BSODs when playing video while OCed

I

Ivris

n00b
Joined
Sep 28, 2007
Messages
14
I may have infodumped a bit too much in my original post. In any case, I'm still looking for answers concerning an irregular BSOD that I've been getting with my overclock, and wonder if anyone knows what's going on.

As mentioned before, my system's a Q6600 (B3) on an eVGA 680i (A1) mobo. Overclock is set to 9/333 and RAM is synced to it at 666. Cooling is a Freezer 7 Pro in an Antec P180B case with an extra fan blowing in on the Northbridge heatsink and video card. With ambient of 20 after idling for half an hour, core temps are 48, 43, 38, 39, and are 72, 69, 67, 67 after 10 minutes of Prime95 (according to SpeedFan 4.33 +15 degrees).

Further tests have given me a bit more information. I can Prime95 for at least 36 hours and Memtest86+ for the same amount of time without error. If I attempt to start any sort of multimedia-related task (movie playback, for instance), however, I randomly BSOD with a "Hardware Malfunction" error (roughly 1 in 15 times). Games run fine except for certain FMVs which give the same problem (the Bioshock opening is one such video).

Unlinking RAM and setting it to run at 800 MHz greatly reduces the frequency of errors (down to about 1 in 45), but they still occur; relaxing RAM timings doesn't appear to help, as whether at 4-4-4-12 or 5-5-5-15, the problem occurs with about the same frequency. Upping voltages also shows no change; I took vCore up to 1.45 and RAM up to 2.2 with no effect before reverting changes. Undoing the overclock resolves the issue (or at least reduces it below detection).

Given the limited circumstances these BSODs occur in, does anyone have an idea of what component might be at fault?
 
I'm assuming when you say you P95'ed it you stressed all 4 cores. Based on that it's a bit of a conundrum. I'm no expert but it may be the north bridge. If it seems to be running hot, maybe put some active cooling on it, if it's cool possibly up the voltage?
 
Just for completeness what is your graphics setup. i.e. what card and are you overclocking it at all?
 
The video card is an EVGA 8800 GTS 320 MB at stock speeds. I wanted to sort out the CPU overclock before moving on to the video card.

All drivers are, of course, completely up-to-date.
 
have you tested to see if it occurs while stock clocked?

I'm wondering if you are actually suffering from an overclock issue, or something just software related.
 
Slade said:
have you tested to see if it occurs while stock clocked?

I'm wondering if you are actually suffering from an overclock issue, or something just software related.
Click to expand...

Undoing the overclock resolves the issue (or at least reduces it below detection).
Click to expand...

In other words, I reverted to stock clocks and no longer had the issue, making me pretty sure it's overclock related. I left a video file on constant loop while I was out today with the system set back to stock as a test and came back to find it still chugging along, so I'm pretty sure now that it's related to the OC.
 
check to see if your pci-e is locked to 100, it might be that you are overclocking the pci-e and subsequently any pci devices. Those wouldn't be stressed during a prime/memtest run, but would be unstable in everything else related to them.

If that's not the case, I suggest a FSB and NB voltage increase.. other than that, maybe it's a fubar'ed windows install... can't see something passing prime/memtest for 36 hours with all recent drivers for mainstream products randomly BSOD...

With video also, I'd check what codecs you have installed. Try installing a codec pack like K-Lite (but don't install ANYTHING FFD related).
 
PCIe is locked to 100. I found that Auto volts on the Northbridge was setting it to 1.20 even after OC, so I bumped it to 1.35 and am giving that a try now. Auto volts on the FSB set it to 1.4, and setting it any lower resulted in crashes before boot completed, so I left it there. Should I adjust HT or Southbridge voltages?

For codecs, I've been using CCCP, but with FFDShow replaced by separately installed Xvid and h.264 codecs. This was because I didn't like how FFDShow kept stealing FourCC associations, but I didn't know there were any issues with OCing.

Also, how are the temps looking? I've heard that B3 stepping runs about 10 degrees hotter than G0 for a given clockspeed and my numbers seem in line with that, but I've also heard that the Freezer 7 Pro isn't exactly the best HSF for Quads.
 
although your temps are on the high side, if you are prime stable at those temps for all those hours, then I wouldn't worry too much short of having a hot room. A video wouldn't stress anything cpu/mem wise.

With that said the HT and the southbridge could be potential candidates for oc issues, yet they are remote, I'd play with the HT a bit, and if that makes no difference then the sb, although that one is remote if it could do anything bad for an oc.
 
Tried a variety of NB and HT overvolt settings last night -- nothing was stable enough to last for over an hour of video playback (can't say anything for Prime, though as the volts are higher, I'd expect it to last like before but with a temperature cost).

I can't even scale the OC back to a more modest level -- FSB of 1200 instead of 1333, for instance, results in a BSOD before hitting Windows, even on Auto volts for all components (save RAM and NB, which I've bumped up myself). FSB 1100 wouldn't even POST and had to be reset, while 1300 crashed on login. RAM timings were relaxed to 5-5-5-15 for the duration of these tests. My only stable options appear to be 1066 and 1333, with nothing in between, and 1333 gives the video playback problem despite being otherwise rock solid.

It probably would have helped if Windows gave me more information than just "Hardware Malfunction." Is there a chance that the CPU is at fault here?
 
something's not right with either your motherboard settings

You aren't stressing the board much at a 333 setting, and while your temps are high if prime and memtest are not crashing, then I can only suspect the video card or your sound card is acting up at those settings. Try disabling your sound device and run the video. I'd also monitor your voltage readings for any spiking out of spec prior to the BSOD.
 
Got it, I think (three hours stable with video on constant loop and counting, where the farthest it ever got before was one and a half). I had to max out the FSB volts at 1.5 AND bump both the HT and NB voltages a few notches to get it stable. My final voltages are:

vCore: 1.3625 (shows as 1.31 in SpeedFan and CPU-Z)
vDIMM: 2.1
FSB: 1.50
HT: 1.35
PCI-E/NB: 1.35
nForce/SB: 1.5 (Auto)

This change has resulted in a 3-4 degree increase in idle temperatures (again, after a half-hour of idling), but surprisingly, a tiny drop in load temps (reading below 70 degrees across the board after 10 minutes of Prime). I might be able to drop the vCore and maybe RAM voltages a bit more; I have to Prime this a few more hours to see if this configuration holds for non-video tasks.

I was hesitant to push FSB past 1.4 for such a modest overclock, but it seems to have done the trick (1.4 with the NB and HT overvolts was still BSODing, and 1.5 without the other overvolts did the same). Are there any major consequences for burying the needle on the FSB volts like this?
 
You must log in or register to reply here.
Back
Top