Message boards :
Number crunching :
Flakey AMD/ATI GPUs, including RX 5700 XT, Cross Validating, polluting the Database
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 20 · Next
Author | Message |
---|---|
Bluerazor Send message Joined: 22 May 99 Posts: 15 Credit: 3,889,427 RAC: 12 |
New drivers today, but don't expect a fix. As promised, it's on the known issues (finally): https://www.amd.com/en/support/kb/release-notes/rn-rad-win-19-12-3 |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
As promised, it's on the known issues (finally): made clickable bottom of the known issues list 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 |
what a cop out too. SETI isnt the only project having issues with these cards. Instead of naming SETI specifically, they should have said something like some OpenCL compute applications will provide incorrect results. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
what a cop out too. SETI isnt the only project having issues with these cards. Instead of naming SETI specifically, they should have said something like some OpenCL compute applications will provide incorrect results. . . Exactly there is NO maybe about it ... they are giving out nothing but rubbish ... Stephen :( |
Wiggo Send message Joined: 24 Jan 00 Posts: 36654 Credit: 261,360,520 RAC: 489 |
A new 1 decided to pester me, along with the usual culprits. :-( JohnDoe 9166075 Cheers. |
Wiggo Send message Joined: 24 Jan 00 Posts: 36654 Credit: 261,360,520 RAC: 489 |
I was either very very lucky yesterday (UTC time) as only 2 regulars were of a very minor annoyance to me, or more have shut their GPU's down (or have they been finally locked out?). Cheers. |
Mr. Kevvy Send message Joined: 15 May 99 Posts: 3805 Credit: 1,114,826,392 RAC: 3,319 |
I was either very very lucky yesterday (UTC time) as only 2 regulars were of a very minor annoyance to me, or more have shut their GPU's down (or have they been finally locked out?). I think it's working.. between my pestering (JohnDoe 9166075 is the latest and thank you) and word getting out. [AfZ]TomServo1 is the latest to reply and indicate the GPU is now disabled. |
Eric Korpela Send message Joined: 3 Apr 99 Posts: 1382 Credit: 54,506,847 RAC: 60 |
I've made a change to the validator that raises the effective quorum for overflow results. The should limit the number of successful cross validations for these GPUs. That change is still missing a few, so I need to check the overflow detection mechanism. No, my mistake. It is working. The results I was looking at were before the change. @SETIEric@qoto.org (Mastodon) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
I've made a change to the validator that raises the effective quorum for overflow results. The should limit the number of successful cross validations for these GPUs. Thank you for that. That should also help reduce the server load. Grant Darwin NT |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
That should also help reduce the server load. Not be so sure about that. Will take more time to clear the quorum for overflow results. But sure will reduce the pollution of the DB giving more time to they fix the driver. |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
I've made a change to the validator that raises the effective quorum for overflow results. The should limit the number of successful cross validations for these GPUs. . . Many thanks, hope all this does not spoil your Christmas too much. Stephen :) |
Wiggo Send message Joined: 24 Jan 00 Posts: 36654 Credit: 261,360,520 RAC: 489 |
I must of just been lucky yesterday as today I can report 3 new culprits with 1 being a Linux job so it's certainly not just Windows thing. :-( fredi 7913572 Linux Recedham 954834 TomasFraus 8445239 Cheers. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I must of just been lucky yesterday as today I can report 3 new culprits with 1 being a Linux job so it's certainly not just Windows thing. :-( Good find. I thought there was the same issue with the Linux hosts. But rare to find one using the 5700 on the Seti project. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Chooka Send message Joined: 13 Dec 12 Posts: 10 Credit: 7,913,101 RAC: 0 |
That latest AMD driver update 19.12.2 was garbage. Caused my system to crash. That running 1 wu on a Radeon VII. I reverted back to 19.9.2 and it's been stable since. |
Wiggo Send message Joined: 24 Jan 00 Posts: 36654 Credit: 261,360,520 RAC: 489 |
That latest AMD driver update 19.12.2 was garbage. Caused my system to crash. That running 1 wu on a Radeon VII. I reverted back to 19.9.2 and it's been stable since.But at least your cards, so long as you stay with your current drivers, are not effected by the huge problem that effects the newer RX pose just out of the box and so far no solution is in site (other than AMD now stating that their RX cards are not "SETI Friendly"). But I would be very wary about newer drivers on their older kits now. ;-) Cheers. |
Justin Turner Arthur Send message Joined: 20 Oct 03 Posts: 12 Credit: 3,929,052 RAC: 2 |
That is good info. Looks like fredi's computer 8867025 is running the PAL OpenCL driver for Linux (probably from the AMDGPU-Pro package). As ROCm driver cards still aren't chosen in the ATI multibeam client's plan class, the only results we'd get from those will be from anonymous hosts, and that's only when ROCm finally gets Navi support. So the big question is how this continues to work on the macOS OpenCL runtime according to observations posted to this thread. My guesses are one of these: - There's an issue with Navi support at the PAL layer. Both the Windows driver and AMDGPU-Pro use the PAL layer provided by device drivers. - Both the Windows and AMDGPU-Pro OpenCL compilers are shipped with the same GCN bitcode mistakes. I don't know if the Apple stack uses PAL at all. |
Eric Korpela Send message Joined: 3 Apr 99 Posts: 1382 Credit: 54,506,847 RAC: 60 |
Since yesterday, we've fallen back to the old server so anonymous platform apps should be able to get work. Since this morning we should have had the validator that requires 3 results for overflow results. Merry Christmas! Now I just need to explain why some workunits like this one are sneaking through. @SETIEric@qoto.org (Mastodon) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
Since yesterday, we've fallen back to the old server so anonymous platform apps should be able to get work. Since this morning we should have had the validator that requires 3 results for overflow results. Merry Christmas!Thanks for all your efforts. We're just very curious as to how the Scheduler from Beta ended up on main, after all the complaints about it's issues over at Beta about it? Grant Darwin NT |
Tom M Send message Joined: 28 Nov 02 Posts: 5126 Credit: 276,046,078 RAC: 462 |
Since yesterday, we've fallen back to the old server so anonymous platform apps should be able to get work. Since this morning we should have had the validator that requires 3 results for overflow results. Merry Christmas! +42 A proud member of the OFA (Old Farts Association). |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
Since yesterday, we've fallen back to the old server so anonymous platform apps should be able to get work. Since this morning we should have had the validator that requires 3 results for overflow results. Merry Christmas! . . Thanks for all the hard work, and I hope whoever was responsible buys you guys a nice dinner out (and wine) :) . . Have a happy New Year! Stephen :) |
©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.