If this can help in your discussion where I feel completely lost here some elements:

I am used to get in average 19.5% less points that what my system claims. I have all devices in W7 64 bit and all processors are of the core i7 types.

My last Results Stats at the moment look like:

Total FAAH WU = 941
of which:
-Valid 548
-Inconclusive - 12
-Pending Validation - 8
-In Progress - 353

I have no experience in mixing projects. I always run one project full at a time. When a target is reached I switch all machines at the same time by switching all machines to a new and same profile. Only during a very short transition of a few hours two projects mix, the finishing one and the new one. It is easier to track what happens when you have many machines all running on the same project. As the machines are similar it is much easy to correlate between them and understand what is the problem.
As a result since I started with Boinc frankly I nearly had no problem at all. The only ones I had where due to misunderstandings, or wrong actions from my side (did not read enough the manual).
When projects are mixed, I suppose that the behaviour becomes more difficult to understand when troubles and side effects appear.
So try running one project at a time first FAAH for one or two weeks and the HCC the same and then compare, credits, claims, problems etc..

I get always 20% less then what my machines claim desperately to WCG.

My machines know much better what they deserve, how much they suffer, the high thermodynamical stress I impose on them, the terrible ordeal they go through when I arrive at midnight to check their result. They know that if some did not reach their target an immediate quartz-clock-shock (QCS 100*) in term of an increase of CPU* frequency by 100MHz is imposed until the target is reached. Yes it is a sweatshop atmosphere were WU's must be produced at high rate. So my machines feel frustrated not being rewarded according to their claims and I have to defend them.

*This shock treatment goes from QCS 50 until QCS 1000 where death by CPU* arrest is a high probability, but some can survive for some time.

*Cardiac Power Unit

Hello All
If this can help, the only thing i changed was power management.
When i cam beack to my first power Scheme (maximun performance), credit was fine again.

Highwire,

I have 41 "Valid" single Quorum FAAH WUs run since my response yesterday. These have been run under XP 32, Vista 32 & 64 and Win7 32 & 64. CPU by processor/OS are:

OS WUs Min CPU Max CPU Avg CPU
AMD Duo (no HT) 2.41GHz
XP-32 1 6.3 6.3 6.3
XP-32 2 5.96 5.97 5.97

Intel Duo (no HT) 2.67GHz
XP-32 2 6.5 6.62 6.56

Intel Quad (HT) 2.66GHz
Vista-64 10 4.65 8.23 6.27
Vista-32 11 5.27 7.63 6.02
Win7-64 7 5.1 7.08 5.98

Intel Duo (no HT) 2.53GHz
Win7-32 1 5.65 5.65 5.65

Intel Duo (no HT) 1.83GHz
XP-32 3 7.46 8.43 8.01

AMD Single (no HT) 2.01GHz
Server2003-32 1 8.08 8.08 8.08

Intel Quad (no HT) 2.83GHz
Win7-64 3 4.21 5.07 4.75

Net is, I'm not seeing anything like you are. CPU time looks to hold to processor, not OS

Ok ran a pile of FAAH now on old OS. I'm letting a few HFCC finish then going back to W7. One thing I notice is I've different AV (I thought I'd installed same). I *DID* disable the W7 guard mode and it made no difference (and I use that one on another computer fine), I may uninstall completely. This wouldn't explain why other tasks aren' so affected though.
faah10729_ ZINC11850830_ xMut_ md03380_ 01_ 1-- Valid 4.31 84.5 / 84.5
faah10741_ ZINC11892053_ xMut_ md03380_ 00_ 0-- Valid 4.97 97.8 / 120.3
faah10738_ ZINC11888206_ xMut_ md03380_ 00_ 0-- Valid 4.91 96.6 / 119.7
faah10736_ ZINC11886031_ xMut_ md03380_ 00_ 0-- Valid 4.34 86.3 / 103.3
faah10736_ ZINC11885503_ xMut_ md03380_ 02_ 0-- Valid 4.34 86.2 / 105.3
faah10733_ ZINC11863096_ xMut_ md03380_ 01_ 0-- Valid 4.89 95.7 / 121.8
faah10728_ ZINC11849951_ xMut_ md03380_ 01_ 0-- Valid 4.78 93.6 / 113.5
faah10728_ ZINC11849187_ xMut_ md03380_ 01_ 0-- Valid 5.31 103.9 / 129.4
faah10722_ ZINC11822306_ xMut_ md03380_ 03_ 0-- Valid 3.76 73.6 / 91.2
- you can see what I mean when I said this computer is capable of knocking out FAAH at fair speed, it's easily more than 2x as slow under my W7 64.
EDIT: couple HFCC now in
HFCC_ s1_ 02795259_ s1_ 0001_ 0-- Valid 3.81 74.6 / 92.0
HFCC_ s1_ 02797665_ s1_ 0000_ 0-- Valid 5.24 102.5 / 129.3
q.v. W7 x64 (from earlier post)
HFCC_ s1_ 02660580_ s1_ 0000_ 0-- 11.81 234.6 / 132.5
HFCC_ s1_ 02646787_ s1_ 0000_ 0-- 12.55 253.7 / 136.1
HFCC_ s1_ 02578184_ s1_ 0001_ 0-- 12.50 252.7 / 142.3

again, easily > 2x slower, massive overclaim, whereas old OS, it tends to underclaim,
which if anything is I'd think demonstrates fast memory etc.