sgemm_wg: Run multiple threadblock per core

This commit is contained in:
Hansung Kim
2024-02-27 15:44:04 -08:00
parent 5b1c527186
commit f1e7407d3a
2 changed files with 28 additions and 10 deletions

View File

@@ -147,8 +147,8 @@ int main(int argc, char *argv[]) {
RT_CHECK(vx_dev_open(&device));
// FIXME: hardcoded
uint32_t dim_m = 16;
uint32_t dim_n = 16;
uint32_t dim_m = 32;
uint32_t dim_n = 32;
uint32_t dim_k = 32;
generate_source_matrix(dim_m, dim_n, dim_k);