FX-8350 core crashing under heavy load

milandrag

Reputable
Oct 6, 2015
8
0
4,510
Hey

I have started experiencing some issue with my PC. When i was playing some cpu heavy game (CS:GO, skyforge...) my pc tend to stop showing any picture (loss of signal). If i emmideatly alt-tab out of the game i get the desktop picture back and going back in-game gives me like 5sec or less of picture before going dark again

i have done some testing arround with prime95 and furmark and furmark passed well, while prime95 usually gives me back a "Hardware failure error" thus droping 1 core (#5 or #6). when doing both tests simultaniously, i got a bsod and .dmp file showing crash of "ntoskrnl.exe" (windows kernel according to some googling)

i changed the graphic card to some older one and did the test with same results (thus i believe graphic card is not a problem)

if i disable cores #5 and #6 in BIOS all test pass without any crashes or errors, also games don't go black anymore.

some time ago while i also got a few bsod with something like "ntfs" error codes, but those were gone since i stopped playing games from SSD

the bios setting was 1st set to preset performance (no manual OC took place) but was as of recently changed down to default settings.

my question is, is the CPU bad and should i take it back, or is there just some issue with some setting that doesn't let me use all 8 cores?

my rig:
cpu: FX-8350 4.0GHz Black box
fan: cooler master hyper 212 evo
motherboard: asus M5A99X EVO R2.0
ram: Kingston HyperX Fury 8GB 1600 DDR3 HX316C10F/8
gpu: Gigabyte GeForce GTX 660 (GV-N660OC-2GD)
psu: Tecnoware Black core 650W (FAL654BLC14)
 
Solution
most probably its a power issue since your mobo qualifies for running 125w tdp vicera cpu, however i doubt your psu does.! looks like it fails under heavy load.

psu: Tecnoware Black core 650W (FAL654BLC14)

thats an unknown brand. they dont even mention any real specs of their products.!

get a quality psu for your system

http://www.tomshardware.co.uk/forum/id-2547993/psu-tier-list.html

a quality 650w psu would be more than enough and be able to support any future single gpu upgrade.

Please check the manufacturer for newer BIOS update. It could be a glitch there.
Second, check your bios settings. It could be not enough power to the CPU. If you are ok tinkering with the BIOS, you can try to switch off any AMD's Turbo settings, and then try again with the prime.
Once you do this , type us a small "report" of what you have found.
Update:Latest BIOS on Asus page is version :2501. Check if yours is too old.
 
ok so i disabled amd turbo boost in bios and im running prime now, sofar so good although it always took about 15-20min before a core crashed
i also checked, i have bios version 2201 although i don't really see this could be such a problem since the rig was working normally for about a year
i will give a try also to upgrading bios if problem remains
 
most probably its a power issue since your mobo qualifies for running 125w tdp vicera cpu, however i doubt your psu does.! looks like it fails under heavy load.

psu: Tecnoware Black core 650W (FAL654BLC14)

thats an unknown brand. they dont even mention any real specs of their products.!

get a quality psu for your system

http://www.tomshardware.co.uk/forum/id-2547993/psu-tier-list.html

a quality 650w psu would be more than enough and be able to support any future single gpu upgrade.

 
Solution
ok so even with turbo mode disabled i still got to crash core #5 after 21mins of prime95, although with a different error (illegal summout)

is there any way to test the psu disability further before buying a new one?
 


dont bother and better not. psu is the most crucial pc part an unknow psu will be of low quality. dont endanger your system.! since you have been stress testing you should have mentioned cpu-gpu temps.regardless of them being good or not , you must get a proper psu. (your cpu would throttle with temps above 65c)
 
ok i will consider buying a new one
how much better is it to buy tier 1 to tier 2?

temps during stress testing were (max)
gpu 64°C (fan at 66%)
cpu 50°C (fan at ~70%)
 


temps look good however cant be accurate since cpu is throttling due to insufficient power! the same might be happening to gpu too.!

it will be much better especially now that you are overclocking and you wont be dealing with any longivity issues at least for 5 years!

my picks would be xfx / evga B2 - GS - G2 / seasonic
 
ok so, since the computer worked fine with 2 disabled core, i figured i should swap the cores desabled (to leave the previously crashing #5 and #6 running) and it turns out my black screens returned when i had them enabled
turning them back off instantly solved the issue

i am thinking now that afterall cpu might be a problem, will do some testing, but will most likely take it back to be fixed
 
power ploblems can result on such wierd activity. cpu itself choses which cores to tax more. maybe those are the ones which bear higher loads when enabled. whithin my frequent presence here in this forum, i have never encountered a defective cpu, yet hundrends of defective psu!
 
I agree. It is a common situation. When I overclocked my FX 8120 with not enough power it started freezing, errors in prime95 ,etc. When you disable core #5 and #6 you use less power - which makes the PSU work better (not struggle) and thus you have less issues.As a temporary solution you can downclock (manually) your CPU and reduce a little the voltage for the CPU. This could temporarily let you use your PC but I must imply that PSU is the backbone of a PC.If it doesn't operate within very tiny margins (out of specs) It can fry (YES - FRY) your GPU, CPU or Motherboard which are expensive components.
If you have a multitester and you are ok with electronics - you can test what voltage you get from the power supply - while the PC is running.If that's not for you - you can go to a specialised shop where they can test your PSU. Keep in mind that not only voltage , but also amperage (I) must be in order. Also the PSUs are designed to filter any "electronic noise" that could cause issues with modern PCs.
PS: Modern electronic shops use oscilloscopes to check any electronic device. Even a multitester can't give you the "whole picture". According to me , It's too risky to use your PC at the current state - you will have to swap this PSU.
 
Ok, so i have got myself a brand new PSU, EVGA G2 650W and at 1st everything looked normal, then the problem reapeared though a bit differently.
runing prime and furmark simultaniously produced an error on prime (before replacement was bsod)
running only prime crashes 2 cores throwing out 100 warnings (before replacement was error which stopped 1 core)

so i am currently surprised and clueless what can be wrong but i'm going to start testing ram now, although it already passed 1 memtest95+ (yes i know, should be 7 or 8)

any suggestion as to what else i can try?
 



you have been messing with bios. try resetting to default. also try a different stress testing as prime 95 may lead in unexpected errors with amd FX
 
ok so at first i kinda forgot to reset bios and i was still getting errors, but since i reset it back to defaults, i didn't get any errors on 1st test in prime (i got errors before reset)
right now i'm feeling confidant, that the system will sustain alive, but i will see once i get it back into action (yes, gaming 😛)
thank you for all your explanations and suggestions and of course, your help
 


really glad to hear such good news, you're welcome!:)