Message boards :
Number crunching :
Beta 4.72 adjusts completion time nicely but doesn't keep them
Message board moderation
Author | Message |
---|---|
Brian D from Georgia Send message Joined: 18 May 99 Posts: 5 Credit: 547,809 RAC: 0 |
An aggrevation of mine since joining BOINC was the terrible job it does of calculating the time it will take to complete a work unit. I have 4 hosts with varying Intel cpu types and speed. I can never cache an appropriate number of work units because its calculations are off by factors of 3 to 1 or worse. I was happy to see 4.72's documentation have a feature that adjusts the calculated work time each time a work unit gets completed. This works great, my calculated times slowly drift downward with each finished work unit just as advertised. Problem is that once I connect to upload/download work units, the calculated time gets reset back to the original excessively long estimate. I run a 4 day connect time and suspend network connectivity until I get about 80 percent of my work units completed and then connect. I also run Einstein, Predictor, LHC as well as SETI in various combo's on my 4 hosts. This "bug" is present in all clients and all hosts. Is anyone else noticing this? Is this a bug in 4.72? Will it be fixed in the next beta? |
Jesse Viviano Send message Joined: 27 Feb 00 Posts: 100 Credit: 3,949,583 RAC: 0 |
An aggrevation of mine since joining BOINC was the terrible job it does of calculating the time it will take to complete a work unit. I have 4 hosts with varying Intel cpu types and speed. I can never cache an appropriate number of work units because its calculations are off by factors of 3 to 1 or worse. I was happy to see 4.72's documentation have a feature that adjusts the calculated work time each time a work unit gets completed. This works great, my calculated times slowly drift downward with each finished work unit just as advertised. Problem is that once I connect to upload/download work units, the calculated time gets reset back to the original excessively long estimate. I run a 4 day connect time and suspend network connectivity until I get about 80 percent of my work units completed and then connect. I also run Einstein, Predictor, LHC as well as SETI in various combo's on my 4 hosts. This "bug" is present in all clients and all hosts. Is anyone else noticing this? Is this a bug in 4.72? Will it be fixed in the next beta? This is actually a bug in SETI@home because it does not accurately report the percentage of work completed. Einstein@home is much better in its calculations of the percentage of work left to complete, and therefore BOINC can actually estimate the time for Einstein@home fairly accurately. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
An aggrevation of mine since joining BOINC was the terrible job it does of calculating the time it will take to complete a work unit. I have 4 hosts with varying Intel cpu types and speed. I can never cache an appropriate number of work units because its calculations are off by factors of 3 to 1 or worse. I was happy to see 4.72's documentation have a feature that adjusts the calculated work time each time a work unit gets completed. This works great, my calculated times slowly drift downward with each finished work unit just as advertised. Problem is that once I connect to upload/download work units, the calculated time gets reset back to the original excessively long estimate. I run a 4 day connect time and suspend network connectivity until I get about 80 percent of my work units completed and then connect. I also run Einstein, Predictor, LHC as well as SETI in various combo's on my 4 hosts. This "bug" is present in all clients and all hosts. Is anyone else noticing this? Is this a bug in 4.72? Will it be fixed in the next beta? I believe I read elsewhere that the correction factor is used for the time estimate, but is not carried through to the SETI servers when requesting work. Don't know when you'll see it, but I suspect that it's coming. |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
There is atleast one bug in the correction-factor that was hopefully fixed a couple days ago, so will be included in v4.73 or whatever next build will be... |
T1ger5hark Send message Joined: 26 Sep 02 Posts: 1 Credit: 2,315 RAC: 0 |
Brian D, can you contact me. I have a question about my other acct. on the MSC team. You are the leader of our group, right? Thanks. Warren |
©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.