igdkmd64/i3 540 crashing widows 7 x64, no bsod but occastionally a .dmp file

[idata]

I am getting a crash when playing video, particularly full screen video, on my i3 540 HTPC. Usually there is no .dmp file, but I have gotten 3 of them from the daily crashes in LiveKernelReports. The problem is semi-intermittent, but nearly exclusively happens when playing video. Out of the 50 or so crashes, 1 of them happened at desktop.

When playing the video, the system will freeze. The display will present with ~14 columns, alternating between clean display and garbled. The garbled columns appear like stacked = signs, about 20px high each, with alternating garbage and clean display (iirc.). Anyway, it always presents the same way, leading me to believe it is a video issue. Also, on reboot the machine needs to get through post with video off, then rebooted again to get video out to work. This issue is the same using dvi=>hdmi or hdmi, but hdmi is even touchier.

Memtest x86 runs overnight, far more than 10 passes. Win7 mem check is fine. Running distributed.net 4 threads for a day causes no issues, even when running furmark at the same time. sfc /scannow returns clean. All power saving options have been disabled.

i3 540 stock

asus p7h55m-pro

toshiba 46" lcd - 1080p

4gb ddr3 mushkin, 600W ocz psu,1 wd hdd, 1 dvd

The 3 minidumps all point to igdkmd64.sys. I have them to attach, but can not in this forum software. They look sorta like this, but this is from the previous (2086 build) if I am not mistaken. I have not had a crash produce a .dmp file since then, it usually goes down too fast.

Probably caused by : igdkmd64.sys ( igdkmd64+1cc60 )

!analyze -v

VIDEO_TDR_TIMEOUT_DETECTED (117)

Arg1: fffffa8006a184e0, Optional pointer to internal TDR recovery context (TDR_RECOVERY_CONTEXT).

Arg2: fffff88004833c60, The pointer into responsible device driver module (e.g owner tag). Arg3: 0000000000000000, The secondary driver specific bucketing key. Arg4: 0000000000000000, Optional internal context dependent data.

FAULTING_MODULE: fffff80002c64000 nt

DEBUG_FLR_IMAGE_TIMESTAMP: 4bb3d0e5

FAULTING_IP: igdkmd64+1cc60

DEFAULT_BUCKET_ID: GRAPHICS_DRIVER_TDR_TIMEOUT

TAG_NOT_DEFINED_202b: *** Unknown TAG in analysis list 202b

STACK_TEXT:

fffff880`053b8630 fffffa80`0618a210 : fffff880`040b88a0 fffff8a0`02e21d19 fffffa80`03f48c10 fffff880`04833c60 : watchdog+0xa577 fffff880`053b8638 fffff880`040b88a0 : fffff8a0`02e21d19 fffffa80`03f48c10 fffff880`04833c60 00000000`00000000 : 0xfffffa80`0618a210 fffff880`053b8640 fffff8a0`02e21d19 : fffffa80`03f48c10 fffff880`04833c60 00000000`00000000 00000000`00000000 : dxgkrnl+0x1f8a0 fffff880`053b8648 fffffa80`03f48c10 : fffff880`04833c60 00000000`00000000 00000000`00000000 fffffa80`000009d0 : 0xfffff8a0`02e21d19 fffff880`053b8650 fffff880`04833c60 : 00000000`00000000 00000000`00000000 fffffa80`000009d0 00000000`00000000 : 0xfffffa80`03f48c10 fffff880`053b8658 00000000`00000000 : 00000000`00000000 fffffa80`000009d0 00000000`00000000 00000000`00000000 : igdkmd64+0x1cc60

0: kd> lmvm igdkmd64

start end module name fffff880`04817000 fffff880`051ef3a0 igdkmd64 T (no symbols) Loaded symbol image file: igdkmd64.sys Image path: igdkmd64.sys Image name: igdkmd64.sys Timestamp: Wed Mar 31 15:47:01 2010 (4BB3D0E5) CheckSum: 009DC368 ImageSize: 009D83A0 Translations: 0000.04b0 0000.04e4 0409.04b0 0409.04e4

[idata]

I think it a video driver problem try updating the vide card drivers and mainboard drivers.

Then downoad this OCCT.

http://www.ocbase.com/perestroika_en/index.php?2009/05/21/39-occt-perestroika-310

Do a gpu bench and powersuply bench.

This error can also caused by overheating, the graphic card tend to lockup for a short period of time but enough to trigger this error dump file.

Good luck

[idata]

All drivers are up to date already, by manual inspection and according to the intel software update scanning tool. The latest video is 2104, which I have. Chipset & video bios are all current. Bios is flashed to latest. All the basic bases have been covered.

I doubt it is a heat issue, since I can:

1. Run 4 threads of distrubited.net (max 100% on both cores, as well as the 2 fake HT cores)
2. Run http://www.ozone3d.net/benchmarks/fur/ FurMark at the same time, stressing the GPU directly
3. Do this for 24 hours, in the same enviroment
4. Voltages all look fine, no spikes or drops according to HW monitor and several other monitoring tools.

I will look into OCCT again, but I didn't like it last time I used it. I recall most of the test tabs are disabled on my system, probably it doesn't know what to do with the new chips. Hopefully that has changed.

What would be MOST helpful is figuring out how to get the win7 to give me a dump file in all cases when this kind of error happens, or at least increase the likelyhood of getting one significantly.

[idata]

Still crashing in exactly the same manner. Still not writing .dmp files.

Does anyone have any insight? Is there an actual support line, email or otherwise, to contact about this? Anyone from intel out there listening? ~80K possible pairs of eyeballs, is anyone watching this forum?