SETI MB CUDA for Linux

[glennaxl]

just rebuild my system from using dotsch to ubuntu 10.4.1 and my seti cuda now  is too slow (using only 0.5-1.x cpu - windows is about 7-12cpu) - about 1hr to complete. running gpugrid to see if my cuda setup has issues but grugrid seems to be running perfectly well. So what's the deal here? any ideas?

driver ver: 195.36
boinc: 6.10.17
seti cuda: 2.3
seti app: setiathome-6.08.CUDA_2.2_x86_64-pc-linux-gnu
seti wus: not vlar

[sunu]

Do you raise the priority of the seti cuda app? Also why don't you use a newer graphics driver?

[Metod, S56RKO]

I've recently replaced gfx card with (still old but slightly) better one. This one is 9800GX2, with 2 GPUs and 512 MB graphic RAM per GPU. Now I wonder the following: does SAH CUDA app really make use of all the HW available? Or would it be sensible to try to run, say, 2 CUDA tasks in parallel on same GPU in order to 'increase throughput' (similar to hyperthreading)?

I've observed that it's quite possible to run 2 Einstein@Home CUDA apps in parallel (indeed GPU is not used much there) while my efforts to do so with SAH failed miserably (I doubt it's due to lack of graphic RAM, single SAH CUDA ran just fine on my old 8600 with mere 256 MB gRAM).

Any thoughts?

[sunu]

No, for pre-fermi GPUs there is no advantage in running multiple tasks per GPU, in fact usually the speed is worse.

Your card has two GPUs so you can run two tasks (1 per GPU) simultaneously.

[glennaxl]

Do you raise the priority of the seti cuda app? Also why don't you use a newer graphics driver?

Yah I found that was the problem - NICE is not playing nice at all 

Upgraded the driver and also used crunch3r cuda3 bins as the cuda2.2 doesn't have raised priority.

BTW, I have 1 rig that doesn't raised priority at system start-up (I also set a 20sec delay but doesn't work).  I have to manually restart boinc so setiathome-CUDA_3.0_6.09.x86_64_vlarkill will be raised. My other rigs works perfectly fine though.

[Metod, S56RKO]

At least the app I'm running (x86, 2.2, vlar-kill) has a nasty habit of complaining:[/li]

Code: [Select]

Cuda error 'GaussFit_kernel' in file './cudaAcc_gaussfit.cu' in line 497 : invalid configuration argument.
Seems benign though as most results have validated. Is there any particular reason for this error being reported and app seemingly still operating OK?

Can anybody comment on the error above? I don't think I've got any results without this 'error' while almost all of them validated OK. It's annoying to see thousands of lines of error while meaningful information gets truncated.

[glennaxl]

Do you raise the priority of the seti cuda app? Also why don't you use a newer graphics driver?

Yah I found that was the problem - NICE is not playing nice at all 

Upgraded the driver and also used crunch3r cuda3 bins as the cuda2.2 doesn't have raised priority.

BTW, I have 1 rig that doesn't raised priority at system start-up (I also set a 20sec delay but doesn't work).  I have to manually restart boinc so setiathome-CUDA_3.0_6.09.x86_64_vlarkill will be raised. My other rigs works perfectly fine though.

I solved the issue. Apparently, setiathome-CUDA_3.0_6.09.x86_64_vlarkill doesn't raise the priority if boinc is run by user "boinc" - it needs admin privilege. any thoughts?

[Raistmer]

Maybe boinc user needs right to increase process priority ?

[glennaxl]

Maybe boinc user needs right to increase process priority ?

Isn't it giving boinc user right to increase process priority the same as root? or is there a specific group that I need to add boinc to?

[Raistmer]

Root has many other rights too.
Hope linux has such conception as particular rights assignment for different users/groups.
In windows right assignment can be pretty granular.

[sunu]

Why do you need boinc user and other stuff? Keep the garbage out of your system. Install it in your home directory and go from there.

[Raistmer]

Why do you need boinc user and other stuff? Keep the garbage out of your system. Install it in your home directory and go from there.

In windows service(protected) install required for BOINC work w/o user log on. If linux can launch user's apps (installed in particular user home dir ) w/o user logon - good.
Also, it's good when some kind of autologon enabled. But if PC sits idle and just awaits when user come and do logon - it's not good.

[glennaxl]

Why do you need boinc user and other stuff? Keep the garbage out of your system. Install it in your home directory and go from there.

In windows service(protected) install required for BOINC work w/o user log on. If linux can launch user's apps (installed in particular user home dir ) w/o user logon - good.
Also, it's good when some kind of autologon enabled. But if PC sits idle and just awaits when user come and do logon - it's not good.

My rig is running ubuntu server with no x-window, stripped down kernel (using localmodconfig - 9MB kernel) and boinc running as service. Add some startup script to initialize cuda and other stuff. Everything is running perfectly now, just boinc running as root - i don't think its a risk as i don't have anything in this rig - just pure cruncher

[Metod, S56RKO]

Why do you need boinc user and other stuff? Keep the garbage out of your system. Install it in your home directory and go from there.

In windows service(protected) install required for BOINC work w/o user log on. If linux can launch user's apps (installed in particular user home dir ) w/o user logon - good.
Also, it's good when some kind of autologon enabled. But if PC sits idle and just awaits when user come and do logon - it's not good.

My rig is running ubuntu server with no x-window, stripped down kernel (using localmodconfig - 9MB kernel) and boinc running as service. Add some startup script to initialize cuda and other stuff. Everything is running perfectly now, just boinc running as root - i don't think its a risk as i don't have anything in this rig - just pure cruncher

You could be running BOINC under non-root account and just run that renice script as root. Theoretically that would make your rig even more safe.

[Claggy]

At least the app I'm running (x86, 2.2, vlar-kill) has a nasty habit of complaining:[/li]

Code: [Select]

Cuda error 'GaussFit_kernel' in file './cudaAcc_gaussfit.cu' in line 497 : invalid configuration argument.
Seems benign though as most results have validated. Is there any particular reason for this error being reported and app seemingly still operating OK?

Can anybody comment on the error above? I don't think I've got any results without this 'error' while almost all of them validated OK. It's annoying to see thousands of lines of error while meaningful information gets truncated.

If you read the first dozen pages of this thread, you'll see that the original 32bit Linux Cuda app was broken, it might be that the x86, 2.2, vlar-kill app isn't totally reliable eithier,
If you can, try swapping to a 64bit OS and use the 64bit app instead.

Claggy

Edit: The old broken 32bit Linux Cuda app really should be removed from the first post.