Overclock.net banner

Blue Screen Troubles vs Overclocking (AMD FX-6300 Vishera)

4K views 49 replies 7 participants last post by  Snoopi25 
#1 ·
Hello! It's been a while since I posted on the OC.net forums, and lo and behold here I am w/ a problem I can't figure out. I need some new insight. I hope this is the proper spot for this kind of topic.

So, I'm running the following specs:

CPU: AMD FX-6300 (OC @ 4.5GHz) with a Thermaltake SpinQ VT cooler
RAM: Corsair Vengeance 1600MHz 4GB x2
Motherboard: Gigabyte GA-990FXA-UD3 rev. 4.0
GPU: EVGA GTX 960 SSC
PSU: OCZ ModXtreme 700W
HDD: WD Green (1TB)

This PC has seen various upgrades since day 1, and this is where I'm at right now. So, I decided to go ahead and milk some power out of my CPU. I followed some basic OC guides, and I know every chip is different. It so happens that I always end up w/ the weak chips lol... Now, I've been fighting with blue screens forever, since day 1. Whether it was Windows 7, 8, 10, whether I changed from a GTS250 to a GTX570, to this GTX960, or from an Athlon II x2, Phenom II x4, and FX-6300. I always had some sort of blue screen. Now, they're really intense and out of nowhere. Besides the blue screens, my PC just shuts off randomly, as if someone unplugged the power chord.

The blue screens I usually suffered from are: WHEA_UNCORRECTABLE_ERROR, SYSTEM_SERVICE_EXCEPTION, and CLOCK_WATCHDOG_TIMEOUT. I have learned how to use the windows debugging tools and I had it analyze my dumps (you will see 2 most recent dump analysis that I included). Sometimes the fault is with Ntkrnlmp.exe, and I thought maybe I have some corrupt files. I did defrags, sfc /scannow, sfc /scannow /r, I did an on-the-spot windows upgrade/install... hasn't fixed anything. I dug a little deeper, and the uncorrectable error BSOD, seems to point to the CPU, at a specific memory address. Based on what I could gather, stuff gets stuck in the CPU and the CPU triggers a force shutdown or a blue screen in this case.

A little bit of info regarding my overclocking. You will see my settings in the pictures I have provided. Needless to say, I have all bios and drivers up to date. When I overclocked my CPU I ran multiple Prime95 tests and monitored my temps until I found something stable. Relatively recently I put my CPU to test by decrypting some files from someone's computer. Thing ran 10 hours a day, 100% load, NO CRASHES OR BLUE SCREENS whatsoever. I was impressed. I ran Memtest86+ for 10 hours, no errors in RAM modules. But if I were to play a game, or watch youtube, or both, I blue-screen. Youtube watching barely uses like 5% CPU usage, and I blue-screen, which is odd, because when CPU load was 100% during those decrypting attempts, nothing happened. I also noticed that if I leave my PC idle for a very long time, it would blue-screen.

So, these blue screens happen randomly. Cannot trace them to a specific program. Temperatures are in check, and like I said, I would crash even if just watching youtube, where system is practically idle and temps are low. If I forgot to include some information, please request it, and I will reply asap.

BSODDebug1 9k .txt file


BSOD2.txt 7k .txt file


21741967_10207339502404193_885770122_o.jpg 146k .jpg file


21745105_10207339502524196_1172703699_o.jpg 153k .jpg file


21754727_10207339502764202_932292440_o.jpg 139k .jpg file


21754832_10207339502964207_1063534208_o.jpg 128k .jpg file


21755016_10207339502804203_210305662_o.jpg 138k .jpg file
 

Attachments

See less See more
7
#2 ·
I had the WHEA_UNCORRECTABLE_ERROR error too for a long time, random bluescreens and that stuff, but found the solution, although i don´t know if its gonna solve the other errors.
redface.gif


You have to go into Device Manager and update the screen/video driver to the last one, if you already have the last one no need to reinstall, just click on update and select to look for the las driver on internet and let the process finish.

thumb.gif
 
#3 ·
That doesn't seem to make any sense. I said I already have the latest drivers installed on my system. I get my GPU drivers straight from Nvidia. Unless, Windows 10 has yet another annoying bug where it doesn't trust drivers unless you get them through windows
mad.gif
 
#4 ·
Quote:
Originally Posted by Snoopi25 View Post

That doesn't seem to make any sense. I said I already have the latest drivers installed on my system. I get my GPU drivers straight from Nvidia. Unless, Windows 10 has yet another annoying bug where it doesn't trust drivers unless you get them through windows
mad.gif
I had the lastest drivers installed when i did it and it worked, it didn´t downloaded or installed anything. I guess it´s some sort of verification from windows.
confused.gif
 
#5 ·
Ok, I did that, let's see if it happens again. But in all seriousness, anything else I can look for? Anyone here that can interpret debugging logs like the ones I included. Funny thing is, those logs clearly point to the CPU triggering the BSOD. For the longest time, I had this conspiracy theory that ever since windows 10, windows updates have been so intrusive that if they need to install something and you're using the PC, they'll force a BSOD on you, so that when PC reboots they get priority, kinda like the government LOL... because every time my PC would reboot, I would see Superfetch or other such update related processes hog up my HDD read speed for like 5-10 minutes. I disabled a LOT of windows stuff, like Windows Defender, compatability telemetry... those 2 were cancer incarnated... So happy to live w/out them.
 
#6 ·
Is your CPU-NB voltage stock? It's for the memory controller.

Could also be a faulty or tired Power Supply Unit. Capacitors perform worse as they age.

Touch your chipset heatsink. Does it burn?

Find ErP in your BIOS and disable it.

Just some possibilities
 
#7 ·
Quote:
Originally Posted by 1216 View Post

Is your CPU-NB voltage stock? It's for the memory controller.

Could also be a faulty or tired Power Supply Unit. Capacitors perform worse as they age.

Touch your chipset heatsink. Does it burn?

Find ErP in your BIOS and disable it.

Just some possibilities
Now that you mention it, i had that motherboard (990FXA-UD3+Phenom II X4 966) and it has VRM and Northbridge overheating issues
thinking.gif
 
#8 ·
Thanks for the pointers... does NO ONE look at the stuff I uploaded? XD I attached pictures of my BIOS settings, including all voltages, etc. etc. You're basically looking at my BIOS settings if you open up those pictures. Yes, I left NB voltage stock, but I changed the motherboard profile to extreme voltage idk what it's called. This motherboard does run hot chipsets. I saw the north bridge can get as hot as 60C during hardcore operation, but from what I could gather, this is normal for this board and nothing to worry about. The north bridge is sitting right next to the exhaust fan, so any heat produced gets thrown out anyway. What is this ErP you speak of? And my power supply CAN be tired as you suggest. It's been running since 2012 I believe. But I never used it to its full capacity. And people here told me 2 years ago that it's a ticking timebomb, that I can't SLI 2 GTX 570's cause it's not enough. I called such BS... I bought a Kill-a-Watt, plugged it into the wall, went ahead and SLI-ed my 570's and never drew more than 500W out of the wall, so I was still WELL under the specified wattage, even with its 80+ efficiency. Now, the thing doesn't even reach 300W when I game (I only have 1 GPU now, the 960), usually stays at 260 or so, so I highly doubt I'm putting a lot of strain on it.
 
#9 ·
Adjust Load Line Calibration. Having this on Extreme causes your voltage to be a-ok under full load but under partial/no load conditions it will be lower than sufficient. This explains the bluescreens you're having.

People usually recommend raising RAM voltage to 1.6 on FX for stability. Wouldn't hurt. NB Core voltage needs raising to get your RAM to 2000+ and NB speed over 2400 to really get performance up. See Minotaurtoo's recent thread for more information.

Your PSU is now a ticking time bomb+2. Seek replacement. Consult Power subforum.

I looked at the pictures. I should be embarrassed but I'm not. Attachments aren't efficient.
 
#11 ·
Quote:
Originally Posted by chrisjames61 View Post

Nobody is going to take the time to look at your attachments. Take five minutes to do it the right way and post them inline. Then you may get more input.
See, if I were to actually throw those pics in, some other admin or whatever will say it looks too messy. Would you like me to post the Debugs as a reply as well? those will surely take 2 pages of this thread.
Quote:
Originally Posted by 1216 View Post

Adjust Load Line Calibration. Having this on Extreme causes your voltage to be a-ok under full load but under partial/no load conditions it will be lower than sufficient. This explains the bluescreens you're having.

People usually recommend raising RAM voltage to 1.6 on FX for stability. Wouldn't hurt. NB Core voltage needs raising to get your RAM to 2000+ and NB speed over 2400 to really get performance up. See Minotaurtoo's recent thread for more information.

Your PSU is now a ticking time bomb+2. Seek replacement. Consult Power subforum.

I looked at the pictures. I should be embarrassed but I'm not. Attachments aren't efficient.
Thanks for the info, but if I raise the NB and RAM voltage, wouldn't that mean that under high load, the calibration will raise that voltage EVEN MORE? I wasn't able to get a stable overclock on the CPU unless I turned LLC on extreme, that's true. My workers would keep crashing in Prime95. In terms of PSU replacement, I could just get a Corsair 550W gold or something, I might consider it. The RAM only clocks max to 1600MHz, and again, I chose a default profile from the motherboard for memory OC that set the timings and voltage to 1.5V. Maybe I should do 1.6V like you said.

P.S. now please excuse me, this thread will get flooded w/ pictures because that's what chrisjames61 wants.
 
#14 ·
Quote:
Originally Posted by 1216 View Post

Raise vcore. Find LLC setting that keeps it stable
imo medium llc for the gigabyte board is the best for OCing. Of course you will need to raise the voltage a bit most likely too.

could also drop the multiplier of ht link 1 and run it at 2400.

and i would always recommend raising nb multiplier to 11 or 12 (2200 or 2400) and the nb voltage to 1.3
 
#15 ·
Thank you all for the input, I'll put it into practice first chance I get (school is painful). But like I said, I adjusted Vcore and LLC until I got a stable CPU overclock. My CPU overclock is STABLE, and like I mentioned before, the CPU itself was tested without any crashes. What's possibly happening is communication is suffering between the CPU and the RAM. Could be RAM voltage needing an extra 0.1V. But from what everyone here is saying, the NB might need some tweaking. CPU alone is fine. I raised the Vcore in an attempt to see if CPU clock was still unstable, and I still blue-screened, so the CPU as an issue can be checked off as not a possibility.

For now I increased RAM from 1.5V to 1.6V, let's see if situation improves at all.

OH and what's the difference between NB Voltage and NB Core Voltage? As you can see in the pictures, one is currently set to 1.1V and the other to 1.2V from stock. Which one should I increase if I clock the NB higher?
 
#16 ·
Quote:
Originally Posted by Snoopi25 View Post

Thank you all for the input, I'll put it into practice first chance I get (school is painful). But like I said, I adjusted Vcore and LLC until I got a stable CPU overclock. My CPU overclock is STABLE
Quote:
Originally Posted by 1216 View Post

Adjust Load Line Calibration. Having this on Extreme causes your voltage to be a-ok under full load but under partial/no load conditions it will be lower than sufficient. This explains the bluescreens you're having.
Please get HWinfo64 cpu vcore values at idle and under heavy load (prime, ibt) and tell us.

And NB Core is cpu memory controller (+L3?)
NB is chipset
 
#17 ·
I used HWMonitor and CPU-Z for this. Here's what happens when I just run the 3 different tests of Prime95. Granted, numbers fluctuate as the CPU is doing other things during test, they either bump up or drop down, I tried to find a stable place to take screenshots. As you can see also from the HWMonitor, the "CPU" temp that's at like 65C is actually the north bridge. Funny thing is, that thing will run at that temp whether it's idle or during load. Sometimes it drops down to like 40C, again, idle or load, it's really random. It's not related to my room temp, that's for sure. I also included an idle pic (first one).




 
#18 ·
Is your cpu throttling? Disable APM and see if it's still dropping to 3 GHz

You have set 1.375 as your vcore in bios while MAX is showing 1.452. Extreme LLC isn't doing you any favors. Try 40 or 60 percent and see how it works out. You'll probably need to raise Vcore a bit but not to 1.45. Try 1.4, see how prime runs. You're looking for minimal vcore fluctuation
 
#19 ·
yea you really want to stop using extreme LLC on that board and retune it in with medium llc, im on the 990fx gaming gigabyte board which is pretty similar and have had far better results this way.

you want the NB voltage raised from 1.1-1.3 and the frequency to 2200. again i might try dropping HT link to 2400 if everything else fails and see where that lands you on the blue screens.

also as 1216 said above you need apm off, i think it drops the frequency whenever you go above a certain power draw. even putting you below stock clocks.

can you pass prime 95 blend test for lets say 30min without any cores failing, can you pass prime 95 small FFTs for 30min without throttling or failing on any cores? run hw monitor in background so we can see clocks, temps, and voltages from start to end of test. (how long were the tests ran in previous pics?)

you say the cpu OC it is stable, but obviously something is causing instability.
 
#20 ·
Like I mentioned when starting the thread, just using the settings in the BIOS pictures I provided, I ran Prime95 heat test for 30 mins and was fine. I DECRYPTED (that's brute force 100% CPU load) some files for 10 hours at a time for 2 days, and I did NOT crash at ALL, which is very interesting if you ask me. I will try those settings you guys suggested as soon as I get a chance to. It might be a couple of days. The pictures that I threw up with the HWMonitor right now and those tests, I just started test, took screenshot, stopped test. Just to see how high voltage and clocks go. Where is APM? I'm blind lol... I believe the stability issues arise from some link somewhere, either the NB, or something else, because that's the only component that remained at stock V's and clocks. Back when I did the overclock, was not able to get a stable OC without LLC at 100% though, I don't know why. I think it was the only option that would give the CPU enough juice w/out having high voltage. At least, that's what I thought at the time.
 
#22 ·
Hi There,

Try this.... I have a similar board....

Make a FULL CLEAR CMOS in the motherboard... Change the CLEAR jumper and turn on the PC... Nothing will happen.... Put back the jumper in normal mode...

After that enter BIOS and setup the basic stuff.. Year, date, hours.... Dont mess with the BIOS at this point... We need the PC stock....

Than go to "Voltage Control" and set you DDR3 voltage to 1.65v.

Save the changes and use the PC...

Good Luck !
 
#23 ·
Ok, so I went ahead and initially only increased my RAM voltage to 1.6 instead of 1.5... things went pretty well, there was a very minuscule boost in performance (in terms of how snappy things were), but I just blue-screened again while having multiple tabs open in Chrome, doing homework. Funny thing is, I didn't blue-screen yesterday, and whole day I was gaming.

I went ahead and applied all the other changes, set LLC to medium, changed VCore to 1.4, changed NB frequency to 2200, dropped HT link to 2400, and changed North bridge voltage to 1.3 from 1.1. I also disabled APM and SVM... PC booted, and everything seemed really snappy, like, opening windows, applications, etc... I decided to open Prime 95, and HWMonitor and CPUz... My CPU temp shot up from idle to 50C in like 2 seconds, and then one of the workers stopped working... What should I try to fix it?

 
#24 ·
UPDATE: So, I just went ahead and changed some settings again. I set the Vcore back to 1.375 like it was before, re-enabled APM, and set LLC to Extreme again. Ran Prime 95 for about a minute, and workers didn't stop this time. I CANNOT run this overclock w/out LLC on extreme, workers will stop during testing. I have had this issue when I first started to overclock this CPU and when I messed around only w/ the CPU, no other settings. This is the CPU-Z and HWMonitor results:

 
#25 ·
Quote:
Originally Posted by Snoopi25 View Post

UPDATE: So, I just went ahead and changed some settings again. I set the Vcore back to 1.375 like it was before, re-enabled APM, and set LLC to Extreme again. Ran Prime 95 for about a minute, and workers didn't stop this time. I CANNOT run this overclock w/out LLC on extreme, workers will stop during testing. I have had this issue when I first started to overclock this CPU and when I messed around only w/ the CPU, no other settings. This is the CPU-Z and HWMonitor results:

that just means your voltage is too low, and looks like your cooler is not up to the task of the clocks you want.

unfortunately imo your overclock with APM on is going to be worse than running stock as your min frames are gonna take a big hit every time the multiplier drops.

also 50c is well within the safe range, though i am interested where it would hit with the voltage raised to where it needs to be stable after running for 15-30min.
 
#26 ·
Ok, this is NOT good... I left the computer on sleep mode for a while, and when I came back and tried to turn it on, the tower turned on, but didn't get any signal. The PC then went through a series of on, off, on, off by itself, and I decided to just press the Reset button. It rebooted, and I went to the BIOS and wanted to change the APM to off again. Did that, saved, entered OS, tried putting it on sleep. Tried waking it up again, same thing. This time, I got a screen that said UEFI Dual Bios and the PC rebooted. Upon reboot, it resumed windows (strange if u ask me), and then I wanted to try it again. This time, when I tried waking it up, it had a long sequence of really fast beeps, and then rebooted. So I freaked out and put all the BIOS settings to how they were prior to this thread opening lol. I'm either doing something wrong, or I ALWAYS, and I mean ALWAYS get the cheapest chips that aren't OC-able... like, I was never able to get that good chip, always had the runts.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top