HI BDAqua,
You and I had a thread together on the issue of high temps on CPU A. I could not use the howto link you sent me once I got in because my model was different than that author's. Fortunately my model was far easier than that one and I posted a how to here. I think i called it quad 2.5 thermal paste. I have not yet run thermal recalibration yet as Arctic silver(all versions) specifies a break in period that can be as long as a month for some chips. I suspect that if I run it now, it will quit without making changes because of the 8 C disparity of the dies at idle. At fifty percent load for extended periods(overnight) in a non air conditioned room the disparity drops down to about 2-3 C 52ish on CPU A and 56 ish on CPU B. The fans, however, are running at half speed(annoyingly loud).
In answer to your question: No, I did not change the location of each board as I did one at a time to have an adjacent reassembly model. I did, however, mess up and boot the machine without the nvram battery and it booted but ran with even higher fan speeds. I did run the full battery of OF tests and it checked out fine save two bad pieces of memory. It balked twice at running the CPU A pump fan test but passed on the third go. The machine now boots to normal fans but they instantly rev up as you login. I sent an email to Arctic silver begging them for PPC application instructions so I don't expect much from that direction.
On the bright side, If I do this again I will make a pictorial howto and see if one of the mac devoted sites will publish it. This morning i realized I do have a way to confirm your memory of CPU A being on the upright topside. I can just run the CPU A and CPU exhaust fan OF test and should detect with my hand each fan shutting down and then restarting. I was just hoping someone had a link as this job, while ultimately pretty easy)on my model, is not one I care to keep repeating. CPU A is fine and I want to leave it alone and only focus on the one that is now out of line.