New overclocker looking for help (BSOD)

Kageromero

Reputable
Mar 9, 2014
22
0
4,510
Two days ago I overclocked my 4770k to 4.5, first time overclocking anything, ran a stress test for an hour, perfectly fine, max temps at 80c, no issues, went the day gaming and it performed great and never crashed.

And then there is tonight...it seems every 30 minutes of usage it'll crash or BSOD, gaming or not. Stability test lasted 30 minutes till I got out of bed and noticed it had crashed. A few things to note I think may be important

My cpu voltage is currently set to 1.275, but according to AIDA64, has been going between 1.274-1.278 during the stress test, not being consistent. My OS is currently on a SSHD which has been laggy and horribly slow ever since I upgraded from my i7 950, was never sure if it was a drive issue, mobo issue, or cpu issue as all were new, but seeing as my games all played with the expected fps, I don't think there was an issue with the cpu itself. My powersupply is a few years old now, so not really sure how great it's doing. Another thing to note is that when I first put my computer together, it was so slow on boot and starting up, I thought something had to be wrong and took it over to the nearest canada computers to check out, they found nothing wrong, I brought it back, and it still had all the same slowness issues, so I brought it back, they turned in on there and it started up incredibly fast, while still staying slow back when I brought it home. I have also tried increasing the voltage up to 1.29, and while temperatures were fine, the crashing persisted.

If anyone could help or provide guidance with anything listed here, it would be greatly appreciated.
 
Solution


if your running your ram at default 1333mhz then you shouldnt need to adjust any voltages since the qpi/vtt should be at 1.05v and the ram is at 1.50v. they are within 0.50v of each other. if your 1600mhz setting is an xmp profile and it is using 1.65v dram then the qpi/vtt should be at 1.15v at least and up to 1.20v. but, again, some have...
Using a asus Maximus VI formula motherboard. I have tried the asus suite in the past, I reset default settings and had it uninstalled before I did this manual overclock. I'm not set to xmp, but I will go lower my ram back down to 1333, was just thinking of that as well. Did another stress test just now at 4.4ghz but without stress testing the memory and it went pretty solid.

Edit: Just lowered the memory back to down 1333, had a very quick and speedy start up, my hopes are high! Time to stress test...
Also, from what I've seen in my bios, I can disable C states, but I can't find c3/6 specifically, is it just disabling that? And from what I've seen in that link the other person posted, since I am in manual mode shouldn't having it set to auto be okay?

Also, what if I left my memory at 1600mhz, but gave it an extra .5v?
 


you can likely still run your ram at 1600 if your passing stress test. but if you want to find a real stable overclock you should leave EVERYTHING at stock, disable c3/c6, disable c1e and speedstep or eist for now, and keep moving your multiplier/ratio up until you cannot pass a stress test. this way you can find your stock svid vcore limited overclocked which should always be rock stable since it was tested at the factory. usually its modestly decent, 4.1-4.5ghz depending on sandy/ivy/haswell. then you can start messing with voltages and other settings to push it up more as long as your still within the thermal envelope.

you can re-enable all the power savings after since figuring out vcore is the main goal for maunal overclocking and you dont want ultra low power states messing with your vcore. if your unstable after power savings, your likely close and need just a few minor voltage adjustments to be stable.
 
im not sure if it still applies to haswell, but dram voltage is supposed to be within 0.5v of the qpi/vtt voltage(vccio). at least for sandy bridge, the stock qpi/vtt is 1.05 which allows 1.50v dram dimms to run fine. many people are on the fence about using 1.65v dimms as it requires the qpi/vtt to be pushed up to 1.15-1.20v, which is past the intel recommended 1.08v. many have ran 1.20v qpi/vtt, myself included, and haven't had too much problems. but some believe this can burn up the intel memory controller. my ram wants a 1.65v dram with the intel xmp model, but i use 1.50v and slightly looser timings as 1.65v was messing with my overclock and adding 1-2c to cpu temps.

for these reasons, you should leave the memory at super slow 1333 and stock voltages(1.50, or 1.35v if you have low voltage dimms) so you can rule out qpi/vtt or dram messing with the stability of your cpu overclock when initially finding your cpu limits.
 
Alright sorry for the late replies, had a busy really day. Ran a stress test last night with cpu at 1.28v and 4.4ghz, ram at 1333 and gave it the extra .5v, and disabled c3/c6. Need to run a longer test tonight, but last night it ran for 2 hours without crashing, temps peaking above 80 a few times, but very rarely and then quickly dropping back to mid 60's so think it may finally be stable!
 


if your running your ram at default 1333mhz then you shouldnt need to adjust any voltages since the qpi/vtt should be at 1.05v and the ram is at 1.50v. they are within 0.50v of each other. if your 1600mhz setting is an xmp profile and it is using 1.65v dram then the qpi/vtt should be at 1.15v at least and up to 1.20v. but, again, some have had issues with 1.65v dram affecting overclocks and or higher than 1.08v qpi/vtt possibly being the culprit for burning up the memory controller.

though my ram is rated for 1.65v, i run it at 1.50v@1600 with slightly looser timings of 9,10,9,28,1t

you say 1.28v. did you manually adjust that or is that what the svid is automatically putting the vcore to if you leave vcore on auto with the 44x muli/ratio?

passing 80c? what cpu cooler do you have? how well was the thermal paste applied? what ram sticks do you have?
 
Solution