The Project Credit Comparison page (http://boincstats.com/stats/project_cpcs.php) is a great feature. In addition to the fields comparing one project to one different project (e.g. Rosetta@home vs. Einstein@home), would it be possible to have a field for each project showing its average credit per time compared to all other projects (e.g. Rosetta@home vs. Einstein@home, SETI@home, WCG@home, etc.)?

For example, the sum of each value in Rosetta@home's array divided by the length of the array (minus 1, to avoid self-comparison).

Comments from the peanut gallery-The biggest problem I see with Credit Comparison charts is it doesn't take into account projects with optimized apps,64bit apps and your ability to utilize such. The charts because of this may lie,but something is better than nothing

As I understand it, the charts (especially the second) are essentially showing how efficient each project's application is relative to those of other projects. It takes credits, which can equated with FLOPS, and compares that with seconds of CPU time.

So I don't think the Project Credit Comparison charts are lying at all; their purpose is to effectively quantify the difference between how optimized and robust different projects' applications are.

I can agree to a point.....why I think it is misleading is it may say x project gives the best credit.....but people may use this to decide which project to crunch when in fact project y may give them better credit performance based on their architecture ,OS,and whether they can get an optimized app to work (win may be easy but other OS can stump many).

To give another example: Windoze may kick butt on 1 project but Linux on another....how is this ever accurately displayed?

Yet another example-You are comparing a project using fixed credit vs a project using benchmarks....how is that factor ever compensated for? Through total output? You may be doing more flops on 1 project that actually grants less credit than another.

IMHO the 'best' way is to run project x vs y on the same host,all variables considered, and make your own judgement over a period of time and only use credit charts as a rough guide More refinement is just a waste of energy unless the 'true' breakdowns can be done.(Actually doing that would be a great Boinc project!)

More refinement is just a waste of energy unless the 'true' breakdowns can be done. (Actually doing that would be a great Boinc project!)

This thread demonstrates how contentious the cross-project stats problem has become. This issue has become the elephant in our living room.

Until some degree of sanity is introduced, the whole cross-project BOINC cobblestone system will become meaningless.

Maybe each project can give out what it deems as reasonable credit, but before they are applied to cross-project stats a correction factor would be applied.

. . . that way, 1 cobblestone of WCG or Rosetta would take equal effort as the real easy projects [we all know which ones]

That would take the steam out of some of the cobblestone mills.

Perhaps WUProp or WUProj or whatever they are called will implement this. It seems they have taken a step in that direction.