no GPU tasks anymore!

Exard3k
Exard3k
Joined: 25 Jul 21
Posts: 66
Credit: 56155179
RAC: 0
Topic 226090

E@H doesn't send me any new GPU tasks. Noticed that the processing ran dry this morning. I aborted some CPU tasks before because BOINC is a bit too optimistic on my CPU power and I got more tasks than I could handle before the deadline. I didn't check if I can get new CPU tasks because I got plenty left.

 

</p>

<p>Fri 24 Sep 2021 11:11:01 AM CEST | Einstein@Home | Sending scheduler request: To fetch work.<br />
Fri 24 Sep 2021 11:11:01 AM CEST | Einstein@Home | Requesting new tasks for NVIDIA GPU<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | Scheduler request completed: got 0 new tasks<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | No work sent<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | No work is available for Binary Radio Pulsar Search (Arecibo)<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | No work is available for Binary Radio Pulsar Search (Arecibo, GPU)<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | No work is available for Gamma-ray pulsar search #5<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | No work is available for Gamma-ray pulsar binary search #1 on GPUs<br />
<strong>Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | (reached daily quota of 13 tasks)</strong><br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | Project has no jobs available<br />
Fri 24 Sep 2021 11:11:02 AM CEST | Einstein@Home | Project requested delay of 53946 seconds</p>

<p><br />

 

What is that daily quota entry? Never seen that before.

solling2
solling2
Joined: 20 Nov 14
Posts: 219
Credit: 1577607973
RAC: 19940

Exard3k schrieb: (...) What

Exard3k wrote:

(...)

What is that daily quota entry? Never seen that before.

The FAQ section, bottom left of this page, may give you some hints.

More interesting though is for what reason all of your GPU tasks end up with Computation error?

Exard3k
Exard3k
Joined: 25 Jul 21
Posts: 66
Credit: 56155179
RAC: 0

solling2 wrote:More

solling2 wrote:

More interesting though is for what reason all of your GPU tasks end up with Computation error?

 

Oh I just realized I got errors on the GPU tasks. I didn't pay attention to that because aborted tasks end up there too. Guess my system update messed something up as the GPU is running fine otherwise. I'll look into it. Thanks for the tip!

 

Just in case it's a driver issue...where do I see if BOINC detects the GPU? If BOINC doesn't "see" my GPU, I should be able to narrow it down.

 

edit: boinccmd --get_host_info lists my GPU, looks all good.

 

edit2: Done another "update", everything fine now. compute works, no errors with the new tasks. Wierd stuff happens. But thanks for hinting me to the FAQ and my failed tasks.
 

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3958
Credit: 47046462642
RAC: 65123077

you are in a 24hr timeout due

you are in a 24hr timeout due to errors. you wont be able to get any tasks until the 24hr timer is up.

_________________________________________________________________________

Exard3k
Exard3k
Joined: 25 Jul 21
Posts: 66
Credit: 56155179
RAC: 0

Ian&Steve C. wrote:you are

Ian&Steve C. wrote:

you are in a 24hr timeout due to errors. you wont be able to get any tasks until the 24hr timer is up.

 

edited my last posting. Seems like I passed that timer now. Not sure what happened in the first place, but server sent me new work 2min ago. I also sent completed CPU tasks back, maybe sending valid tasks back was the catch?

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3958
Credit: 47046462642
RAC: 65123077

your error indicates that the

your error indicates that the app could not find the GPU. maybe a driver update. or maybe the driver crashed.

_________________________________________________________________________

Exard3k
Exard3k
Joined: 25 Jul 21
Posts: 66
Credit: 56155179
RAC: 0

Ian&Steve C. wrote: your

Ian&Steve C. wrote:

your error indicates that the app could not find the GPU. maybe a driver update. or maybe the driver crashed.

 

Yeah probably that. I ran a full pacman -Syu and I've seen Nvidia drivers and openCL were updated.

The joys of a rolling release distro. Sometimes things just get f**** up ;) I was about to load an old snapshot, but seems like all is running fine again. boinccmd is a really good tool I never used on Windows.

Stray_Trons
Stray_Trons
Joined: 19 Jun 21
Posts: 2
Credit: 84611975
RAC: 0

I ran in into a similar issue

I ran in into a similar issue on 14 Sept. I noticed my average completion stats had started to crash.  Turns out Windoze 10 decided to patch without asking permission.

The short version is the RTX3070 driver and audio driver were damaged during the patch. Still have no idea what the audio driver and BOINC have in common, but MicroSucks never cases to amaze me about how many things it can screw up.

Finally finished troubleshooting this morning and reset the project, we will see if BOINC performance and my benchmark stars align.

(Sorry, I had to do that pun.)

Dave. S

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3958
Credit: 47046462642
RAC: 65123077

Stray_Trons wrote: I ran in

Stray_Trons wrote:

I ran in into a similar issue on 14 Sept. I noticed my average completion stats had started to crash.  Turns out Windoze 10 decided to patch without asking permission.

The short version is the RTX3070 driver and audio driver were damaged during the patch. Still have no idea what the audio driver and BOINC have in common, but MicroSucks never cases to amaze me about how many things it can screw up.

Finally finished troubleshooting this morning and reset the project, we will see if BOINC performance and my benchmark stars align.

(Sorry, I had to do that pun.)

Dave. S

reboot the computer to reload the GPU drivers, if you haven't already.

_________________________________________________________________________

Exard3k
Exard3k
Joined: 25 Jul 21
Posts: 66
Credit: 56155179
RAC: 0

Yeah I can confirm that

Yeah I can confirm that rebooting after updating video driver stuff is actually useful :)

HeatForScience
HeatForScience
Joined: 7 Jan 18
Posts: 1
Credit: 667553645
RAC: 1202843

I had a similar problem

I had a similar problem though my situation is more complicated. I have headless 4 GPU system with various GPUs I've scrounged up over the last couple of years: A Tesla K20, K40, GTX 970, and GTX 1070ti. used almost exclusively for BOINC crunching, plus the occaisonal VR gaming session.

I've run into funky driver issues since the Tesla cards aren't explicitly supported inside the GEFORCE drivers that for the 1070 and 970. But loading the cards one at a time, rebooting, and verifying the driver loads properly seems to work. Unfortunately, when windows decided to update the drivers, everything went off the rails.

I found this article describing how to prevent Windows from updating drivers for devices with specific hardware IDs and plan to give it a try. Fingers crossed.

 

 



 

Comment viewing options

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