Prime 95 fatal error, cores stop working , no overclock , ram tested FX 8350

[sunzzofman]

Hey guys i really need some help with my computer. So the other day my PC froze randomly, no blue screen it just completely froze. I'm not using any turbo or overclock settings with my motherboard, the CPU is at stock 4 ghz

iv also noticed my PC just doesn't seem to be running that well in games etc. So when i run prime 95 i get a FATAL ERROR: Rounding was 0.5, expected less than 0.4 and this can be on any cores , sometimes two of them , or other times none of them. I ran memtest last night for around 8 hours with 2 pass's (and half way through the next) with 0 errors.

Whats causing this? CPU temp doesn't go over 60 c and im using the corsair closed loop water cooler , motherboards at 35 c . iv got my specs in my signature .

(Just tried prime 95 blend again and cores 2 3 and 4 stopped, cpu usage went down to 60% and prime continued to run.

i'v even blocked up some of my fans to see if temp was making the cores stop by 5+ c more and there fine. it seems to be random :S

any help is appreciated love you guys 😀

EDIT : Ran memtest last night for 8 hours and it did 2 passes with no errors, pc just froze again :S

http://postimg.org/image/qlxtgndc3/

 

[lukelaw]

I don't know what the problem is but your voltage is a little high, on stock you only need 1.35 max and 1.325 min

 

[sunzzofman]

I don't know what the problem is but your voltage is a little high, on stock you only need 1.35 max and 1.325 min

but i haven't touched anything :S its all stock lol so why is it higher then it needs to be?

 

[sunzzofman]

bump 🙁

 

[lukelaw]

AMD boost probably was enabled in the BIOS at some point in time

 

[titanHUNTER]

Turbo boost has nothing to do with it. I have the 8350 and I've never had any issues running Prime95. Never go above 43 in fact (after 45 minutes of Prime95). I actually want to use my PC, not just stress test it for hours! lol

It may just be a bad chip. How long ago did you purchase it? I would RMA it ASAP!! If the cores are not functioning, then the CPU is not functioning. I guess it happens. It was not used/refurbished was it? Either way, RMA it.

 

[titanHUNTER]

Also, the comment about voltage may be the reason also. Go into your BIOS and check the voltage.

 

[Guest]

Is anyone else thinking MSI motherboard?
Because I'm thinking MSI motherboard.

Jokes aside, hop in to your BIOS and check the readings. Perhaps your mobo is over or under shooting the voltage causing the instability. Fortunately if that's the case it would be a simple fix of just setting it to the default for that chip.
(I'm not sure if it varies between 8350s, but for mine that was 1.3625v).

You can check that using CoreTemp. Under a bit of load it will list the stock voltage.

Start by opening CPU-Z and/or HWMonitor and look at the CPU VCORE (Or, 'Core Voltage' for CPU-Z) reading.
Watch this reading as you run another stress test, only for about a minute, it's going to reach the load voltage almost as soon as you start so you don't need to watch it gradually rise like temperatures.

I suspect this because of something/s called Vdrop/Vdroop. Put simply:
Vdrop = The slight drop in voltage compared to what it's manually set as in the BIOS, under normal conditions, seen in various software programs (like the two above).

Vdroop = A further drop in voltage under full load.
Usually this isn't much. Say around 0.02 either way. But if it's enough to significantly go under stock voltage that could be where the problem lies.

I know that I had some issue with the CPU LLC (Load Line Calibration) causing it to underpass the labelled stock voltage, which made one of my cores fail. More voltage isn't a good thing, it ups temps, power consumption and depending on how much, a little bit of chip degradation, but as long as it's over stock both normal and under load things should be fine.

If your BIOS has the option. Set the manual CPU voltage to stock, and find the CPU LLC setting, moving it to High/Ultra High (75% works best for me), this will essentially do the job of controlling voltage for you, allbeit slightly over shooting the mark a little.
If not, do the same by setting the manual voltage, and keep upping it by tiny amounts until both normal and under load CPU-Z/HWMonitor reports the voltage above stock. Only slightly above stock, let's not go too far here.


Another thing to mention is that your mobo is a 970 chipset, which isn't so great for overclocking.

Of course this is me speculating. The only other things I can think of are:
- Your board needs a BIOS update to fully support the FX line.
- Your board is getting too hot, the VRMs are exceeding a certain temperature and causing it to throttle the voltage, therefore causing instability.
- Your board is an MSI board (ahuehuehue ijest)
- Your board is this one (Quick Google): http://www.tomshardware.co.uk/forum/330572-30-970a-warning

I'm blaming the board a bit too much, but seemingly your CPU isn't exceeding thermal limits, and your power supply is more than capable, your ram is seemingly fine by your tests. It's the only other thing I can point the finger at.

 

[Guest]

.... Ah crap, this is a two week old post isn't it?

*sadface*