Not a problem at all..A little history on that machine: It has run since last May with first a pair of 8 core Sandy Bridge E5-2687W's then for a short time a pair of Ivy Bridge 10 core E5-2670 V2's, then a pair of 8 core IB E5-2687W V2's and finally these 12 core IB E5-2697 V2's. Through all that time on one OS install and on the same install of WCG. The output on each type of cpu was always relevant to the others factoring in cores and speed untill day three with the 12 cores. At this point I'm going to say that there are many things I don't know in this world but I'm pretty "up" on what certain cpu's should and shouldn't do in WCG. The software of BOINC and it's design is my weak point but I'm very strong on the hardware end..This just didn't make sense to me and why I posted. There is something weird going on and it's either in the way the server is reacting to those WU I showed or??? That's why I wanted the genius's here to take a look at it. Just so we're clear, I welcome any comments because as I said before, thats how we all learn. If I disagree I will say so but I don't get upset at differing opinions..

Well, if it's the same installation of BOINC (I think I mean the same machine identifier, but I'm uncertain and more than happy to be corrected) then I agree that what you're seeing is strange. Even more so if you have seen the expected increase in performance each time you have upgraded the machine.

I'm probably two floors below the bottom of the barrel I've been scraping, but the only straw I've got left to clutch at is some sort of throttling built into the motherboard. (Good grief, that was really mixing metaphors!)

Is there any chance that there's some sort of (perhaps variable) throttling occurring due to high internal temperatures in the CPU(s)? Or something else along those lines? I would have thought that you would have seen it if you've been playing around trying to optimise things, but I have to ask!

Have you tried asking on your team forum? It sounds like the sort of issue your fellow speed freaks (no disrespect intended) might have fallen over.

I hope you get it sorted out -- it must be very frustrating.

A couple of good points have been made, actually along similar lines to my original 4th bullet. You talked about a "constant machine", but that can't really be the case at the moment because FAAH is actually two different apps, VINA and Autodock, and we have no control over which ones we get and when and the mix has been changing over time. These two different apps can behave very differently on different machine architectures. You just might be very unlucky with one of them. But due to the fact that WU complexity is also changing quite erratically at the moment I think you might have some difficulty pinning it down.

But still, it's worth investigating if you still have the historical data needed to generate some stats and you enjoy a challenge. Good luck!

If you look at the WCG global stats, it says that the average PPH of run time for Fight Aids @home = 95.58 and for CEP2 it is 151.04.

That is a huge difference.

Running all CEP2 for a comparison would prolly be a really bad idea on a beast with so many cores like yours Dave.

When a new project appears, I expect your points will rocket upwards.

Now that is what I call a proper challenge for you !

If Shistosoma comes back, you could prolly finish the project on your own. The average for that is even higher at 160.23.

Off to my kip. Its really late here.