After many BSOD ive confirmed that attempting to restart a workunit causes an instant BSOD. Ive tried even starting just one WU to check if it might be a powering up to much at once issue and it still causes a BSOD. Can anyone give me a hint as to why this may be and any solution as not being ables to suspend work so i can use the GPU's is a major hassle.

My setup is as follows
Window Vista Ultimate 64bit
Nvidia Drivers (happens with both) 182.50, 185.68
BOINC 6.6.20
2X GTX 295's
This Host

Yell if you need any more info to debug.

Can anyone with Vista 64 try to reproduce this? (and post your driver version and GPUs)

MrS

---

Scanning for our furry friends since Jan 2002

I am very close to same setup but I only have 1 GTX295, I'll be back in a bit and edit this post with my results.

OS: Vista Ultimate 64
BOINC: 6.6.20
GPU: EVGA GTX 295
DRIVER: 182.50

{edit}
Works without BSOD for me. Tried a couple of different scenarios ...
Suspended tasks individually / Resumed: Pass
Suspended project / Resumed: Pass
Suspended GPUGrid and WCG project / Resumed: Pass
Suspended left in memory / Resumed: Pass
Suspended took out of memory / Resumed: Pass

What I didn't do was after each type of suspension I did not run an intensive game / stress test/ benchmark.

Steve

It seems to have stopped doing it now im really at a loss. I think i have gremlins as it works better now i have upped the overclock.... go figure. Stable over 24hrs this time.

Current overclock
Core clock 690
Shaders 1490
Memory 1290

had more problems with the simple factory overclock that was considerably slower.

to go a little OT ...

I have read in a couple of places that the bang for your buck on OC is, in order of performance increase, shader then core and finally memory which has very little effect. My current OC, 48 hrs stable at stock voltage, is 642/1554/1054 fans at 60% with GPU temps 70c/69c, ambient is 22c. This card is not a factory OC.
Steve

Scratch that it may be more stable now but a resume of GPUGrid still kills things. Stopping and starting my other main project ABC@Home does not.

On the Bluescreen i get

STOP: 0x0000007E

dxgkrnl.sys

plus a lot of hex numbers if that helps any. I have to rebuild the system in a new tower when i get time. Ill try single cards to see if i may have a slightly iffy card causing the issue.

to go a little OT ...

I have read in a couple of places that the bang for your buck on OC is, in order of performance increase, shader then core and finally memory which has very little effect. My current OC, 48 hrs stable at stock voltage, is 642/1554/1054 fans at 60% with GPU temps 70c/69c, ambient is 22c. This card is not a factory OC.
Steve

My current OC is now 700/1500/1300 at 80% fan GPU is sitting around 68-69c, ambient is high 50c's.
The case is badly overcrowded and airflow very restricted. The side isnt even on i really need to move this into my new case as the system does not fit in my current case. Most of the cables hang out the side and above and below the GPU fans.

Thanks for the tests, crash! I think there's no need for anyone else to test this.

@Clownius: although you already tried powering up just one WU you could try to take out on eof the cards completely (and thus reduce base power draw) and try again.

MrS

---

Scanning for our furry friends since Jan 2002

That will happen when i move the system into its new tower. Ill also confirm the cards are on completely separate rails. Im looking at anything and everything now.

The PSU should be ok its a 4 rail 1500W. If that cant handle it nothing can i may even need to go dual PSU (my new case supports that).

Well, that should be enough raw power for 4 of these 295 babes :D

MrS

---

Scanning for our furry friends since Jan 2002

I was suffering from the same BSOD's as well recently, and also the same issue of BOINC never shutting down correctly on exit. 6.6.36 appears to have corrected both issues so far.