Project: 10632 (Run 12, Clone 45, Gen 48)

[uncle fuzzy]

Here's another of the dreaded immediate failures with a 10632.

Code: Select all

[16:44:58] Folding@Home GPU Core -- Beta
[16:44:58] Version 2.09 (Thu May 20 11:58:42 PDT 2010)
[16:44:58] 
[16:44:58] Build host: SimbiosNvdWin7
[16:44:58] Board Type: Nvidia
[16:44:58] Core      : 
[16:44:58] Preparing to commence simulation
[16:44:58] - Looking at optimizations...
[16:44:58] DeleteFrameFiles: successfully deleted file=work/wudata_07.ckp
[16:44:58] - Created dyn
[16:44:58] - Files status OK
[16:44:58] sizeof(CORE_PACKET_HDR) = 512 file=<>
[16:44:58] - Expanded 28912 -> 163067 (decompressed 564.0 percent)
[16:44:58] Called DecompressByteArray: compressed_data_size=28912 data_size=163067, decompressed_data_size=163067 diff=0
[16:44:58] - Digital signature verified
[16:44:58] 
[16:44:58] Project: 10632 (Run 12, Clone 45, Gen 48)
[16:44:58] 
[16:44:58] Assembly optimizations on if available.
[16:44:58] Entering M.D.
[16:45:04] Tpr hash work/wudata_07.tpr:  1901016416 945023205 1605366722 3914600737 2752169938
[16:45:04] Working on 582 p2750_N68H_AM03
[16:45:04] Client config found, loading data.
[16:45:05] Starting GUI Server
[16:45:07] mdrun_gpu returned 
[16:45:07] NANs detected on GPU
[16:45:07] 
[16:45:07] Folding@home Core Shutdown: UNSTABLE_MACHINE
[16:45:10] CoreStatus = 7A (122)
[16:45:10] Sending work to server
[16:45:10] Project: 10632 (Run 12, Clone 45, Gen 48)
[16:45:10] - Error: Could not get length of results file work/wuresults_07.dat
[16:45:10] - Error: Could not read unit 07 file. Removing from queue.

I've completed hundreds of this series, including quite a few 10632s, but the times I have seen this problem of 5 immediate deaths, lately it's always with a 10632. I've had the same problem on different cards in different machines.

For this particular issue: XP32, GTX460@1650 shader (for 2 months), also running an 8800 GTS 512 and smp on Q6600@3.4

[sortofageek]

Hi, uncle fuzzy!

So far, two have returned results for this WU. One got full credit, the other received none.

[uncle fuzzy]

I kinda figured that. I just wanted to point out, again, that something is a little "different" about this WU. Many people have seen this problem and others have run them. It looks like they are being over-issued (each PRCG to multiple clients). I crashed, 2 finished it, only one for points. Strange Project.

[bruce]

When folks have an error, a new copy gets issued. I think this can cause your "over issued" problem.

There may be a correlation with Win-64 and/or specific drivers. What are you running?

[uncle fuzzy]

XP32, 258.96

I had these lines on all 5 failures, so did I send any error report?

Code: Select all

[16:45:10] Sending work to server
[16:45:10] Project: 10632 (Run 12, Clone 45, Gen 48)
[16:45:10] - Error: Could not get length of results file work/wuresults_07.dat
[16:45:10] - Error: Could not read unit 07 file. Removing from queue.