Message boards :
Number crunching :
Credit Calculations
Message board moderation
Author | Message |
---|---|
Mark Grosser Send message Joined: 18 Oct 99 Posts: 34 Credit: 1,211,299 RAC: 0 |
I have started running Seti@home again after about a 4 year absence. I just have two questions about things that might have changed. The first I am sure has been asked many times but a search of the message boards only got hits on brief mentions of it. I used to be under the impression that credits were awarded based on the number of computations required for that work unit. With that thought process a unit that takes a long time to process would get more credit than one that takes a short time. Two different units which take about the same amount of time would receive about the same credit. Since returning and looking at the units I got credit for a found the credit varies for work units taking about the same amount of time. Two different work units crunched on the same computer which took almost the same amount of time resulted in one getting 95 credits and the other received 165. That seems like a big difference for units that took about the same amount of time to process. I am not a power cruncher and the days are past where I can have the latest most powerful computers on the market. I won't lose any sleep over how much credit I get. I am just happy to have my computers contributing something again. I am just curious how the credit is calculated now. My second question is about the validation process. It used to take 3 computations to form a quorum for validating a unit. I now see that it only takes two and if the first two differ a third unit is sent out. What triggered that change? Thanks... |
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
IIRC it always has only taken 2 to validate. Others may be able to explain how credit is awarded, that remains a dark mystery to me and many others. |
Mark Grosser Send message Joined: 18 Oct 99 Posts: 34 Credit: 1,211,299 RAC: 0 |
IIRC it always has only taken 2 to validate. Others may be able to explain how credit is awarded, that remains a dark mystery to me and many others. Yes but it seems to me they would send out the same workunit to 3 different computers at the start. Now it appears it is only sent to a 3rd computer if the first 2 return different results. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Hello Mark, This is like walking on glass, tread lightly, lol http://setiathome.berkeley.edu/forum_thread.php?id=79418&postid=1778812 This will get you started on the whole credit thing. Zalster |
Mark Grosser Send message Joined: 18 Oct 99 Posts: 34 Credit: 1,211,299 RAC: 0 |
Hello Mark, Trust me despite me leave of absence I have been around long enough to know how touchy the subject of credit calculation can be. Thanks for the link Zalster |
W-K 666 Send message Joined: 18 May 99 Posts: 19062 Credit: 40,757,560 RAC: 67 |
IIRC it always has only taken 2 to validate. Others may be able to explain how credit is awarded, that remains a dark mystery to me and many others. In the beginning Seti sent out 4 tasks on initial replication. Validation was done after 3 tasks returned. If Validation successful, credits were calculated for all three tasks using cobblestone method. If Validation not successful, waited until fourth task returned, credits were calculated for all four tasks using cobblestone method. Extra tasks were sent out if necessary. Granted Credits were calculated by throwing out, top and bottom claims, and average calculated of remaining tasks became granted credits. If there were more than one linux hosts then credit granted was usually <5 CR when Windows hosts claimed between 25 and 30 credits. Benchmark calculation was, and possibly still is, very low on Linux hosts. This is the main reason why cobblestone credits were dismissed. |
rob smith Send message Joined: 7 Mar 03 Posts: 22202 Credit: 416,307,556 RAC: 380 |
When a Work Unit is first sent out to us it is as TWO tasks, to different people. Now comes the fun: When the two results come back they are compared with each other. If the results are sufficiently similar then they are declared valid, credit is awarded. If one of the results is late back then a third task is sent out to someone else. If one of the results is an "error" result then a third task is sent to someone else. If the first two task do not agree they are declared "inconclusive", a third task is sent out to someone else. And so it repeats until either there is a pair of tasks that validate, or the number of tasks generated reaches the limit defined, I can't remember what those limits are, and they can be defined (potentially) per-Work_Unit. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
And yes, your memory is correct: there was an intermediate step between WinterKnight's answer and Rob's answer, when the initial replication was 3. |
metalius Send message Joined: 8 Oct 02 Posts: 48 Credit: 32,239,717 RAC: 15 |
... lol Yes, of course - LOL. SETI credits are lower and lower and reached a milestone, marked as absurdity. CreditNew "rules". Since 2005 I participate in SETI/Einstein by principle fifty-fifty. Just take a look to my signature below... If just try to multiply SETI RAC by 10... Yeezzz, that is almost exactly Einstein RAC. :oD |
Mark Grosser Send message Joined: 18 Oct 99 Posts: 34 Credit: 1,211,299 RAC: 0 |
Thanks for the answers everyone. I read the thread that Zalster posted the link to and in there I found a link to the wiki explaining Creditnew. I can see why people don't like it. The engineer in me just wanted to know how it was being done. As long as I know my computers are returning valid results I am happy. I do like the fact that you can now look at the task result details and compare them to the other computers. The fact that I have 1 task as inconclusive made me figure out roughly how the current validation process works. I don't think we were able to do that 4 years ago. As for the validation process my thoughts were confirmed as to the numbers. Does anyone know why it was changed from 3 to the current 2 tasks (terminology corrected from my first post) initially sent out? |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
Does anyone know why it was changed from 3 to the current 2 tasks (terminology corrected from my first post) initially sent out? I think it was a case of it wouldn't affect the science adversely, but it would reduce the size of the database & the load on it's associated servers(s). Grant Darwin NT |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Does anyone know why it was changed from 3 to the current 2 tasks (terminology corrected from my first post) initially sent out? And to get more work done, more quickly. The hosts which were previously wasting time on the redundant third replications could work on the first or second copy of a new task instead. With improved consistency of returned results, you can get three jobs done for the price of two. |
AMDave Send message Joined: 9 Mar 01 Posts: 234 Credit: 11,671,730 RAC: 0 |
Links you may find helpful:
>Â Â Message 1771187 (and its follow-up) >Â Â Computation credit >Â Â BOINC FAQ Service |
©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.