CPU only stable when extremely underclocked [SOLVED]

[Brovahkiin]

I had random crashes every now and then for a while, recently they became very common while gaming and after excessive testing i found that the CPU must be the problem...

I got an AMD fx-6100 running @ 3300mhz on default stock setting if I run prime95 stress test (or any game that stresses cpu up to 100%) the PC crashes always within one minute running it

I underclocked my CPU first to 2300mhz resulting in crashes after 4-5mins of prime test and now i got down to 1600mhz and so far i got no crash at all but ofc the performance suffers alot and i have no idea what could be causing these crashes...

CPU is about 4 years old but never was overclocked or anything but underclocked right now.

I first thought the CPU wasn't getting enough volts so I tried to set CPU VCore higher without positive impact

I don't even want it overclocked just back to 3300mhz or at least something around 2800mhz so i don't have to get a new CPU

My system specs are:

OS: Win 7 64bit SP1

MB: ASRock 970 Extreme 4

CPU: AMD fx-6100 (Zambezi 32nm Technology)

Ram: 8gb dual channel DDR3 @665mhz

GPU: Radeon R9 380 4GB GDDR5

[Prowlcorp]

Yeah....they really should make a small gadget to test cpu since its one of the most expensive things to diagnose

[_I_]

lower the core voltage
1.1 - 1.3v is normal for stock speeds, many boards default to 1.4v

the asrokc 970 ectreme4 isn that good for overclocking
its a weak config 4+1 with vrm hetsink
it will throtlte when the vrms overheat

make sure there is a fan over the mobos vrm heatsink near the cpu

a good read here

[Alexalmighty502 Sup Acc]

what are the temps of your whole pc use a program like HWMonitor to check it

[Brovahkiin]

引用自 _I_：

lower the core voltage
1.1 - 1.3v is normal for stock speeds, many boards default to 1.4v

the asrokc 970 ectreme4 isn that good for overclocking
its a weak config 4+1 with vrm hetsink
it will throtlte when the vrms overheat

make sure there is a fan over the mobos vrm heatsink near the cpu

a good read here

I never wanted to overclock i just want back to stock 3300Mhz for my CPU (this setup ran for 3 years without any trouble) and again all hardware temperatures are not even close to critical values highest i ever got was 70°C on GPU and 55°C on CPU...

If I'm unlucky and no VCore tweaking or anything else helps I just go and buy a new CPU thinking about fx-6300 since I have to keep 95W in order for my PSU to work properly and having a little upgrade for 100€ (getting the same cpu would even cost more)

Or maybe anyone has a suggestion for a better price/performance CPU for 95W? in general I was quite happy with the 3,3ghz six core for gaming and getting the most expensive could result in a bottleneck effect so maybe something around 100€ is fine...

Even if I had to rebuy the ASRock 970 Extreme 4 this would be only another 85€ so it's still better than a whole new PC i think

Anyways thanks for suggestions and help it's much appreciated, I'll update as soon as I tweaked some more around and maybe got lucky...

At least it's stable at 1600MhZ for now :-/

[Brovahkiin]

引用自 Alexalmighty502 Sup Acc：

what are the temps of your whole pc use a program like HWMonitor to check it

Currently running Prime95 and my detailed temperatures are as follows.

MB:
SYSTIN max 36°
CPUTIN max 50°
AUXTIN max 87° but stable at 51° to 53°
TMPIN3 max 40°

CPU: max 39° (on 1600Mhz got up to 55° on 3300Mhz before crash)

HDD: max 29°

GPU: TMPIN0 max 57° (could go up to 70° maybe 75° while in game)

AUXTIN seems to be the only value that got to high at some point

[Alexalmighty502 Sup Acc]

do you crash when auxtin is high or is it a random crash with no relation to temps?

[Brovahkiin]

I don't think it's AUXTIN related since it did not crash right now at those 87° and it always crashed after just a minute of prime95 on 3300mhz when it didn't even had time to rise much but I'll try and clock it back to 3300 and keep an eye on AUXTIN temp

[Revelene]

Kernal power does not necessarily mean hardware failure.

If you are using default settings for voltage, you are probably just using too much voltage.

It is never a good idea to run a synthetic stress test like Prime95 on stock settings.

If you are going to run enthusiast level programs that are designed for enthusiast level testing... at least go in knowing why you are doing it in the first place.

Synthetic benchmarks, like Prime95, are worst case scenerio and will very rarely ever be encountered in everyday use... especially if you game and don't even overclock.

That said...

Are you experiencing the crashing outside of stress test? You said your motherboard beeped a certain amount of times? Look in your mobo manual to find out what it means.

[Brovahkiin]

引用自 Revelene：

Kernal power does not necessarily mean hardware failure.

If you are using default settings for voltage, you are probably just using too much voltage.

It is never a good idea to run a synthetic stress test like Prime95 on stock settings.

If you are going to run enthusiast level programs that are designed for enthusiast level testing... at least go in knowing why you are doing it in the first place.

Synthetic benchmarks, like Prime95, are worst case scenerio and will very rarely ever be encountered in everyday use... especially if you game and don't even overclock.

That said...

Are you experiencing the crashing outside of stress test? You said your motherboard beeped a certain amount of times? Look in your mobo manual to find out what it means.

The crashes occured mainly while in game for CPU hungry games but also sometimes while browsing but this just happend frequently for a while and later only on games then again for some time it ran for a couple of days up to some weeks without any crashes even when in game.. currently on some games the pc just crashes after just a minute in game every time as in prime 95 and i know it's worst case scenario but so are some games at the moment^^

interestingly sometimes when the pc crashed quite often windows tries a system restore and after that (not every time but quite a few times) i had fewer or no crashes for a while which i can't understand if it has to be hardware related usually newer drivers won't do harm so i updated them but no changes with that...

Mobo beeping was only after i produced alot of crashes repeatedly via prime95 but I'll research on that too

EDIT: I'm going to lower VCore as suggested and see if it does any good but yet again on stock settings it ran 3 years without any failures if it's damaged and can't do with stock voltage I might really be better off with a new CPU :-/

[Fluffy]

he said voltage is at 1.2375 and some people on here saying lower the voltage 1.1-1.3 is normal etc.. hes at stock voltage and has this problem after running the cpu for 3+yrs on stock voltage.. i dont think people are reading this post.. amd specs put safe vcore between 1.2-1.4v

kernal power + problems hes having running at stock = most likely hardware failure/hardware issues which includes heat but heat doesnt seem to be the issue... there is some issue with cpu or ram or mobo.. psu possible but less likely

[Revelene]

引用自 Brovahkiin：

引用自 Revelene：

Kernal power does not necessarily mean hardware failure.

If you are using default settings for voltage, you are probably just using too much voltage.

It is never a good idea to run a synthetic stress test like Prime95 on stock settings.

If you are going to run enthusiast level programs that are designed for enthusiast level testing... at least go in knowing why you are doing it in the first place.

Synthetic benchmarks, like Prime95, are worst case scenerio and will very rarely ever be encountered in everyday use... especially if you game and don't even overclock.

That said...

Are you experiencing the crashing outside of stress test? You said your motherboard beeped a certain amount of times? Look in your mobo manual to find out what it means.

The crashes occured mainly while in game for CPU hungry games but also sometimes while browsing but this just happend frequently for a while and later only on games then again for some time it ran for a couple of days up to some weeks without any crashes even when in game.. currently on some games the pc just crashes after just a minute in game every time as in prime 95 and i know it's worst case scenario but so are some games at the moment^^

interestingly sometimes when the pc crashed quite often windows tries a system restore and after that (not every time but quite a few times) i had fewer or no crashes for a while which i can't understand if it has to be hardware related usually newer drivers won't do harm so i updated them but no changes with that...

Mobo beeping was only after i produced alot of crashes repeatedly via prime95 but I'll research on that too

EDIT: I'm going to lower VCore as suggested and see if it does any good but yet again on stock settings it ran 3 years without any failures if it's damaged and can't do with stock voltage I might really be better off with a new CPU :-/

Definitely cannot compare an intensive game and Prime95. You will never encounter the same work load in a video game that Prime95 produces... In short, Prime95 hammers the CPU with difficult calculations based on different instruction sets.

Hopefully lowering the voltage on the core will help! The way that is sounds, it does seem like hardware failure is more likely... but you never know.

I'll be waiting on the update!

[Revelene]

引用自 Fluffy：

he said voltage is at 1.2375 and some people on here saying lower the voltage 1.1-1.3 is normal etc.. hes at stock voltage and has this problem after running the cpu for 3+yrs on stock voltage.. i dont think people are reading this post.. amd specs put safe vcore between 1.2-1.4v

kernal power + problems hes having running at stock = most likely hardware failure/hardware issues which includes heat but heat doesnt seem to be the issue... there is some issue with cpu or ram or mobo.. psu possible but less likely

I didn't see the post that says his voltage. Still don't see it, but I'll take your word.

If that is the case, then hardware failure definitely seems more likely, but I still stand that one should never run synthetic stress tests with stock settings.

It also doesn't hurt for him to try lower voltage. I've seen many computers pass by me that had issues running higher voltage with age, especially AMD processors.

[Brovahkiin]

It seems like anything, from games to browsing with lots of tabs opening at once or prime95, that makes the CPU load go up to 100% for a certain time produces a crash..
I figured the more I underclocked the longer it took in prime95 to produce a crash having none at 1600mhz...

For the RAM I did alot of Memtest and switched them out with older and it's not bad RAM maybe the Ram voltage could be the source but I don't think it's likely

I try lowering the voltage a little but i doubt this will help

[Brovahkiin]

引用自 Brovahkiin：

Sorry forgot about motherboard it's ASRock 970 Extreme 4

default cpu voltage is 1,2375 and NB voltage is 1,1750 I tried to increase both values step by step but had no good results only way to make cpu stable seems running at 1600mhz (only tested 2000mhz and it crashes again)

I don't think it could be heat related since I never had more than 55°C on CPU even on stress test (probably aswell because it crashes before raising over 60)

Thanks so far for your quick suggestions I'll check event viewer right now

No Problem here you go :)

Also does NB voltage need to be altered according to cpu voltage?

[Fluffy]

even if the ram voltage is producing problems it will cause errors in memtest if your passing memtest for long periods of time almost guarantees its not ram then

edit: im not sure what you mean by does NB temp voltage need to be altered according to cpu voltage.. the only time you raise north bridge voltage is for stability issues with oc (possibility to reach higher oc/unlock cpu cores for am3) and for overclocking nb frequencies... your nb frequency would be the same regardless of cpu multiplier.. your at stock (same as mine) voltage for 970 chipset.. and u stated you tried increasing that as well

what northbridge frequency are you running at btw... can see that in hwinfo ex im at nb freq: 2200mhz@1.175v at 1400mhz x8 cpu frq (c1e enabled) if its running above your ht clock it can cause stability issues