ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-04 13:41:24 +00:00
Code Issues Packages Projects Releases Wiki Activity

[TileEngine] Support for sparsity in codegen (#2128)

Browse Source 
* Added sparsity flag in codegen

* remove comments

* clan formatted

* added sparsity as runtime argument

* updated README

* updated stream config variable

* fix typo for tail_num in hot loop
This commit is contained in:
Khushbu Agarwal
2025-04-28 18:19:23 -07:00
committed by GitHub
parent 4094ad158a
commit 768c99eca9
4 changed files with 56 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
tile_engine/ops/gemm/README.md
View File

@@ -20,24 +20,25 @@ make tile_engine_gemm -j
## tile_engine_gemm inputs
```
-m m dimension (default:3840)
-n n dimension (default:4096)
-k k dimension (default:2048)
-stride_a Tensor A stride (default:0)
-stride_b Tensor B stride (default:0)
-stride_c Tensor C stride (default:0)
-split_k SplitK value (default:1)
-v No validation: 0, Validation on CPU: 1, Validation on GPU: 2 (default:2)
-warmup Number of iterations before benchmark the kernel (default:50)
-repeat Number of iterations to benchmark the kernel (default:100)
-timer gpu:gpu timer, cpu:cpu timer (default:gpu)
-init Value for initializing tensor- random: 0, linear: 1, constant(1): 2 (default:0)
-pipeline possible values are: compv3, compv4, mem (default:compv3)
-scheduler possible values are: intrawave, interwave (default:intrawave)
-epilogue possible values are: cshuffle, default (default:cshuffle)
-pad_m Pad in m direction - true/false (default:false)
-pad_n Pad in n direction - true/false (default:false)
-pad_k Pad in k direction - true/false (default:false)
-m m dimension (default:3840)
-n n dimension (default:4096)
-k k dimension (default:2048)
-stride_a Tensor A stride (default:0)
-stride_b Tensor B stride (default:0)
-stride_c Tensor C stride (default:0)
-split_k SplitK value (default:1)
-v No validation: 0, Validation on CPU: 1, Validation on GPU: 2 (default:2)
-warmup Number of iterations before benchmark the kernel (default:50)
-repeat Number of iterations to benchmark the kernel (default:100)
-timer gpu:gpu timer, cpu:cpu timer (default:gpu)
-init Value for initializing tensor- random: 0, linear: 1, constant(1): 2 (default:0)
-structured_sparsity Sparsity for tensor - 0:false, 1:true (default: 0)
-pipeline possible values are: compv3, compv4, mem (default:compv3)
-scheduler possible values are: intrawave, interwave (default:intrawave)
-epilogue possible values are: cshuffle, default (default:cshuffle)
-pad_m Pad in m direction - true/false (default:false)
-pad_n Pad in n direction - true/false (default:false)
-pad_k Pad in k direction - true/false (default:false)
Note: pipeline, scheduler, epilogue, pad_m, pad_n, pad_k should be one of the options specified in instance_combination.json
```