Sign in with
Sign up | Sign in
Your question

2600k can OC to 4.5ghz, but BSODs when using the integrated GPU

Last response: in Overclocking
Share
December 28, 2012 7:39:58 PM

I've had this 2600k for a year and a half, and since day 1 I've had it at 4.5GHz with a +0.150V overvolt. I recently got a copy of Action!, a piece of game recording software that takes advantage of my 2600k's integrated GPU to offload the task of encoding H.264 video away from the CPU.

Only problem is, when I record use this, I eventually get a BSOD (something about a secondary processor not receiving an interrupt at the proper interval). I've been timing how long it takes from record start time to the BSOD, and it seems to correlate to how much overvoltage (more volts = lasts longer) and overclocking (less overclocking = lasts longer) I've done to the CPU. My motherboard (Asrock z68 Extreme4) has an option for iGPU overvolting (up to +100mV), which doesn't seem to have too much of an effect on BSOD times. I've gone as high as 4.5GHz, +0.200V, which brought the temps up to the mid 70s. I haven't tried stock clocks yet, even at 3.8GHz my FPS dips unacceptably low.

Is there anything else I could do? Am I just unlucky and have a 2600k with an unoverclockable iGPU?
a b K Overclocking
December 29, 2012 4:26:20 AM

You know, I'm not absolutely positive how that works, but I know the iGPU has to be correlated to the main processor somehow. Seeing as you're adding voltage and it's staying stable longer. I haven't used the iGPU on my chip so I couldn't tell you, but I know that at 4.5Ghz I was stable at +0.020v on offset. I run at 4.4Ghz with -0.025v offset and never have any problems. Seeing as you're so close to my chips uphill voltage wall, maybe backing down to 4.3Ghz and running the overvolt you had would keep it stable? Can you downlock the iGPU at all without it BSODing?

I'd try a few different things, but running at a +0.2v overvolt isn't one I'd run for long.
December 29, 2012 4:37:16 AM

I've done 4.0GHz, at a +0.200V overvolt, and it lasted about 45 minutes before the BSOD. I don't see any options for setting the clock of the iGPU separately from the main ratio...
Related resources
a b K Overclocking
December 29, 2012 4:47:35 AM

That's weird as couple be. You certainly have sparked my interest here. I'm going to do some investigating. But while I do, I want you to try something for me.

Drop down clear to stock settings. Run the same setup and see what happens. See if the iGPU still gives you fits like that. Like I said I don't have a lot of experience tinkering with the intel iGPU like I have with a few A4's... But I'd imagine they run similarly. I know the iGPU will add more heat to that coolers workload but it shouldn't be that much. I wonder if the iGPU voltage is just bounced off of the main voltage going into the CPU? Either way, see what happens when you're at stock. If it still gives you a fuss it could be a few things other than the CPU/iGPU but if it runs right; then there is an issue there somewhere and we'll have to track it down.

It may be that running the iGPU that hard while overclocked is just going to cause instability between the iGPU and CPU. Kind of like when you overclock an FSB and the northbridge just isn't stable enough to handle the clock when the processor is.
a b K Overclocking
December 31, 2012 5:26:01 PM

You may need to adjust the frequency and timings of your RAM, iGPU performance has a lot to do with system memory
December 31, 2012 8:29:56 PM

I just got the opportunity to try recording with stock clocks and voltage. I managed to make it almost 2 hours until it gave me another BSOD. A major improvement, but it still crashes. Do you guys think that perhaps it's a driver issue? Or since it seems to correlate a lot with the clock and voltage, it's an overclock issue?

@hdeezie80, I'm running 4 sticks of G.Skill DDR3-1600MHz ram (4x4GB). They're properly set to the right frequency, voltage and timing in the UEFI. Could this be a sign they're starting to malfunction?
a b K Overclocking
January 2, 2013 9:06:52 PM

You can thread that out by running memtest386 on each of the sticks for a while and see if there is one giving you an issue.

I'd definitely try updating the drivers, but if ti's crashing at stock as well; there's something wrong somewhere. Could it be the actual application having an issue and causing a crash?
!