leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 25 Packages Projects Releases Wiki Activity

fix split_k_mode and add reduction kernel for f16 input/accum/output (#896)

Browse Source
This commit is contained in:
Manish Gupta
2023-03-30 15:31:08 -04:00
committed by GitHub
parent bc36122c3f
commit 660a05f581
4 changed files with 41 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/library/src/reduction/init_reduction_operations.cu
+2
View File
@@ -42,6 +42,7 @@ namespace library {
///////////////////////////////////////////////////////////////////////////////////////////////
// CUTLASS Reduction Instances //
///////////////////////////////////////////////////////////////////////////////////////////////
void initialize_reduce_add_linear_combination_f16_f16_f16(Manifest &manifest);
void initialize_reduce_add_linear_combination_f32_f32_f16(Manifest &manifest);
void initialize_reduce_add_linear_combination_f32_f32_f32(Manifest &manifest);
void initialize_reduce_add_linear_combination_f64_f64_f64(Manifest &manifest);
@@ -52,6 +53,7 @@ void initialize_reduce_add_linear_combination_cf32_cf32_cf32(Manifest &manifest)
//
void initialize_all_reduction_op(Manifest &manifest) {
initialize_reduce_add_linear_combination_f16_f16_f16(manifest);
initialize_reduce_add_linear_combination_f32_f32_f16(manifest);
initialize_reduce_add_linear_combination_f32_f32_f32(manifest);
initialize_reduce_add_linear_combination_f64_f64_f64(manifest);