5 #pragma HLS interface port = a mode = m_axi offset = direct bundle = gmem0 6 #pragma HLS interface port = b mode = m_axi offset = direct bundle = gmem1 7 #pragma HLS interface port = output mode = m_axi offset = direct bundle = gmem2 10 #pragma HLS cache bundle = gmem0 line_count = 16 line_size = 16 bus_size = 32 ways = 1 num_write_outstanding = 2 rep_policy = \ 12 #pragma HLS cache bundle = gmem1 line_count = 32 line_size = 16 bus_size = 32 ways = 1 num_write_outstanding = 2 rep_policy = \ 13 tree write_policy = wt 14 #pragma HLS cache bundle = gmem2 line_count = 16 line_size = 16 bus_size = 32 ways = 1 num_write_outstanding = 4 rep_policy = \ 15 tree write_policy = wb 21 for(
unsigned c_tile = 0; c_tile <
tile_rank; c_tile++)
23 for(
unsigned r_tile = 0; r_tile <
tile_rank; r_tile++)
32 for(
unsigned i_tile = 0; i_tile <
tile_rank; i_tile++)
43 running += a[aIndex] * b[bIndex];
volatile int output[DIM_Y][DIM_X]
void mmult(int *a, int *b, int *output)