Gregor Thalhammer <gregor.thalham...@gmail.com> writes: >> Am 05.04.2016 um 22:08 schrieb Martin Weigert <mweig...@mpi-cbg.de>: >> >> Hey Gregor, >> >> Thanks, upgrading to the latest github version (bbb2ca7dd) got rid of the >> INVALID_VALUE error, >> however the host-image copy still is off and empties the input data ndarray >> as a side effect >> >> #--------code >> from __future__ import print_function >> import numpy as np >> import pyopencl as cl >> >> #host -> image >> data = np.random.randn(100,100).astype(np.float32) >> fmt = cl.ImageFormat(cl.channel_order.R, cl.channel_type.FLOAT) >> im = cl.Image(ctx,cl.mem_flags.READ_WRITE, fmt, data.T.shape) >> >> cl.enqueue_copy(queue,im,data, origin = (0,0), region = data.T.shape) >> >> #image -> host >> out = np.empty_like(data) >> cl.enqueue_copy(queue,out, im, origin = (0,0), region = data.T.shape) >> >> >> print(np.allclose(out,data)) >> # prints False >> # data is zeroed out >> #-------- >> > > Hi Martin, > > I can confirm that your original snippet, essentially > > #OK > data = np.random.randn(10,10).astype(np.float32) > im = cl.image_from_array(ctx, data) > out = np.zeros_like(data) > cl.enqueue_copy(queue,out,im, origin = (0,0), region = (10,10)) > print np.allclose(out, data) > > is ok for me, but host to device image transfer via > > cl.enqueue_copy(queue, im, data, > origin = (0,0), > region = (10,10), > ) > > seems to have no effect at all. I did not look further into detail.
Possibly fixed by 587d8fd. Could you give that a shot? Thanks! (and thanks for reporting/looking into it, both of you!) Andreas _______________________________________________ PyOpenCL mailing list PyOpenCL@tiker.net https://lists.tiker.net/listinfo/pyopencl
