It is hard to make a global statement based on a few samples, and apply that to thousands and thousands of other setups, sometimes you can be right...and sometimes not. Took me just few hours with linx to find below OC points, and have checked some with prime, and like on my i940 I record tests.
4.2 HT on, vcore 1.28 pass linx 10 runs of 5100mb, 2 notches less fails 1st pass.
4.3 HT on, vcore 1.31 pass linx 10 runs of 5100 mb (max memory), 1 notch less failed 8th run of same, later confirmed 9 hrs prime stable blend.
4.4 HT on, vcore 1.34 pass linx 10 runs of 5100 mb, 1 notch less failed 6th run, confirmed later with 9 hrs prime stable. Lowered vcore 1 notch in bios, still says 1.34 cpuz load, and still passed prime 13 hrs.
4.5 HT on, vcore 1.38 pass linx 5 runs, 45 mins prime, did not run either longer 2nd temps.
On both my i940 and i950 I need at least same and when tested more vcore for linx at max memory, than prime.
Showing a prime blend and linx stable screenshot together, how would you know they did not simply choose a vcore that both would run at

I have a 13 hour run with prime blend at vcore that failed linx max memory.
Only way to prove anything would be to video a 13 hour run of prime at vcore where prime passed but linx failed, then keep video going and run linx and capture the bsod. But time demands vary considerably from my interest in the subject.
Since I have now heard quite a few arguing both ways, I assume it is possible for some cpu/mobo/setups to need more vcore for linx or prime, especially given difference between 2 programs and given extra heat of linx.
Or maybe I should get into the internet spirit, and say everyone is crazy that found it the opposite of my setup

.
Bookmarks