On Sep 30, 2009, at 8:20 PM, Lynn W. Taylor wrote: > Paul D. Buck wrote: >> Which means, for what little it matters, that those metrics >> violate the standard as defined in the BOINC documentation >> A moving average is not a standard ...it is however a way to >> deflate the credit awards over time.., why this is seen as a good >> thing has still not been explained. > > ... and it hasn't been explained because there is no reason to do so.
Actually I can think of several reasons to do so ... one is sufficient ... because if we state it is one thing in the documentation and then use a formula to calculate it that does not match that definition then the reason why should be stated of the definition changed. A second good reason is that I asked ... it would just a good if you asked, or anybody else asked ... but the fact that someone has asked should be sufficient ... The third is because credits are important to some. A stable cobblestone allowing one to compare a cobblestone earned in 2000 with one earned to day would be nice ... with deflation you cannot do that ... > The cobblestone standard says "you calculate credit by taking the > results of two standardized benchmarks and multiplying by time." Yep, that is part of the definition... how is Eric's deflationary multiplier that debases this an improvement? > Counting FLOPs gives us a more repeatable credit system, but it > still goes back to the definition of a Cobblestone, and like it or > not, Cobblestones are defined by Whetstones and Dhrystones. And I am not the one running away from that hard definition ... > If you'd thought it through, you'd realize that the standard (good > or bad) is not the moving average. The moving average is the rate > of conversion from FLOPs to "true" Cobblestones. > > You can argue that basing Cobblestones on Whetstones and Dhrystones > is stupid, but you can't argue with a definition. No, I cannot, and I not ... because as you stated above Eric's formula is not in the definition. I can't really get on the boards and in one sense since I don't do SaH it is a little moot for me, but others in SaH have demonstrated that the "pay" over time is slowly decreasing that means that with the same size tasks the same amount of work done the pay is less ... and Eric's write up on this states that this is an intended effect. So, he is the one breaking the linkage between the benchmarks and the pay ... not I ... I just asked, how is this a good thing? Your answer carefully avoided the actual question ... it is not about how Cobblestones are defined ... it is how a formulation that slowly debases the payment for work done is a good thing when that formulation simply proves that there is NO linkage between the benchmarks and the awards ... _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
