I've being crunching for S@H for a couple of years CPU only, recently got "the bug" and got me a couple of GPUs to speed up the process.

I set up a unit which has sucessfully been running mainly for GPUGrid, and I would like to use it for S@H also, the problem is most of the GPU WUs (SETI@home Enhanced v6.10 (cuda_fermi)) I get run for about 15 seconds and get aborted.

The error messages are:
--------------------------------------------------
<message>
Incorrect Function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
setiathome_CUDA: Found 2 CUDA device(s):
Device 1 : GeForce GTX 680
totalGlobalMem = -2147483648
sharedMemPerBlock = 49152
regsPerBlock = 65536
warpSize = 32
memPitch = 2147483647
maxThreadsPerBlock = 1024
clockRate = 1124000
totalConstMem = 65536
major = 3
minor = 0
textureAlignment = 512
deviceOverlap = 1
multiProcessorCount = 8
Device 2 : GeForce GTX 680
totalGlobalMem = -2147483648
sharedMemPerBlock = 49152
regsPerBlock = 65536
warpSize = 32
memPitch = 2147483647
maxThreadsPerBlock = 1024
clockRate = 1124000
totalConstMem = 65536
major = 3
minor = 0
textureAlignment = 512
deviceOverlap = 1
multiProcessorCount = 8
setiathome_CUDA: CUDA Device 2 specified, checking...
Device 2: GeForce GTX 680 is okay
SETI@home using CUDA accelerated device GeForce GTX 680
setiathome_enhanced 6.09 Visual Studio/Microsoft C++
libboinc: 6.3.22

Work Unit Info:
...............
WU true angle range is : 0.429018
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_GetPowerSpectrum 0.00221 0.00000
v_ChirpData 0.01775 0.00000
v_Transpose4 0.00266 0.00000
FPU opt folding 0.00387 0.00000
CUFFT error in file 'd:/Projects/SETI/seti_boinc/client/cuda/cudaAcc_fft.cu' in line 62.

</stderr_txt>
--------------------------------------------------------

The unit is a:
i7-2600K (8 cores)
Win 7 Ultimate 64
2 EVGA GeForce GTX 680 (2048MB)
Boinc 7.0.28
Driver 304.48
http://setiathome.berkeley.edu/show_host_detail.php?hostid=6614088

I updated Boinc to 7.0.28 and the driver to 301.42 and then to the Beta 304.48 and got the same results...

Any merciful cruncher might have any idea what may be going wrong?

---

We have already seen that the 680's do not really like the stock CUDA Fermi app and error out all work.
http://setiathome.berkeley.edu/forum_thread.php?id=68588

But we do know that the optimized apps do not have the same problem and will definitely run the work faster as well.
http://lunatics.kwsn.net/index.php?module=Downloads;sa=dlview;id=330

---

Thanks for your response arkayn, one more question though.

Does installing the lunatics app interfere with the hability to run other Boinc GPU projects? I would like to continue using this unit to run GPUGrid.

---

Thanks for your response arkayn, one more question though.

Does installing the lunatics app interfere with the hability to run other Boinc GPU projects? I would like to continue using this unit to run GPUGrid.

Nope, it only installs the optimized apps in the setiathome.berkeley.edu folder and does not touch any other project.

---

Thanks for the reply (and including the appropiate link), I installed the app, we'll see tomorrow how the next batch does.

Regards

---

Quick update.
Installing the Lunatics app seems to have solved the problem.
No more errors.

Thanks & Regards

---