Message boards :
Number crunching :
Not Understanding what Credit is...
Message board moderation
Author | Message |
---|---|
Seaworthy Comic Send message Joined: 21 Aug 05 Posts: 8 Credit: 175,873 RAC: 0 |
Can someone just clear this up for me a bit; I can't find anything on the web that makes much sense. So I understand that the amount of credit one gets for a work unit depends on the amount of CPU time used to process it. Doesn't that reward people who have slower CPUs? It doesn't seem to take into account that faster CPUs process more data units and therefore are more valuable. If I am completely wrong or if there's something I'm missing or not thinking about, please let me know! Thanks! 4.26 Ghz total crunching power! |
Speedy67 & Friends Send message Joined: 14 Jul 99 Posts: 335 Credit: 1,178,138 RAC: 0 |
Can someone just clear this up for me a bit; I can't find anything on the web that makes much sense. Take a look at "computation of credit" in the Boinc Wiki.. Greetings, Sander |
cjsoftuk Send message Joined: 3 Sep 04 Posts: 248 Credit: 183,721 RAC: 0 |
Theory has it that a 450Mhz CPU that spends 10,000 CPU seconds doing work will get as much credit as a 2.66Ghz CPU that spends 10,000 CPU seconds doing work. In reality the values differ slightly, but on average it is about right. Take this example: 450Mhz CPU = 15,000 CPU seconds/WU 2.66Ghz CPU = 5,000 CPU seconds/WU The 2.66 Ghz machine will run 3 WUs in the time it takes the 450Mhz to run 1, and the 450Mhz will claim 3 times as much as the 2.66 Ghz per WU. This balances it out. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Can someone just clear this up for me a bit; I can't find anything on the web that makes much sense. It's based actually on the number of floating-point operations. That is estimated by a measure of your CPU speed (from the benchmark) and wall-clock time. Granted credit is then based on the middle-speed of all the machines that report (drop the fastest and slowest, average what's left). That said, you'll read more about how the system is unfair than anything else, how some machines crunch better than they benchmark, or benchmark better than crunching, but I think it all averages out sooner or later. |
Seaworthy Comic Send message Joined: 21 Aug 05 Posts: 8 Credit: 175,873 RAC: 0 |
Thanks folks, that clears a lot up to me! 4.26 Ghz total crunching power! |
Tern Send message Joined: 4 Dec 03 Posts: 1122 Credit: 13,376,822 RAC: 44 |
Theory has it that a 450Mhz CPU that spends 10,000 CPU seconds doing work will get as much credit as a 2.66Ghz CPU that spends 10,000 CPU seconds doing work. No. The theory is that a 450MHz CPU that is otherwise identical (cache, instruction set, EVERYTHING) to a 2.66GHz CPU will take almost 6x longer to do the SAME WU, but will get the same credit for it. So if the 2.66GHz CPU takes 10,000 seconds to do a WU and gets 20 credits, the 450MHz CPU will take 60,000 seconds and will get 20 credits. Otherwise, slower computers would get more credit per result, although doing them slower, and that is exactly what the benchmarks are there to prevent. The formula and all the details are in the Wiki. The plan is for everyone to get the same amount of credit for the same amount of work, not for the same amount of time spent doing the work. There are problems, as the benchmarks don't measure exactly the same things being done by the CPU as when crunching a result, and they aren't terribly accurate to begin with. Thus all the discussion about improving them or replacing them with something better. |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Can someone just clear this up for me a bit; I can't find anything on the web that makes much sense. Disappointment sets in ... and I thought we were so clear in the Wiki ... Seriously, though we are not perfect, the place to start with most of these questions is the Wiki ... we add to it each day, and if I am real well, we add a LOT ... but, fundamentally, the Wiki is a collection of what we know, gathered since mid-BOINC Beta test. The best news is that this is no longer the work of only one person (used to be me only), but instead has nearly 25 editors, each with different skills and knowlege, and we build on the other's work. So, John may add a new page, I will add a little, refine his words, and someone else will follow me ... And, many people send me e-mails with suggestions, I roam the boards on projects and when I see something good I "steal" it and add it to the Wiki ... for example you may ask a new question and someone like Ingleside will answer it with a good explanation that really does explain it well. So, I copy that over and try to make it read as well as I can ... And, then we refine it ... |
©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.