Message boards :
Number crunching :
Dual GPUs - failed to calculate likelihood
Message board moderation
Author | Message |
---|---|
Send message Joined: 14 Jun 11 Posts: 5 Credit: 140,086,219 RAC: 0 |
I'm running an AMD 1100 Hex with dual HD 5830s. Operating system is Ubuntu 10.10 - 2.6.35-30, Catalyst 11.9, Boinc 6.10.58 The system runs full Milkyway and Collatz GPU WUs only, timeslicing is at 30 minutes. Collatz has no problem running two GPU WUs simultaneously. On Milkyway some simultaneous WUs process OK and some fail with a "failed to calculate likelihood" error. The WUs in question are ps_separation_82_2s_mix4_ . . . Generally the first of a pair of processes fail and the second works. Sometimes it appears both work. I have yet to observe the system processing one MW and one CC WU at the same time, so don't know what will happen there. Any help would be appreciated. Thanks. |
Send message Joined: 8 Feb 08 Posts: 261 Credit: 104,050,322 RAC: 0 |
You have several validation errors too. I would check for a heat problem first. Downclock gpu and mem a bit and see if the problem disappears. Milkyway is known for stressing the gpu more than Collatz. |
Send message Joined: 13 Aug 10 Posts: 10 Credit: 115,945,904 RAC: 0 |
I've got one WU that failed and it's with the same problem as yours: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=31051023 The "mix" ones are supposed to be a new set of WUs aren't they? http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2662 Most likely cause is that there is something wrong with the new WUs themselves. |
Send message Joined: 14 Jun 11 Posts: 5 Credit: 140,086,219 RAC: 0 |
Thanks for the responses. Not sure what you mean by several validation errors. Fairly sure it's not a heat problem - currently running at 58c with CC WUs, MW sends it to about 62c. About a month ago I had this system set up with a Regor 250 running 2 cores at 3GHz. Had the same problem then with MW, presumably before the "mix" WUs were issued. Made a great CC only rig so I didn't bother any further Now I have the hex in it and it seems a shame to waste all those cores. |
Send message Joined: 15 Jul 08 Posts: 383 Credit: 729,293,740 RAC: 0 |
Thanks for the responses. The validation errors are unfortunately normal and it's been said they're due to the handing of the stderr.out in BOINC/MW. They did not happen in the old ATI app. Heat has nothing to do with it. |
Send message Joined: 14 Jun 11 Posts: 5 Credit: 140,086,219 RAC: 0 |
I've checked the validation errors and only one applies to the rig in question. They appear infrequently and seem to have nothing to do with the likelihood error. I now find that the error always occurs with simultaneous MC and CC WUs. I should have mentioned that I am running the GPUs in Crossfire. This may not be significant but I have never been able to acccess the cards independently under Linux. This is (probably) a limitation in the Catalyst and/or AMD-APP-SDK-v2.5 (not really relevant here but I had that problem when mining.) So it's either a Linux problem or a Boinc / MW problem. If anyone is successfully running Linux / ATI Crossfire with multiple GPU WUs I would appreciate knowing how you did it. Thanks. |
©2024 Astroinformatics Group