Message boards :
Number crunching :
Milkyway@home separation(modified fit) 1.36 coming up as invalid
Message board moderation
Author | Message |
---|---|
Send message Joined: 29 Mar 09 Posts: 2 Credit: 14,947,811 RAC: 0 |
Almost all the work units are coming up invalid on my computer or validation inconclusive. I have 4 GTX 780ti and two GTX Titans all on one motherboard. Any ideas? Thanks |
Send message Joined: 18 Jul 09 Posts: 300 Credit: 303,562,776 RAC: 0 |
Almost all the work units are coming up invalid on my computer or validation inconclusive. I have 4 GTX 780ti and two GTX Titans all on one motherboard. Any ideas? Thanks All your invalid work units are the Modified Fit variety. Simply stop running them. |
Send message Joined: 24 Jan 11 Posts: 708 Credit: 543,747,953 RAC: 131,691 |
You've run into the well known problem with BOINC and MW on heavily stressed and powerful systems. The computer doesn't have enough resources available at the end of task completion to properly write out the state file. That is the only reason the tasks are invalid. The tasks likely are generating valid results but the computer is so fast that it starts crunching another task in the slot the previous task occupied and finished but is unable to finish the task state file because the system is so heavily stressed for resources it is unable to do so. I generate about 3% error rate on the invalids, primarily the 1.36 application but it can happen to the 1.02 application also mainly because the 1.02 task takes 5 times as long to finish compared to the 1.36 task. You are not polluting the science database with bad results, the invalids just waste compute cycles. As the other poster states, you can just stop processing those tasks in your preferences so you don't want to waste compute power. I can live with the 3% error rate on my systems since I do generate valid results for the 1.36 tasks mostly. Cheers, Keith |
Send message Joined: 29 Mar 09 Posts: 2 Credit: 14,947,811 RAC: 0 |
Thanks. Did that. |
©2024 Astroinformatics Group