Message boards :
Number crunching :
Anything relating to AstroPulse (2) tasks
Message board moderation
Previous · 1 . . . 10 · 11 · 12 · 13 · 14 · 15 · 16 . . . 50 · Next
Author | Message |
---|---|
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
The ancients talk of Astro Pulse but I think it is a myth. |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
So I’ve had a few APs pop up. 2 of the systems are just immediately “Aborting†the AP tasks without even running them (runtime = 0). What usually causes this? Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
I got one, woo hoo |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
So I’ve had a few APs pop up. Either you don't have the OpenCL drivers loaded or the tasks are 100% radar blanked. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
got 9, only 1 was 100% blanked |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
So I’ve had a few APs pop up. i have the nvidia 396.51 drivers on both systems in question. i assume that has openCL right? what does "radar blanked" mean? see this system here, you can check the tasks (i dont know what i'm looking at for task details) : https://setiathome.berkeley.edu/results.php?hostid=8559920&offset=0&show_names=0&state=6&appid=20 and here: https://setiathome.berkeley.edu/results.php?hostid=8561893&offset=0&show_names=0&state=0&appid=20 these are the 2 new systems that were just brought online in the past 2-3 days. each on nvidia drivers 396.51 (installed from ppa) each on Ubuntu 18.04.1 LTS each with Petris special app v0.96 (but i think this only applies to the v8 app, not v7. the v7 app is still there from the stock zi3v package each with a 1080ti and a a handful of 1060s this linux computer seems to be running them ok: https://setiathome.berkeley.edu/results.php?hostid=8432395&offset=0&show_names=0&state=0&appid=20 nvidia driver 396.45 installed from nvidia run file it is on Ubuntu 17.10 with Petris special app v0.96 (but i think this only applies to the v8 app, not v7. the v7 app is still there from the stock zi3v package has 2x 1060 the Windows systems running SoG/Lunatics also seem to be running AP tasks fine this one: https://setiathome.berkeley.edu/results.php?hostid=8433872&offset=0&show_names=0&state=0&appid=20 nvidia driver 391.35 windows 7 x64 2x 1080ti this one: https://setiathome.berkeley.edu/results.php?hostid=8555582&offset=0&show_names=0&state=0&appid=20 nvidia driver 397.64 windows 7 x64 2x 1060 Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Wiggo Send message Joined: 24 Jan 00 Posts: 34763 Credit: 261,360,520 RAC: 489 |
So I’ve had a few APs pop up.I can see from your computer list that 2 of your computers do not show any OpenCL component with your driver version and that is the reason that they are being aborted. Install the missing OpenCL component and all should be well. Cheers. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Yes, that is what I suspected. The OpenCL component of the drivers didn't get installed. If you have the ppa installed for the 396 drivers. Then sudo apt install nvidia-compute-396 This package provides a set of libraries which enable the NVIDIA driver to use GPUs for parallel general purpose computation through CUDA and OpenCL. [Edit] The Arecibo telescope is also used as a high power radar transmitter. When we are recording on the Alpha recorder, when the main dish pops out a radar pulse, we encode that time interval into our recordings so that the data is marked as corrupted by the radar pulse. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
So I’ve had a few APs pop up.I can see from your computer list that 2 of your computers do not show any OpenCL component with your driver version and that is the reason that they are being aborted. that's weird that the 396.45 drivers would include the openCL components and the 396.51 drivers dont. i usually just install the drivers by adding the nvidia proprietary PPA repo, then simply "sudo apt-get install nvidia-drivers-396" and the PPA autoselects whichever version. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
Yes, that is what I suspected. The OpenCL component of the drivers didn't get installed. If you have the ppa installed for the 396 drivers. Then nvidia drivers are obviously installed. nvidia-smi reports 396.51 as such. and the v8 app crunches cuda90 just fine. i've never had to do a separate install for nvidia-compute-[version] vs just running nvidia-drivers-[version]. about the "radar" jargon, what i'm asking is, how do i tell if a specific work unit is "radar blanked". i don't see that verbiage in any of the associated WU info. this is a task that failed : https://setiathome.berkeley.edu/result.php?resultid=6893323875 was it "radar blanked"? and how can you tell? Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
this is a task that failed : https://setiathome.berkeley.edu/result.php?resultid=6893323875 https://setiathome.berkeley.edu/result.php?resultid=6893323875 Look the exit... 201 (0x000000C9) EXIT_MISSING_COPROC So is not radar blank related. |
Wiggo Send message Joined: 24 Jan 00 Posts: 34763 Credit: 261,360,520 RAC: 489 |
Both your rigs with driver 396.51 have no OpenCL support listed, which means that you have CUDA support (but no OpenCL which is needed for Astropulse). Cheers. |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
this is a task that failed : https://setiathome.berkeley.edu/result.php?resultid=6893323875 what would it say if it was? Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
what would it say if it was? AFAIK the exit will be like any other WU: 0 (0x00000000) The crunching will be very fast... few seconds in my host. I don`t do AP anymore to post an example. |
Wiggo Send message Joined: 24 Jan 00 Posts: 34763 Credit: 261,360,520 RAC: 489 |
As I've already explained, those 2 setups have no OpenCL support in the drivers (it will say so if it was just like it shows on your other rigs) and that is what the app is looking for and when it can't find that support the app exits with the "EXIT_MISSING_COPROC" in your Stderr output for that task (which is what you are getting).this is a task that failed : https://setiathome.berkeley.edu/result.php?resultid=6893323875 Either install the missing OpenCL part of your drivers or stop accepting Astropulse work. Keith Myers has already posted how to get that OpenCL component installed. ;-) [edit] BTW I picked up 11 AP's this morning. Cheers. |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
i recognize that. but im still on the hunt to understand the exact behavior. nothing wrong with wanting to know exactly how things are working and how you can tell certain aspects of a WU by looking at it. I know the driver is missing the OpenCL component now. I'm not disputing that. i can remedy that once i get home but it IS strange that i've never had this issue before. using linux and older driver versions (390 and older), i've always just run "sudo apt-get install nvidia-driver-[version]". it installs, i reboot, and on my way. why would previous driver version install the opencl component for me, while now it does not? something is different, given an identical install. there has to be a reason for that. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Wiggo Send message Joined: 24 Jan 00 Posts: 34763 Credit: 261,360,520 RAC: 489 |
From what I've heard it has something to do with the 18.04 version you're using on those 2 rigs, your other Linux rigs are older versions and are not effected by this. Cheers. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I think there is something wrong with the 396.51 release from the ppa. Too many people reporting the OpenCL component missing and not getting installed when the driver moved from 396.45 to 396.51. The output in stderr.txt will specifically report 100% blanked for a radar blanked AP task. It will run very fast and give you maybe 2 credits. Don't have any examples to show since they would have been long cleared by now. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
sudo apt install nvidia-compute-396 Returns unable to locate package. What now? just tried this from the other thread: sudo apt-get install ocl-icd-libopencl1 seems to have worked. we'll see when another AP task rolls around Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
You could try 'ldd' on the AstroPulse app and see if it reports anything missing. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.