GA-X48-DS5 unstable on anything but stock in dual-channel mode

[Sigtran]

In single channel, no problems whatsoever, in dual-channel lockup under load (Prime95 for example, but even in memtest 2.01 test#4) with same bios settings. In single-channel mode, with all voltages on default, can clock my Q9550 to FSB od 412MHz (3.5GHz) and remain stable. In dual-channel mode, it is stable only to 340MHz, more then this, windows boots-up but locks afterwards within minutes (even with the CPU multiplier lowered).

Strange thing is that when in dual-channel I boot-up on stock (8.5x333) and overclock in Windows (using setFSB or EasyTune), board is stable until FSB of 395MHz, but setting then same (or even lower, for example 360MHz) FSB in BIOS is NOT stable, i.e. lockup under load happens!!! WTF!

Memory is BTW on the supported and tested memory list for this board - Kinston KHX6400D2ULK2/2G, and worked without problems on X38-DS4 and Asus Rampage Forumla (X48) board in dual-channel mode. Tried different slots, different memory, always same problem.

I am not alone with this issue, quite a few people having the same problem
http://www.ocforums.com/showthread.php?t=557945

BTW, I am not a novice in overclocking, I managed to OC this Q9550 to over 4GHz on AIR on Rampage Formula (quite lot of tweaking experiance needed for this), but this X48-DS5 is giving me headaches.

Any idea? I suspect some BIOS issue, but not quite sure. I have logged the case on GGTS on 25th of May, but no response yet (emailID 612404).

[Sigtran]

Gigabyte recently supplied me with the latest beta BIOS for X48-DS5, F6e, and situation is improved.

Now, using the same settings as before, overclock in dual channel is 100% stable somewhere up to FSB of 436MHz , giving me about 3.7GHz CPU frequency. Whatever they have changed in BIOS, it sure improved Q9xxx overclocking...