[Haskell-cafe] NVIDIA's CUDA and Haskell

[Johannes Waldmann]

Trevor L. McDonell <tmcdonell <at> cse.unsw.edu.au> writes:

> hmm... so libcuda and libcudart are in /usr/local/cuda/lib ...

actually libcuda is in /usr/lib/nvidia-current ,
unbeknownst to ./configure.
I think this comes from package "nvidia-current(-dev)" in ubuntu.

I could solve this with

LDFLAGS='-L/usr/lib/nvidia-current ' cabal install

note the space after dirname - otherwise ./configure
(in cuda-0.3.2.2) fails.

It still feels strange that I can build the examples from
NVIDIA_GPU_Computing_SDK/C/src/  without modifying LDFLAGS.  
When running, I need to set LD_LIBRARY_PATH=/usr/local/cuda/lib64
but that's expected.

Anyway, when running "accelerate-examples --cuda" I get mixed results,

...
fold-maximum: Ok
fold-minimum: Ok
fold-2d-sum: Failed: stack overflow
fold-2d-product: Failed: stack overflow
scanseg-sum: Failed: 
*** Internal error in package accelerate ***
*** Please submit a bug report 
at https://github.com/mchakravarty/accelerate/issues
./Data/Array/Accelerate/CUDA.hs:58 (unhandled): 
CUDA Exception: unspecified launch failure

I guess I should report them as advertised.


Is there a way to find out the expected results/failures 
for a recent accelerate package install?
(So I can avoid reporting them.)

Thanks - Johannes.