Self Mutilation, can you fix it?

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0
Topic 195531

Hello everyone, I just put a new video card into my machine

Im running an Nvidia GTX570 here and tweaking with the settings a little bit to see what does what to help or whatnot.

anyways my question is... well I have two actually.

#1 IF it set something wrong the card will lock up for a few seconds, reset itself to I guess you could call it, factory default levels and recovers and continues right on, and I get the little windows message saying it just did so. The problem is, any computations that were in the process immediately get killed with a computational error message and it starts a new one. My question is, is there any way to resurrect those if I can catch them before the get to the server. Id rather not be killing a ton of apps having them cancel out like that. Can I reset the project or something to start them from scratch or does reset do all the einsteins going on the machine?

This leads me to my next question. I finally got einstein to feed me some binary data to gpu crunch and the card is barely moving. The milky way data ran the thing at like 99 percent processor useage, where the einstein data has it in the upper teens / lower 20's. There's tons of viv memory left too. IS there any way to get it to utilize the card more? Am I missing something here or is that all it's going to do?

Thanks
Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Jord
Joined: 26 Jan 05
Posts: 2952
Credit: 5893653
RAC: 4

Self Mutilation, can you fix it?

Quote:
My question is, is there any way to resurrect those if I can catch them before the get to the server. Id rather not be killing a ton of apps having them cancel out like that.


No, once a task has run its course and it's either completed, or in error, it'll stay having that outcome.

Quote:
Can I reset the project or something to start them from scratch or does reset do all the einsteins going on the machine?


Resetting the project will cause all present tasks to be erased from the machine, yet with Einstein's resend function on, it's possible you'll get them all again. But a bit overkill, in this case.

BOINC has the redundancy built-in that if something goes wrong with work on your machine, that it'll eventually get sent out to another machine.

Quote:
IS there any way to get it to utilize the card more?


CUDA BRPs are still being tested. Only when the developers figure that whatever data was sent back is good enough for them, may this specific app become mainstream. Until then, let's just wait it out.

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0

thanks for the answer. I was

thanks for the answer. I was hoping I could reset just 'one' task on my machine since I know the fault was caused directly by me, and not programming or something of that nature. I just didn't want the folks seeing a ton of rejects coming back at them and thinking its a programming issue when it was me trying to figure out how to get my card working at it's best.

Ill leave em run as is then.

Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.