On Fri, 7 Oct 2011 17:56:40 +0200, Nick Rayrider <nick.rayri...@googlemail.com> wrote: > This is what the profiler outputs. Can anyone confirm it? > > # CUDA_PROFILE_LOG_VERSION 2.0 > # CUDA_DEVICE 0 Quadro FX 580 > # CUDA_PROFILE_CSV 1 > # TIMESTAMPFACTOR fffff6d463c55b20 > method gputime cputime occupancy memtransferhostmemtype > fill 3.072 17 0.042 > memcpyHtoD 5.024 14 0 > kernel 2.208 12 0.333 > memcpyDtoH 5.024 26 0
Your mistake is here: empty_pinned_array = gpuarray.to_gpu(data) You're misunderstanding the (Python) semantics of this. This deletes the pinned array you allocated just above and then transfers data into a newly-allocated non-pinned array. To use existing storage, you need to use a memcpy of some sort. Andreas
pgpgGnVzgX8BF.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda