New twist on an old issue!
It wasn't the board! It was the RAM!
Got it in the ASUS and ran Memtest all night again.
no issues.
Installed windows (same as with the DFI / as I didn't install XP before the upgrade, I had to reinstall Vista again before I could validate / that's when the issues started).
So, I ran Sandra and SuperPi and .... a few others.
Temps were all fine.
Sanded the CPU and used some Arctic Silver Ceramique with my old heat sink ... dropped 18 C baby!
Then decided to up the FSB as the RAM is DDR2 1000.
So, I kicked the FSB up to 250 and cut the CPU multi down to 11.
Set all the voltages cut the HT multi to 4X, did this that and the other.
Memtest had errors out of the gate.
Dropped the bus to 230 and all others the same.
Ran till test 6 and errored in the last 90%.
Dropped the RAM to 210 ......... same issues in the last 90% of test 6.
Adjusted the clock skews, upped the voltages, repeated this that and the other ...
Still won't pass #6.
Dropped to 201! and let it run over night. Set everything else to where it would be at default 5-5-5-15 timings, HT Multi @ 5X, CPU at 15X for 3.015, voltages at just a hair over stock.
***** I must say, the voltage options are phenominal in the Crosshair! ****
Anyway, let it run all night ... 2 or 3 errors the next mo.
~ 205 ... errors in test 6!
Worst case, 100% stable at the default 'Auto' settings in the BIOS.
The RAM was used / free anyway.
I have to assume that the RAM is crapped out and won't run at higher than DDR2 800
Will have to see what the board will do with good RAM later, when I have some more $$ burning a hole.
I still think I can get that CPU to 3.5 with air. But will have to hold off.