like chrome, player, game, word...) - instant crash to black screen. No bsod, just black screen. And no errors in prime95.

Same thing with OCCT and IBT.

I also ran x264 encode for about 30 min and it worked great until i launched vovie player (mpc-hc) - instant crash

 

-CPU temps are ok - 45C under full load

-there is no dust in pc case

-i checked if all power connectors were plugged to mbo and gpu.

-I ran memtest - 0 errors

-i flashed bios to older version

-i upped voltages to northbridge and ram

-i also tried to oc it to 4,[email protected]+0,025, +0,050 and [email protected]+0,050 - same thing

 I have no idea why is this happening. Here is log from OCCT power test. Ignore the dip in the mid, I accidentaly alt-tabed from occt.

http://imgur.com/a/VkWyb

 

8350, 990xa-ud3, kingston-1600 2x4gb, xfx-550w, msi 7950 boost, win8.1 x64. All latest drivers and win updates installed.

Tell me if you need me to post some other stats like bios settings, hwmonitor etc.

Help!

 

the motherboard you are using, the Gigabyte GA-990FXA-ud3, how old is that board? do you know which revision it is?

THe older versions of this particular board had some overheating vrm´s and northbridge issues, this causing voltage throttle, and could cause crashes, wenn stress testing.

It about 9 months old, same as the rest of the rig. Rev 3.0

Btw, i just droped cpu to 2.0ghz without lowering the voltage and it seems to be stable now.

What is your PSU rated at?

80+ bronze. Model is xfx pro550w 

I just tried with 4.0(default) with core boost disabled and with voltage at +0,050. Same thing. OCCT psu test (both gpu and cpu) - same thing

1. Download HW monitor

2. Open your cabinet and check the temps on mosfets nearby

3. Fault may be with your PSU. Run some other benchmark like Cine bench. If it works fine then there many be a software bug. 

red square - while idling, heatsink ts only warm, but under load it gets really, really hot. I cant hold my finger on it for more then a second. Cooling on that heatsink is really good. Also cpu fan is larger than cpu heatsink and blows directly over the socket to the heatsink.

Green square - hot both while idling and in load

Orange - also a heatsink - cold all the time

Will run cinebench in a minute.

Forgot link.... http://imgur.com/oPM2bhg

HW monitor under cpu load:

http://imgur.com/Qs7RCHQ

Cinnebench passes bot cpu and gpu test. I can also launch other programs while it works and system isnt crashing.

you could try to disable LLC (load line calibration) in the bios then try again. also disable cool n quiet.

Already tried with disabled cnq along with all other power saving features and turbo boost. LLC can be set to: auto, normal, regular, medium, high, ultra high. No idea whats the difference between normal and regular xD

Will try it now.

Same thing with different LLC settings. I tried normal, regular and ultra high. Here are the ss's of bios settings:

http://imgur.com/a/F8hWj

its hard to say what it could be, you are allready using the latest bios?

I know the Revision 3.0 can be horrible.. with overclocking.. so i hope that its not your motherboard, but it could be.

How does the system works furter with playing games and such ? Big chance that your motherboard get overheated wenn running prime. throttles the voltage, and causing a crash.. its a well known issue with the Rev 3.0

Zero problems while playing games or doing anything else besides stress testing. Even prime wont crash for some time if it is the only app running. But as soon as I launch another demanding application - instant crash even if it was running for only couple of seconds before launching another aplication. So I dont think the heat is the problem.

Tried all released bios versions.

I'm starting to feel this could be some driver or software bug rather than hardware...

and whats happening wenn you use Aida64 stress test, and passmark performance test?

Because the black screen, is a bit bottering me.

I'm kinda scared it might be to motherboard overheating like you said:

So far I ran prime95 in 3 situations:

4.0 @ +0,000v - Seems to go for half-hour to hour without crash. H264 encode also black screens after some time.

4,4 @ +0,025v - Run's for about 10-30 minutes without any errors then black screens.

4,4 @ +0,050v - No errors, black screens after couple of minutes.

I touched vrm heatsink after the crash and it is EXTREMELY hot.

 

What would be a good alternative to this mobo in case I RMA it back to shop. Sabertooth 990fx rev 2.0 costs about the same as this one.

The Asus 990FX Sabertooth R2.0, is a rock sollid board. Its one of the best  AM3+ boards on the market.

Grab it if you can ☺

Sorry for my late reaction

Looks like you got one of those garbage rev 3.0 motherboards.Sucks. Get the sabertooth like Mistery said,It has a 5 year warranty and probably the best am3 + mobo.