Wow, bad situation.
I flashed a working win 7 x64 setup with the DX58SO and the ex965 to bios rev 5020 but in my case it would boot afterwards.
I'm waiting for the next rev bois to continue my troubleshooting with the 5020 problems I've seen.
The worst is you don't know if your troubleshooting has zapped the cpu.
So, step one would be to separate the ex965 from the mb. Then try to boot / repair the bios with a known good cpu.
Be sure to handle that ex (pensive) 965 with tlc.
You can do the usb / F7 boot with no cd drive.
You could check the mb processor socket pins for damage.
I'd check all the power connectors for good seating. Including that sorry 4 pin Molex next to the first Pcie slot.
So, to be clear, you ran win 7 setup all the way with the 965 / 5020 combo, with it's restarts, then it stopped
I'd post some questions about the * watchdog / failsafe * operations to see if it ever cuts off the boot completely
in certain situations.
And maybe post some questions about microcode update problems.
If this path is closed, ( no replacement cpu ) I guess all you could do is beat on Intel support for replacements.
As for the ex980 I'd sure make sure I had a board that would boot to 5020 first.
With the problems posted here, I' say that the 5020 bios wasn't very safe yet, and maybe waiting for the next rev
Intel bios might be good.
With as much money as that Gulftown is at stake, I'd think a cheapo i7 920 as a test pony would be a good investment.
PS Unless you've got a really great relationship with Intel support.
- 965 was OK.
- DX58SO was hosed. Even battery removal and CMOS recycling would not resurrect this motherboard.
- Installed 5020 or later BIOS using USB-floppy install method on a new DX58SO. Then Gulftown installed and runs fine.
- The Gulftown is a prodigious processor. It routinely generates 20-25,000 points per day in Folding@Home while also driving two GTX 460 graphics cards.