VLAR Tasks End In Error |
![]() |
| log in |
Questions and Answers : GPU applications : VLAR Tasks End In Error
| Author | Message |
|---|---|
|
I'm running a Tesla C1060 card and cuda work units generally take about about 13 minutes each.VLAR tasks error off after about 1 hour and 40 minutes. Machine is a i7 with 960 cpu & 6gb RAM. OS is Windows XP Pro x64. How can I improve the time required and eliminate the errors on the VLAR tasks? | |
| ID: 1074963 · | |
|
Could you post a link to one of those VLARs? On this host, I have only found "Aborted by user", "Error while downloading" and "Abandoned" tasks. The only "Error while computing" was a -12 error with an angle range of 3.619464 (1791242503). | |
| ID: 1074985 · | |
|
I have no work units at present. Maybe when server is back up. How would I send a link? | |
| ID: 1075123 · | |
|
Just copy the link from the online tasks page after an erroneous task has been reported. If you highlight the URL when posting and then click the "URL" button, the link will be active for the readers. | |
| ID: 1075140 · | |
|
CUDA isn't optimized for VLAR WU's so you'll see very poor performance on VLARs. | |
| ID: 1075200 · | |
|
I switched to an optimized version of Seti and vlar tasks now all go to the CPUs. | |
| ID: 1079688 · | |
I switched to an optimized version of Seti and vlar tasks now all go to the CPUs. Optimized applications don't send VLARs to the CPU on their own. Are you sure that you didn't use a rescheduler? And if not, tasks with the suffix .vlar are sent to the CPU by the server, whether it's for the stock application or not. Gruß, Gundolf | |
| ID: 1079698 · | |
Questions and Answers : GPU applications : VLAR Tasks End In Error
| Copyright © 2013 University of California |