Fair point regarding 'over' rewarding with credits. I sort of wonder if the extension of the logic might review why SETI (and Einstein) have 'devalued' their work units during this past cycle (by about 10% or so). Is there a thought that other projects need more participation?

One thing Climate does to deal with the long cycle work units they have (even their shortest work units are two or more times the length of Astropulse) is to allow 'trickle' awards on their in process work units. Not only does that provide more frequent 'rewards' regarding the process of the work units, but it compensates folks for work units that go 'bad' for no fault of the client. Just a thought here.

Offering more credits would only cause problems with other BOINC projects that are not offering as much, and would encourage them to higher the amount of credits that they offer to attract any number of lost users. If every project did this, there would be a credit inflation problem which would devalue each contribution.

---

Fair point regarding 'over' rewarding with credits. I sort of wonder if the extension of the logic might review why SETI (and Einstein) have 'devalued' their work units during this past cycle (by about 10% or so). Is there a thought that other projects need more participation?
...

24 days ago, Eric posted graphs in the New Credit Adjustment? thread which show why the S@H adjustment is downward. Note in the second graph that when 5.27 was released here about a year ago, the overall granted credit for this project shifted above the true cobblestone rate. If you remember, the 5.27 release had a client-side multiplier downward adjustment but the data indicates it didn't decrease enough. The server-side adjustment has probably removed that unintended bonus by now, and the rate should remain stable in future (barring release of a BOINC client which produces significantly different benchmarks).

I don't do Einstein, but I'll note that the credit_per_cpu_second statistic was added to BOINC by Bruce Allen, and he recommended the kind of comparison display of that data to be used at stats sites. That suggests Einstein adjustments might be based on those comparisons.

                                                                Joe

One of the things I read some months back over in the Einstein newsgroups had to do with what seemed to be something of a 'work measurement penalty' for AMD processors. I'd seen some anecdotal results to suggest this might be going on and as a result did some resource shifting. Then, with the issues of the past month for SETI, I've done a bit more of that. By now, my big players are Spinhenge, Malaria, Climate, Rosetta, with my lesser players being World Grid, SETI, and Einstein. That's almost a reverse of my accumulated numbers which at the moment still are Einstein, SETI, Climate, Rosetta, World Grid, Spinhenge, Malaria. Forecasting things out to the end of the year, it looks something like Climate, Einstein, Rosetta, Spinhenge, SETI, World Grid, Malaria.

The thing is, I've done SETI from back in the old days -- something like 7 and half years of it.

I don't do Einstein, but I'll note that the credit_per_cpu_second statistic was added to BOINC by Bruce Allen, and he recommended the kind of comparison display of that data to be used at stats sites. That suggests Einstein adjustments might be based on those comparisons.

                                                                Joe

---