I am running CPDN along with Seti and LHC on an AMD Athlon XP 2000+ computer. Each trickle on CPDN takes almost exactly 12 cpu hours, so for each hour's trickling I'm getting 6.25 credits (75/12). When I look at my stats on SETI and LHC each CPU hour is worth around 10 credits. Maybe the figure of 0.007 credits/timestep needs to be adjusted! 0.01 perhaps!
<img>

hmm, that's interesting, I wonder if with the newer versions and improved benchmarking the .007 is being "underestimated" now?

&gt; hmm, that's interesting, I wonder if with the newer versions and improved
&gt; benchmarking the .007 is being "underestimated" now?
&gt;
&gt;

SETI and LHC take an ensemble style approach to granting credit to individual results.

The actual credit granted is generally an average of the highest and lowest valid results requesting credit.

LHC has recently rewritten their validator to grant 50% credit to those returned results that are not truly invalid but also do not agree with the ensemble.

Thus, credit granted at other projects can not be reduced to a constant per cpu hour.

It has been noted in several threads at SETI that slow machines request much greater credit than the same WU processed by fast machines.

At LHC their science package has "uncovered" major differences in floating point percision among the various cpu/motherboard/OS combinations. This has twice force LHC to rewrite their validator for a broader interpertation of what is a valid result.

The best way to think of credit granting between CPDN and the others is the difference between a flat tax and a graduated tax system.