update, I've found out that MBM with an updated DLL can directly read the t-bred's in die diode, so far so good.
After aquiring an older stick of corsair XMS 3200 (cas 2.5) I've settled on speed of 2.362gig, 175x13.5,1.85v for the waterblock testing.
The setup is an iwaki MD 20RZ pump with evaporative cooling and a 20L res(with my airconned room temps stay pretty constant)
The fist thing tested was the micro pin block, still in "original" condition from when I made it despite thinking of ways to improve it in the mean time (that will come soon)
With a water temp of 22c degrees the (reported,MBM) idle temp was 29c and load (using cpu burn) was 34c.
Next was my all copper "reference" block, which is based on the WW principle, but with the channels turned 90 degrees.
it handily outperformed the micro pin block, with the same water temp, it's idle was 23c and load was 27c (obviously the on-die diode is reporting a temp somewhat lower than in reality)
With that evidence in hand, I performed the previously thought of mods to the micropin block , pin area was reduced to an area about 2mm larger than the core of the t-bred, 7x6 pins down from the previous 13x9. I also filled either side of the slot under neath the centeral barb with epoxy, reducing the size of the inlet from 3mmx21mm to around 3mmx6mm.
Now I have a far higher volecity jet that is better focused on the pin area, and the removal of the pins outside the "critical" area over the core (where they're doing next to nothing) allow for a more un-obstructed water flow away from the die area.
Now to see if I made it better or worse
The results were nothing short of impressive, it went from loosing to the reference block by 7c, to out performing it by 1c
not bad, not too bad at all
