invalid CUDA driver API error No.: 4294967295

[thibaultlanglois]

Hello, I just installed cl-cuda and was trying to run the vector-add example and got this error:
invalid CUDA driver API error No.: 4294967295
[Condition of type SIMPLE-ERROR]

Restarts:
0: [RETRY] Retry SLIME REPL evaluation request.
1: [*ABORT] Return to SLIME's top level.
2: [ABORT] abort thread (#<THREAD "repl-thread" RUNNING {100ABF0033}>)

Backtrace:
0: (CL-CUDA.DRIVER-API::GET-ERROR-STRING 4294967295)
1: (CL-CUDA.DRIVER-API::CHECK-CUDA-ERROR CL-CUDA.DRIVER-API:CU-INIT 4294967295)
2: (MAIN)
3: (SB-INT:SIMPLE-EVAL-IN-LEXENV (MAIN) #)
4: (EVAL (MAIN))

I run linux 16.04, the last version of cuda (8.0) and SBCL 1.2.12.
I could compile the cuda tests (NVIDIA_CUDA-8.0_Samples) so I think the libs were installed correctly.

Any suggestion ?

[gos-k]

Have deviceQuery of NVIDIA CUDA Samples worked?

[thibaultlanglois]

Well, I haven't checked before and it is not OK.
I think the error message means that the version of my driver is not compatible with the version of the cuda lib I installed (8.0).
The nvidia driver I use (340.101) is according to nvidia web site, the last version available for my board (Quadro FX 580).
Should I try to install an older version of the cuda libs ?

./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
cudaGetDeviceCount returned 35
-> CUDA driver version is insufficient for CUDA runtime version
Result = FAIL

[gos-k]

Yes