On Freitag 09 April 2010, Dan Goodman wrote:
> Hi all,
>
> I'm having a problem using multiprocessing and pycuda on linux. I guess
> I'm getting something fairly basic wrong, but couldn't find any
> documentation. I expected the example below to work (and it does work on
> Windows), but on Linux nothing happens, and nvcc gets stuck using 100%
> processor usage and never returning:
>
> from numpy import ones, int32
> import multiprocessing
> import pycuda
> import pycuda.autoinit as autoinit
> import pycuda.driver as drv
> import pycuda.gpuarray as gpuarray
> from pycuda.compiler import SourceModule
>
> def doit(x):
>
> code = '''
> __global__ void test(double *x, int n)
> {
> int i = blockIdx.x * blockDim.x + threadIdx.x;
> if(i>=n) return;
> x[i] *= 2.0;
> }
> '''
>
> mod = SourceModule(code)
> f = mod.get_function('test')
> x = gpuarray.to_gpu(ones(100))
> f(x, int32(100), block=(100,1,1))
> return x.get()
>
> if __name__=='__main__':
>
> # doit(1)
>
> pool = multiprocessing.Pool(1)
> result = pool.map(doit, [0])
> print result
>
> I'm wondering if it has something to do with the way that
> multiprocessing uses forking on Linux but not on Windows?
>
> Thanks for any suggestions!
>
> Btw, I'm on 64 bit Linux, GTX 280, and get the same problem with
> OpenSUSE and Ubuntu.Works for me on 2.5 with mp 2.6.2.1. What nvcc? Mine: nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2009 NVIDIA Corporation Built on Fri_Feb_19_19:12:59_PST_2010 Cuda compilation tools, release 3.0, V0.2.1221 Andreas
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ PyCUDA mailing list pyc...@host304.hostmonster.com http://host304.hostmonster.com/mailman/listinfo/pycuda_tiker.net
