SLI Problem

I´m using 2 Nvidia GTX 460 in SLI and it seems as if the FAH Tracker uses the same WU for each GPU. When the WU is finished i get the following error on GPU0:

[16:54:13] Working on 2 PEPTIDE (1-42)
[16:54:13] Client config found, loading data.
[16:54:13] Starting GUI Server
[16:54:13] Setting checkpoint frequency: 500000
[16:54:13] Setting checkpoint frequency: 500000
[16:54:31] Posted data.
[16:54:31] Initial: 0000; - Uploaded at ~106 kB/s
[16:54:31] - Averaged speed for that direction ~106 kB/s
[16:54:31] - Server has already received unit.
[16:54:31] + Sent 0 of 1 completed units to the server
[16:54:31] - Autosend completed
[16:55:41] ***** Got a SIGTERM signal (2)
[16:55:41] Killing all core threads


When I start GPU 0 again i get the following:

16:55:49] Gpu type=3 species=30.
[16:55:49] Loaded queue successfully.
[16:55:49]
[16:55:49] - Autosending finished units... [December 23 16:55:49 UTC]
[16:55:49] + Processing work unit
[16:55:49] Trying to send all finished work units
[16:55:49] Core required: FahCore_15.exe
[16:55:49] + No unsent completed units remaining.
[16:55:49] - Autosend completed
[16:55:49] Core found.
[16:55:49] Working on queue slot 02 [December 23 16:55:49 UTC]
[16:55:49] + Working ...
[16:55:49] - Calling '.\FahCore_15.exe -dir work/ -suffix 02 -nice 19 -priority 96 -nocpulock -checkpoint 3 -verbose -lifeline 2772 -version 630'

[16:55:49]
[16:55:49] *------------------------------*
[16:55:49] Folding@Home GPU Core
[16:55:49] Version 2.15 (Tue Nov 16 09:05:18 PST 2010)
[16:55:49]
[16:55:49] Build host: SimbiosNvdWin7
[16:55:49] Board Type: NVIDIA/CUDA
[16:55:49] Core : x=15
[16:55:49] Window's signal control handler registered.
[16:55:49] Preparing to commence simulation
[16:55:49] - Ensuring status. Please wait.
[16:55:51]
[16:55:51] Folding@home Core Shutdown: CLIENT_DIED
[17:01:01] teps (3%).
[17:02:40] Completed 2000000 out of 50000000 steps (4%).

So the GPUs basically do the same work. The client Stats for GPU 0 won´t show any finished WU. Any idea for that Problem?

AFAIK SLI has no effect on the F@H GPU3 Beta client, since it's not used. My 2 GTX460s are running fine as GPU0 and GPU1.

It looks like you may have old work in queue which may be causing the problem.

I would nuke the whole Tracker folder and do a fresh install with the current version (3.44). Tracker makes it too easy... ;>

You didn't post quite enough of your logs to see what the client was doing on startup, which may be relative since the client crashed!

Post back what you find.

Dana