Benchmarks are meaningless, as they are easily modified with a simple text editor.

Subjectively, HDC seems to generate the highest median per hour and GC the lowest. The sighted architecture but also the choice of project......people with power-boxes electing the heavier work plays a role.

With FA@H, the large majority of work done on that project, there is no telling. Sometimes my 10.2 stock per hour is the lowest, sometimes close to median, sometimes way up. Pure chance of PC mix meeting up in the quorum. Having tracked it for a while (to learn understand the discussions), just know that the overall median seems to credit me slightly more than is claimed. Opposed, there will be, that overall get less....such is life which we try to save with all our crunching.

Once HPF2 starts on BOINC, it will be a different ballgame. Groupings of 25 Work Units, with credits awarded / median computed from the 15th WU returned....16-25 will get that median value as I understand. We'll see.

I guess that the definitive test for Keith would be to look at his FA@H quorum results and see where he ranks in the claims as a median. If this is compared to His median claim for other projects and it is observed that his FA@H claim is lower than the others, it would be fair to say that his average score would indeed be higher when working on FA@H.

It works like this... Ones computer is benchmarked to produce say 10.6 PPH. Regardless of the project one works on the computer claims 10.6 PPH. If the computer is better than average at doing a project it will complete the assignment quicker than would otherwise be expected. i.e. If the benchmark indicates completion in 10 hours but is completed in 9, instead of claiming 106 points it will claim only 95.4 points. When this score is placed into a quorum it will be awarded (106+106+95.4)/3 = 102.46 points.

This gives the low scoring machine a 7% increase in points raising the effective PPH from 10.6 to 11.34 PPH and one would observe an overall increase in production when crunching this project from an expected 254.4 PPD to 272.1. Not a lot these figures are conservative. If a high claiming machine were to be included in the example the impact would become far more dramatic.

Those who cheat in order to claim artificially high scores benefit the low claimer far more than they do themselves, weird justice in a way.

Cheers. ozylynx

Hi Keith

Yes your claimed points should be constant/hour regardless of the project. At least between benchmark updates.

Cheers. ozylynx