(1)As shown in the arch.h :#include <thrust/system/cuda/detail/arch.h>", but
the header file does not exist in this route, and therefore I copy the arch.h
from "C:\Program Files\NVIDIA GPU Computing
Toolkit\CUDA\v5.5\include\cusp\detail\device" to the first route.
(2)Disable the duplicate definition of max_active_blocks.
(3)Syntax error was encountered when calling the max_active_blocks, and all due
to the misuse of data type. According to the context, the max_active_blocks has
a data type of size_t that belong to the class of arch, however,
arch::max_active_blocks appears then. As a result, it should be just revised as
return max_active_blocks().
Original issue reported on code.google.com by jiasen.h...@gmail.com on 13 May 2014 at 11:01
Original issue reported on code.google.com by
jiasen.h...@gmail.com
on 13 May 2014 at 11:01