@@ -2661,7 +2661,7 @@ template <int THREADS, int ITEMS_PER_THREAD, int TILE_ROWS, int TILE_COLS, int T
26612661 // const int global_col = base_row; // block offset for col
26622662 if ((base_col + subrow_loop_row + jrow + warp_id < outRows) && (base_row+warp_lane < rows))
26632663 {
2664- // each row hae 32 columns and is offset by 1 to prevent bank conflict during storage into smem
2664+ // each row has 32 columns and is offset by 1 to prevent bank conflict during storage into smem
26652665 char data = smem_data[(subrow_loop_row + jrow + warp_id)*33 + warp_lane];
26662666
26672667 // each 32 columns we have new tile
@@ -2700,7 +2700,7 @@ template <int THREADS, int ITEMS_PER_THREAD, int TILE_ROWS, int TILE_COLS, int T
27002700 // const int global_col = base_row; // block offset for col
27012701 if ((base_col + subrow_loop_row + jrow + warp_id < outRows) && (base_row+warp_lane < rows))
27022702 {
2703- // each row hae 32 columns and is offset by 1 to prevent bank conflict during storage into smem
2703+ // each row has 32 columns and is offset by 1 to prevent bank conflict during storage into smem
27042704 char data = smem_data[(subrow_loop_row + jrow + warp_id)*33 + warp_lane];
27052705
27062706 // each 32 columns we have new tile
@@ -2777,7 +2777,7 @@ template <int THREADS, int ITEMS_PER_THREAD, int TILE_ROWS, int TILE_COLS, int T
27772777 // const int global_col = base_row; // block offset for col
27782778 if ((base_col + subrow_loop_row + jrow + warp_id < outRows) && (base_row+warp_lane < rows))
27792779 {
2780- // each row hae 32 columns and is offset by 1 to prevent bank conflict during storage into smem
2780+ // each row has 32 columns and is offset by 1 to prevent bank conflict during storage into smem
27812781 char data = smem_data[(subrow_loop_row + jrow + warp_id)*33 + warp_lane];
27822782
27832783 // each 32 columns we have new tile
0 commit comments