Jie Liu <jie.liu1(a)tu-dresden.de> writes:
I have a Pycuda code, which deals with two kernels. Both kernels run well
separately, but when I put them together, there is a memory problem
"LogicError: cuMemcpyDtoH failed: an illegal memory access was encountered".
In the second kernel "DotKernel", I can't change the values of any shared
array or global array. Could you please have a look at the code? Thank you
If you look at your kernel log (dmesg), do you see lines starting with
"NVRM: Xid" (or so)? If so, that indicates one of your kernels
encounters a segfault (that you're being told about at the next
dependent operation---in your case maybe the memory transfer---in the
CUDA-Memcheck might be of help in debugging this.