41
edits
Changes
run of kernel with blocks of 2D threads, run 4 times. the runtime, while slightly longer than the run with global memory where shared memory is not initialized for ghost cells, it still takes less time to run than the version with Global memory.
run of kernel with blocks of 2D threads, run 4 times. the runtime, while slightly longer than the run with global memory where shared memory is not initialized for ghost cells, it still takes less time to run than the version with Global memory.
