SOLVED GPU gets locked in a unstable machine cycle. HELP!

I just starting getting constant EUE's with only a very small OC on my core/memory. I had to back it down to almost stock to get it to finish more than 2 in a row without Unstable cores. It's odd because I done 100+ WU's with it at over 50mhz+ core OC than it is now without a single EUE. No idea why it started all the sudden. I thought maybe it was my unstable CPU OC, could be. I have a crappy MB and no HT multiplier settings so my HT link is way too high. I just ordered a PHII 550BE and BIOSTAR TFORCE TA790GX 128M combo from the Egg, I'm thinking it will help a tad

I also learned the hard way that you cannot change Core/Memory clocks while folding or even with it paused. As soon as you resume the WU it will go to unstable core.

The problem is that its not like the card even tries to fold. its that I literally start up the folding client and the it says:

Entering M.D.
Working on Protein
Run: Exception thrown during guardedRun
Run: exception thrown in GuardedRun -- gromacs cannot continue further.
going to send back what have done -- stepstotalG=0
work fraction=0.000 steps=0
logfile size=0 infolength=0 edr=0 trr=23
-writing 635 bytes of core data to disk...
Done:123 -> 124 (compressed to 100.8 percent)
... Done.

[email protected] Core Shutdown: UNSTABLE_MACHINE
coreStatus = 7A (122)
sending work to server.....

Blah blah blah.

It doesnt stop from an error... it just never starts. This is at stock clocks, superclocked clocks, 100000000000THz clocks, anything. It doesnt discriminate.

OK. Here we go again. It did 1 work unit last night, an 1888, and now is back in a cycle of wasting my electricty and making me hate folding.

heres the log.

[20:27:42] *------------------------------*
[20:27:42] [email protected] GPU Core - Beta
[20:27:42] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[20:27:42]
[20:27:42] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[20:27:42] Build host: amoeba
[20:27:42] Board Type: Nvidia
[20:27:42] Core :
[20:27:42] Preparing to commence simulation
[20:27:42] - Looking at optimizations...
[20:27:42] - Created dyn
[20:27:42] - Files status OK
[20:27:42] - Expanded 45408 -> 251112 (decompressed 553.0 percent)
[20:27:42] Called DecompressByteArray: compressed_data_size=45408 data_size=251112, decompressed_data_size=251112 diff=0
[20:27:43] - Digital signature verified
[20:27:43]
[20:27:43] Project: 5771 (Run 8, Clone 100, Gen 842)
[20:27:43]
[20:27:43] Assembly optimizations on if available.
[20:27:43] Entering M.D.
[20:27:50] Working on Protein
[20:27:50] Run: exception thrown during GuardedRun
[20:27:50] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[20:27:50] Going to send back what have done -- stepsTotalG=0
[20:27:50] Work fraction=0.0000 steps=0.
[20:27:54] logfile size=0 infoLength=0 edr=0 trr=23
[20:27:54] - Writing 635 bytes of core data to disk...
[20:27:54] Done: 123 -> 124 (compressed to 100.8 percent)
[20:27:54] ... Done.
[20:27:55]
[20:27:55] [email protected] Core Shutdown: UNSTABLE_MACHINE
[20:27:59] CoreStatus = 7A (122)
[20:27:59] Sending work to server
[20:27:59] Project: 5771 (Run 8, Clone 100, Gen 842)
[20:27:59] - Read packet limit of 540015616... Set to 524286976.

[20:27:59] + Attempting to send results [November 26 20:27:59 UTC]
[20:27:59] + Results successfully sent
[20:27:59] Thank you for your contribution to [email protected]
[20:28:03] - Preparing to get new work unit...
[20:28:03] + Attempting to get work packet
[20:28:03] - Connecting to assignment server
[20:28:04] - Successful: assigned to (171.67.108.11).
[20:28:04] + News From [email protected]: Welcome to [email protected]
[20:28:04] Loaded queue successfully.
[20:28:05] + Closed connections
[20:28:10]
[20:28:10] + Processing work unit
[20:28:10] Core required: FahCore_11.exe
[20:28:10] Core found.
[20:28:10] Working on queue slot 02 [November 26 20:28:10 UTC]
[20:28:10] + Working ...
[20:28:10]
[20:28:10] *------------------------------*
[20:28:10] [email protected] GPU Core - Beta
[20:28:10] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[20:28:10]
[20:28:10] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[20:28:10] Build host: amoeba
[20:28:10] Board Type: Nvidia
[20:28:10] Core :
[20:28:10] Preparing to commence simulation
[20:28:10] - Looking at optimizations...
[20:28:10] - Created dyn
[20:28:10] - Files status OK
[20:28:10] - Expanded 45461 -> 251112 (decompressed 552.3 percent)
[20:28:10] Called DecompressByteArray: compressed_data_size=45461 data_size=251112, decompressed_data_size=251112 diff=0
[20:28:10] - Digital signature verified
[20:28:10]
[20:28:10] Project: 5770 (Run 10, Clone 34, Gen 514)
[20:28:10]
[20:28:10] Assembly optimizations on if available.
[20:28:10] Entering M.D.
[20:28:17] Working on Protein
[20:28:17] Run: exception thrown during GuardedRun
[20:28:17] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[20:28:17] Going to send back what have done -- stepsTotalG=0
[20:28:17] Work fraction=0.0000 steps=0.

What the heck is the issue. I have tried everything short of sacrificing animals to various gods.

Edit: this is only on my 8800gts, my 285 and my cpu are folding away just fine.

Deleted the cores and everything. spent a good 4 hours trying to get it to work. Took card out of system, and never looked back. Either the card is dead/dying, or it just is not meant to fold. The 285 is gonna have to fold solo for now