I'll leave it to the techs to give you an authorized answer, but I already have two questions.

1. What makes you think that BOINC knows the specs of all processors participating to grid computing?

2. Why do you think that what is true for FAAH is not true for all other projects? I think that all applications need more memory than the cache size allocated to a single core, therefore all should benefit from the largest cache.

For there to be CPU matching, server side with HCMD2 already quite expanded, there have to be groups created and the finer the groups, the unhappier the techs become [it being a manual exercise to populate the group tables and assigning CPU's]. On top the more CPU groups the slower results will move through the system, eventually to a point that validation grinds to a halt.

Then, let's not talk about the highly arbitrary... "oh you're a CPU-B, you're not getting an FAAH from me"... or am I misunderstanding?

Sorry but anyone who's got for instance the right version of WinLuxMac will get anything he/she likes to crunch at the project weights WCG hands results out... up until type A of DDDT-2 as those will have to have oodles of internet bandwidth.

As always, something will have got lost in translation.

PS: WCG techs are working on assigning heavy jobs to heavy gear, so something is coming, next year... see posts by knreed.

Well, thanks for your willingness to discuss this!

And just to address your 'over-claiming' comment, the points don't really matter to me. I just want to get as much work done as fast and efficiently as possible!

And while I realize that it is quite easy for us 'addicts' (well, at least I'm addicted... ) to manually tune our hardware and find which projects are best for which computers, I'm looking at this from the others point of view. The majority of the grid (in my head anyway) are 'set and forget' users. They install Boinc and never look at the project again, except to see how far they have come every once and a while. Nothing against that! - but if we can get more work out of their computers then all the better! Just trying to brainstorm the best way to do this.

And to sekerob's comments - Do the CPUs currently have to be matched? I was under the impression that they could just send the WU to whatever CPU requested work and met the minimum requirement for that project...?

Also, what do the techs manually have to add? If it's just updating the list when new CPUs come out, that would not be to difficult or time consuming... Even some volunteers could do it if we were asked. Or at least we could gather the data and post it for quick overview and input by the techs later. (Along with the computers, you also have a vast resource of very willing volunteers, let us not forget that!)

Then to your concern about people not being able to get the tasks they want... I was not trying to suggest that at all. This is all run by volunteers who in my mind have the power to crunch whatever they so desire, I applaud them for joining! I was just suggesting that if there are multiple checks selected, to set the servers to distribute the work as efficiently as possible. Some others on the XS forum have suggested making a check box similar to the 'receive work from other projects' option. Kinda like a 'request work best for my system first' option. That is also kinda inline with my suggestion if that appears to be better to satisfy the masses?

As always, thanks for your time!