Forum > GPU crunching

Latest drivers (NVidia and ATI)

<< < (21/167) > >>

Ghost0210:
could be the higher end cards that this is affecting then
I run 2 GT240's on the 196.75 drivers and the opt apps and so far have had no problems
I've seen it mentioned here in another thread that its possible that the newer drivers work ok for the 240's
both my temps and fan speeds are normal with no crashes

HSchmirPo:
Yes, higher end cards.
I have a 285, 250, and a 8800 GTX, all with optimezed apps. None of them shows VGA restarts.
So I think I have to switch to the stockversion on the second 295 card.

Pepi:
Reading this thread, I am little confused.
Please tell me what will be fastest combination on my GT 9800 Green ( cuda 2.3)
Some says that optimal is 190.38, some other says 190.62 and also 191.07. It is clear that 195.62 is not fastest.
Thanks for reply

RottenMutt:

--- Quote from: Ghost0210 on 06 Mar 2010, 01:15:14 pm ---I used to get the crashes as well especially on the 195 drivers.
Went back to the 191.07 drivers and problem solved
Thought I'd try thee 196.75 drivers as there seems to have been some stability work done by nVidia on this set - had them since thrusday and no crash yet - also reported problem of fan speed issues hasn't affected me
No improvement or reduction of speed either

--- End quote ---
are you running quad sli or higher?  i've got three gtx 295's, one in one rig and two in another, and am only able to get two optimized apps to run without what i call thrashing (invalid results and nvidia driver restarts).
can you get four stock gpu apps to run?
how much slower is the stock app?
i was able to get the quad sli rig to run three apps, but you could feel it trying to leave the kernel (not responding).
don't put your rig to sleep with the 196.75 drivers, this caused my fans not to run.

Raistmer:
you could try latest opt CUDA MB build, x4 one, for your multy-GPU host. Maybe it will achieve more stable results (all GPUs could be loaded with maybe w/o crashes)

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Go to full version