Message boards :
Number crunching :
IS BONIC 4.25 SLOWER?
Message board moderation
Author | Message |
---|---|
William Ross Send message Joined: 23 Sep 04 Posts: 4 Credit: 1,230,436 RAC: 0 |
ON THE SAME CPU BONIC 4.09 WENT THRU 4 TO 5 WU A DAY AVER 102-115 CR A DAY NOW IT GOES THRU 2-3 A DAY AND I CANT FIND WHERE THE CREDIT IS GOING THE SYSTEM HAD A HARD DRIVE CRASH AND WHEN I REPLACED THE HARD DRIVE I PUT BONIC 4.25 ON IT IT HAS BEEN RUNNING 3 DAYS AND THE WEB SIGHT (SETI) SAYS IT HASNT RETURNED A SINGLE WU I KNOW IT HAS. HOW DO I CHECK ON THIS? |
Saenger Send message Joined: 3 Apr 99 Posts: 2452 Credit: 33,281 RAC: 0 |
> ON THE SAME CPU BONIC 4.09 WENT THRU 4 TO 5 WU A DAY AVER 102-115 CR A DAY NOW > IT GOES THRU 2-3 A DAY AND I CANT FIND WHERE THE CREDIT IS GOING THE SYSTEM > HAD A HARD DRIVE CRASH AND WHEN I REPLACED THE HARD DRIVE I PUT BONIC 4.25 ON > IT IT HAS BEEN RUNNING 3 DAYS AND THE WEB SIGHT (SETI) SAYS IT HASNT RETURNED > A SINGLE WU I KNOW IT HAS. HOW DO I CHECK ON THIS? Try not to shout that LOUD ;) But to your question: The benchmark calculation was faulty with the older clients (is it really 4.09, not 4.19?) for Windoze machines. They cheated somehow, and therefore claimed too much credit. That's fixed with the 4.2x versions. Billy now gets the same benchmarks as Thorwald. Gruesse vom Saenger For questions about Boinc look in the BOINC-Wiki |
William Ross Send message Joined: 23 Sep 04 Posts: 4 Credit: 1,230,436 RAC: 0 |
Sorry about the cap lock. changing for 4.19 to 4.25 should not have slowed down the # of units the computer can run should it? |
eberndl Send message Joined: 12 Oct 01 Posts: 539 Credit: 619,111 RAC: 3 |
well, from what I've read, it sorta should...Windows used to skip part of the benchmark check (it was automatically optimized away by the system), causing all windows machines to have higher bench marks than comparable machines running linux, OSX etc etc. Now windows has to run the whole benchmark, which lowers the benchmark rate, and thus the number of WUs you can d/l... they will still take the same amount of time, but you WILL get less credit than before (but it will be more fair to non-windows users) |
Saenger Send message Joined: 3 Apr 99 Posts: 2452 Credit: 33,281 RAC: 0 |
> Sorry about the cap lock. changing for 4.19 to 4.25 should not have slowed > down the # of units the computer can run should it? No, it shouldn't. And some guy on the german Boinc-Forum just said that his crunching speed has increased on his machine with the step from 4.19 - 4.25. I don't track this values on my machines, so I can't tell you what has happened with mine. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
> well, from what I've read, it sorta should...Windows used to skip part of the > benchmark check (it was automatically optimized away by the system), causing > all windows machines to have higher bench marks than comparable machines > running linux, OSX etc etc. Uh, not really. While the benchmark shows Windows to be "faster" because of the optimized-out parts of the benchmark, the actual calculations take as long as they take, and that doesn't get optimized out. Now, I've read about folks using the time to do the calculation and turning that into a replacement benchmark, but if we knew the exact number of math ops in a work unit, we wouln't really need the benchmark, but using the estimate of the amount of work and the actual time taken seems kind of incestuous -- at least the benchmark isn't work, and work isn't the benchmark. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
> Sorry about the cap lock. changing for 4.19 to 4.25 should not have slowed > down the # of units the computer can run should it? > Several things to point out here. 1) The Boinc core clients such as 4.19 and 4.25 only manage the Applications. IMO they shouldn't (significantly) effect the actual amount of time it takes an Application to crunch a WU. 2) Boinc doesn't crunch anything. 3) Applications such as Seti 4.09, PPAH Mfold B125 4.24, and Einstein 4.79 are the things that do the crunching 4) The Claimed Credit is affected by the Benchmarks run. Boinc CC 4.19 and CC 4.25 benchmark a computer differently. It is believed that the new version 4.25 requests a more accurate amount of credit (however it is lower). There has been a problem with this which comes to light when Linux computers are compared to Windows computers in relation to "Claimed Credit". The Linux machine CC versions already had the "more accurate" benchmarks in the software and were therefore requesting LESS credit than a windows machine. Boinc is trying to even out the playing field with this change. hope this helps. tony |
©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.