Message boards :
Number crunching :
Strange results with new GPU (GTX 970)
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Ryan Munro Send message Joined: 5 Feb 06 Posts: 63 Credit: 18,519,866 RAC: 10 |
Just like to understand really why I managed to get so many unvalidated and Validation inconclusive units so it does not happen again, I have only recently installed the 970 in this machine and it had a R9 290 in before, the drivers were fully uninstalled though but thats all I can think of as the issue as it seemed to work fine in the last machine which was media_box. BTW the results on that page are for the last month right? |
Ryan Munro Send message Joined: 5 Feb 06 Posts: 63 Credit: 18,519,866 RAC: 10 |
Just done some milky way at home units and getting validation inconclusive on all http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=596585&offset=0&show_names=0&state=3&appid= So thats seti / Einstein and mikly way I all having issues, card works fine for all other tasks though? I am at a loss now |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
As a matter of interest if it's a driver problem then there is a new version out, 347.25. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Ryan, how big is your PSU on that machine? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Just done some milky way at home units and getting validation inconclusive on all Well, the host is not showing any completed tasks now - valid, invalid, or inconclusive. MilkyWay is a strange project: it appears to me that very few of the posters on their message boards, and even fewer of the project scientists, understand how BOINC works and how to control it. (That's an absurd over-generalisation, of course, and my apologies to anyone caught on the wrong side of it). Specifically, most of the programming seems to be left to the long list of Undergraduate Research Assistants named on their home page: the only person who manages the BOINC Server (with any skill) is Professor Desell himself, and he is clearly too busy to devote much time to it. My belief is that at some point in the dim and distant past, somebody set an obscure configuration policy on their server, and subsequently forgot all about it: Adaptive Replication Tasks are sent out one at a time. If a host is 'trusted', the single returned result is accepted as valid and given credit. If the host is 'untrusted', a second or even third replication will be sent out after the first result is returned (unlike the process here, where all tasks are replicated, and both copies are sent out at nearly the same time). In the meantime the initial result will show as inconclusive. Crucially for this discussion, a newly attached host (or a host with a brand-new video card) will initially be treated as untrusted, and all results returned to MilkyWay by a new host will initially show as inconclusive. So, the initial quote by itself tells us nothing about whether the card worked at MilkyWay. I see that Ryan's machine has earned some credit there, so some tasks must have passed through it and been validated, but we don't know whether they were processed on the GTX 970. It would be worth Ryan's while to process a few more MilkyWay tasks, and report back on their final outcome (valid or invalid) after the extended replication/validation process has reached a final decision. |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
Travis is not even working at the RPI anymore. He moved to the University of North Dakota. http://tdesell.cs.und.edu/ |
©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.