nvidia gpu errors

Cat22
Cat22
Joined: 13 May 21
Posts: 28
Credit: 925384315
RAC: 1632688
Topic 229860

I'm getting  lot of failed GPU units. How can I determine why?

My laptop is an ASUS Zenbook with 2 gpu's an intel and a nvidia

Linux kde on openSuse Tumbleweed

NVIDIA GeForce GTX 1650 Ti Mobile,  Driver Version: 535.54.03

Sample Errors and Info from boinc:

Output file h1_1029.80_O3aC01Cl1In0__O3ASHF1a_1030.00Hz_27859_5_0 for task h1_1029.80_O3aC01Cl1In0__O3ASHF1a_1030.00Hz_27859_5 absent

27-Jul-2023 05:37:25 [---] Starting BOINC client version 7.22.2 for x86_64-suse-linux-gnu
27-Jul-2023 05:37:25 [---] log flags: file_xfer, sched_ops, task
27-Jul-2023 05:37:25 [---] Libraries: libcurl/8.1.2 OpenSSL/3.1.1 zlib/1.2.13 brotli/1.0.9 zstd/1.5.5 libidn2/2.3.4 libpsl/0.21.2 (+libidn2/2.3.4) libssh/0.10.5/openssl/zlib nghttp2/1.54.0

27-Jul-2023 05:37:26 [---] CUDA: NVIDIA GPU 0: NVIDIA GeForce GTX 1650 Ti with Max-Q Design (driver version 535.54, CUDA version 12.2, compute capability 7.5, 3904MB, 3904MB available, 2458 GFLOPS peak)
27-Jul-2023 05:37:26 [---] OpenCL: NVIDIA GPU 0: NVIDIA GeForce GTX 1650 Ti with Max-Q Design (driver version 535.54.03, device version OpenCL 3.0 CUDA, 3904MB, 3904MB available, 2458 GFLOPS peak)
27-Jul-2023 05:37:26 [---] libc:  version 2.37

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4964
Credit: 18744039146
RAC: 7018222

Just read the output file of

Just read the output file of any failed task.  Tells you right there why it failed.

OpenCL error: CL_MEM_OBJECT_ALLOCATION_FAILURE

Not enough memory for the GW tasks.  Since this is a laptop it tries to get memory allocated from the system memory pool.  Which it fails in doing so.

The GW tasks need more than 4GB to run.

Change your project preferences to deselect the GW tasks.  Run only the FGRPB1G and BRP7 tasks.

 

Cat22
Cat22
Joined: 13 May 21
Posts: 28
Credit: 925384315
RAC: 1632688

Thanks!

Thanks!

Cat22
Cat22
Joined: 13 May 21
Posts: 28
Credit: 925384315
RAC: 1632688

So, do i change the GPU

So, do i change the GPU Utilization Factor for GW tasks to 0?

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4964
Credit: 18744039146
RAC: 7018222

Cat22 wrote: So, do i change

Cat22 wrote:

So, do i change the GPU Utilization Factor for GW tasks to 0?

No,  you just unselect it or toggle the checkmark from the list of applications.

Applications

All-Sky Gravitational Wave search on O3 (O3AS)

 

Michael E.@ team Carl Sagan
Michael E.@ tea...
Joined: 17 Sep 05
Posts: 6
Credit: 35032944
RAC: 38038

Thank you for the info! In

Thank you for the info!

In my case, any O3AS tasks for NVidia execute for about a minute and then fail. My GPU is: NVIDIA GeForce GTX 1050 Ti (4096MB) driver: 457.51

I went to Preferences > Advanced Settings Project > and under Applications and unchecked the All-Sky Gravitational Wave search on O3 (O3AS) line.

It did not have any effect - I still got more of those O3AS tasks. I aborted them. Twice! 

Maybe it takes a few minutes to propagate? Some strange browser cache issue?  I will turn off GPU use for now.

 

Cat22
Cat22
Joined: 13 May 21
Posts: 28
Credit: 925384315
RAC: 1632688

Are you using web pref's?

Are you using web pref's? Check in boinc/options/computing prefs, make sure you have selected the Einstein project first tho in the boinc projects tab

mikey
mikey
Joined: 22 Jan 05
Posts: 12691
Credit: 1839095536
RAC: 3727

Michael E.@ team Carl Sagan

Michael E.@ team Carl Sagan wrote:

Thank you for the info!

In my case, any O3AS tasks for NVidia execute for about a minute and then fail. My GPU is: NVIDIA GeForce GTX 1050 Ti (4096MB) driver: 457.51

I went to Preferences > Advanced Settings Project > and under Applications and unchecked the All-Sky Gravitational Wave search on O3 (O3AS) line.

It did not have any effect - I still got more of those O3AS tasks. I aborted them. Twice! 

Maybe it takes a few minutes to propagate? Some strange browser cache issue?  I will turn off GPU use for now.

Also be sure to Yes is NOT checked in this setting on that page:

Allow non-preferred apps:
yes

no

If no work for selected applications are available, accept tasks from other applications?

 

Michael E.@ team Carl Sagan
Michael E.@ tea...
Joined: 17 Sep 05
Posts: 6
Credit: 35032944
RAC: 38038

CAT22 wrote: Are you

CAT22 wrote:

Are you using web pref's? Check in boinc/options/computing prefs, make sure you have selected the Einstein project first tho in the boinc projects tab

Thank you @CAT22!  I never noticed that in decades of BOINC use. I always assumed doing a Save would update the web prefs for the current project. Another usability issue is entering percentage numbers instead of selecting the number of CPUs. Sorry, but I worked in software products for decades.

Michael E.@ team Carl Sagan
Michael E.@ tea...
Joined: 17 Sep 05
Posts: 6
Credit: 35032944
RAC: 38038

The Advanced View > Options >

The Advanced View > Options > Computing Preferences point to the WCG web site even if I select the Einstein project first (bug?). So I do not think I can change any Einstein Preferences for Projects. 

Good news about the O3AS tasks for NVidia on my Win 10 desktop PC - if I reboot the PC it seems to process the O3AS tasks without error for almost a day. 

Cat22
Cat22
Joined: 13 May 21
Posts: 28
Credit: 925384315
RAC: 1632688

What version of boinc are you

What version of boinc are you running? The "Use Web Preferences" button is missing on some boinc versions

Oh, and I assume you are running Windows? I only use Linux so I wont be much help in any windows environment

Comment viewing options

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