Using Tesla as secondary card

Moderators: slegrand, Site Moderators, PandeGroup

Using Tesla as secondary card

Postby TimStaley » Wed Sep 10, 2008 5:46 pm

Hi there,
first time posting here:

I'm trying to setup FAH GPU 6.20 console version on an XP-32bit machine. However, as far as I can make out it is impossible to use a Tesla C870 card? (It's a work machine and it might as well do something when not in use)

Here's the thing - my primary card (to run the display, Tesla cards have no video output!) is a geforce 7300 - so no CUDA support. Fine I thought, read the FAQ. Ran FAh... -gpu 1 -forcegpu nvidia_g80 ... all started up ok.
However as soon as the core is loaded I immediately get " mdrun_gpu returned -1 / UNSTABLE MACHINE " EUEs. Reading these forums suggests that to fix the bug you must connect a monitor to the secondary card, but like I mentioned already the Tesla has no video output.

Anyone figured out a way around this bug, by chance?

Cheers,
T.
TimStaley
 
Posts: 1
Joined: Wed Sep 10, 2008 5:36 pm

Re: Using Tesla as secondary card

Postby Hom3br3w » Thu Sep 11, 2008 1:43 pm

My understanding is that -gpu 1 will try to find the second CUDA available video card try -gpu 0 as it should detect the first Cuda card (your tesla card as the 7300 has no cuda)
Hom3br3w
 
Posts: 3
Joined: Tue Sep 09, 2008 11:12 am

Re: Using Tesla as secondary card

Postby shdbcamping » Thu Nov 20, 2008 8:45 am

TimStaley wrote:Hi there,
first time posting here:

I'm trying to setup FAH GPU 6.20 console version on an XP-32bit machine. However, as far as I can make out it is impossible to use a Tesla C870 card? (It's a work machine and it might as well do something when not in use)

Here's the thing - my primary card (to run the display, Tesla cards have no video output!) is a geforce 7300 - so no CUDA support. Fine I thought, read the FAQ. Ran FAh... -gpu 1 -forcegpu nvidia_g80 ... all started up ok.
However as soon as the core is loaded I immediately get " mdrun_gpu returned -1 / UNSTABLE MACHINE " EUEs. Reading these forums suggests that to fix the bug you must connect a monitor to the secondary card, but like I mentioned already the Tesla has no video output.

Anyone figured out a way around this bug, by chance?

Cheers,
T.

Did you get the Tesla card running ? I'm trying to find info to see if getting one for my 2nd PCIe slot would be worth the money.
I would appreciate any info you could provide.
Sean
shdbcamping
 
Posts: 519
Joined: Mon Nov 10, 2008 7:57 am

Re: Using Tesla as secondary card

Postby MtM » Thu Nov 20, 2008 9:02 am

TimStaley wrote:Hi there,
first time posting here:

I'm trying to setup FAH GPU 6.20 console version on an XP-32bit machine. However, as far as I can make out it is impossible to use a Tesla C870 card? (It's a work machine and it might as well do something when not in use)

Here's the thing - my primary card (to run the display, Tesla cards have no video output!) is a geforce 7300 - so no CUDA support. Fine I thought, read the FAQ. Ran FAh... -gpu 1 -forcegpu nvidia_g80 ... all started up ok.
However as soon as the core is loaded I immediately get " mdrun_gpu returned -1 / UNSTABLE MACHINE " EUEs. Reading these forums suggests that to fix the bug you must connect a monitor to the secondary card, but like I mentioned already the Tesla has no video output.

Anyone figured out a way around this bug, by chance?

Cheers,
T.


If you're not getting much help posting in the general forum I'd advice to try to send a message about this to Slegrand memberlist.php?mode=viewprofile&u=2833

He's the nvidia representative on the forums and he should have the most knowledge how to get you folding on that tesla card :)
MtM
 
Posts: 3054
Joined: Fri Jun 27, 2008 2:20 pm
Location: The Netherlands

Re: Using Tesla as secondary card

Postby Noegzit » Fri Oct 09, 2009 8:52 pm

I had yesterday exactly the same problem with an ATI card and a Nvidia card.

Try to use -gpu 0 and not -gpu 1

The reason is that your primary (and only) graphic device is not a cuda device so it doesn't count.

That was the same thing for me : I tried to use -gpu 0 for the ATI CAL client and -gpu 1 for the nvidia cuda client when I should have used also -gpu 0 for my second graphic adapter (also used as a PPU)

toTOW sent me an explanation on a French forum.
The logic of numbering the gpus seems to be a per driver basis :
So I have to use -gpu 0 for my 1st CAL driver and -gpu 0 for my 1st CUDA driver.
You probably just have to use -gpu 0 cause your tesla is your 1st CUDA device.
Noegzit
 
Posts: 6
Joined: Fri Oct 09, 2009 8:16 pm


Return to NVIDIA specific issues

Who is online

Users browsing this forum: No registered users and 2 guests

cron