Hi all, The problem seems to happen only when N is 32. I tried a few values greater than 32, and did not have the error.
Sorry for the wrong information. Yifei On Wed, Apr 11, 2012 at 9:47 AM, Yifei Li <yifl...@gmail.com> wrote: > Hi, > > I got the above run-time error ONLY WHEN I set N (see below) greater than > or equal to 32. > > > The outline of my program is as follows: > > N = 31 > > # allocate memory on device > dA = cuda.to_device(A) > dR = cuda.to_device(R) > > for i in xrange(iter): > launch_kernel_1(dA, dR, block=(256,1,1), grid=(N,1), shared = 8*N + 4 > * n, texrefs=[mytex]) # n is a number smaller than N > launch_kernel_2(dA, dR, block=(256,1,1), grid=(N,1), shared = 8*N) > > > from the traceback, I can see the exception was thrown when > launch_kernel_2 was executed. > > Any help is highly appreciated. Thanks > > Yifei >
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda