When I started this with the same three computers now being used, the average credit stats were varying from 560's to 620's. Nothing has changed here, same computers, doing the same things and each taking about 10 hours per work unit. As of today, the average credit is down to around 400, and has been pretty low for several months.
I'm using BOINC version 4.19. All computers are running WinXP Pro or Win2000 Pro. Maybe I should downgrade all the computers to Win98SE because they all processed SETI@Home faster on 98SE than on WinXP.
Copyright © 2024 Einstein@Home. All rights reserved.
Average credit keeps sinking
)
Ed,
I can't speak to the SETI issue, but you do say that all 3 rigs are processing Einstein at the same 10hr/WU rate as before.
To the Einstein credit-falling issue, Ed, you're a victim of progress, with a minor short-term assist from the 15 WUs that are pending credit. The 4.19 Einstein benchmarked higher than does the 4.45 version, thus calculating higher claimed credit. So, while your claimed credit remains ~90/WU, most other crunchers have started with or moved to the 4.45. Pre-4.45, most others were also claiming high credit, and so the granted credit for the quorum was accordingly high. Since the credit granted is the mean of the first 3 validated WUs, and the odds are that you're now almost always crunching a WU with computers claiming lower credit, your higher claim will usually be dragged down.
Regards,
Michael
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: When I started this
)
As you probably know, the accumulation of credit is a two step process. The first step is "claimed credit" - which is totally a function of cpu time and benchmarking. The second step is granted credit - a random and somewhat arbitrary process that may increase or decrease the amount of credit you are awarded bases on comparison to the other units in your quorom. You can't "do anything about" granted credit - you just have to be luckily. But you can influence claimed credit to some degree. That is, you can influence benchmarking. BOINC automatically does benchmarking periodically - once a week (if I remember correctly). If BOINC happens to run the benchmarking when your computer is also busy doing something else, your benchmarks will be lower than if your computer had been idle. (Also, if spyware somehow creeps into your system, your benchmarks will go down.) But, the BOINC Manager also allows you to run benchmarks on demand allowing you to make sure your computer is idle. One other thing (and I am not recommending this) - the Seti website has a link called "Porting and optimization". If you go there, you can find a "BOINC client" optimized for your particular processor. Using one of these optimized BOINC will definitely improve your benchmarks.
EDIT: Michael, I see you beat me in - this time!