Message boards : Questions and problems : flops estimate for different app plans
Message board moderation
Author | Message |
---|---|
Send message Joined: 9 Apr 06 Posts: 302 |
For ati13ati I see time estimation change with more completed CPU-only tasks. And reverse, CPU only time estimate changes with each new completed ati13ati GPU task. BOINC 6.10.48 Question is: do separate time estimations for different app_versions implemented already and in what BOINC version if yes ? |
Send message Joined: 5 Oct 06 Posts: 5149 ![]() |
Question is: do separate time estimations for different app_versions implemented already and in what BOINC version if yes ? They're implemented on the server only - do keep up! And only on SETI Beta so far. |
Send message Joined: 9 Apr 06 Posts: 302 |
Question is: do separate time estimations for different app_versions implemented already and in what BOINC version if yes ? I see. This makes task of writing correct app_info host-specific :/ |
Send message Joined: 5 Oct 06 Posts: 5149 ![]() |
Question is: do separate time estimations for different app_versions implemented already and in what BOINC version if yes ? They always were. Re-read the very first post ever written about GPU flop entries: http://setiweb.ssl.berkeley.edu/beta/forum_thread.php?id=1536&nowrap=true#36887 Almost 15 months ago. It says your machine's floating point benchmark, multiplied .... your graphics card's CUDA FLOP count, multiplied .... Edit - in case this is the point of confusion: the values to put in app_info.xml are speed estimates, which relate to your hardware: not counter values relating to the task size. |
Copyright © 2025 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.