ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-02 04:31:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CK TILE ENGINE] Adding GEMM Preshuffle to CK Tile Engine (#2712)

Browse Source 
* Partial Progress : Completed ListBlob

* Additional changes in Listbob

* Partial Progress : Generate Blobs Completed

* Partial Progress : Added Host side code for Preshuffle

* Working code for Preshuffle before Cleanup

* Partial Progress : Cleanup

* Partial Progress : Datatype Validation

* Partial Progress : Warptiles for preshuffle changed from hardcoding to take from config

* Partial Progress : Cleanup

* Partial Progress : Code Cleanup

* Partial Progress : Passing all valid tiles failing for unsupported tiles

* Partial Progress : Working code, testing pending for edge cases

* Partial Progress for testing

* Completed Code

* kBlockPerCu as tunable parameter from config

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Update tile_engine/ops/gemm_preshuffle/README.md

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Partial Progress : Working listkernels

* Partial Progress : Cleanup Working listkernels

* Partial Progress : Single instance

* Partial Progress : Working single instance code

* Partial Progress : Working generate individual instance code

* Partial Progress : Working rewamped code for given config file needed validation and edge case testing

* Partial Progress : Working Code, testing pending

* Removing LOGS file

* Working code

* Minor changes to GEMM Preshuffle : Restructured

* Minor Changes in Preshuffle

* Changes to Jenkins File

* Changes to Jenkins file to consider new architecture

* Changes to Jenkins file for fixing CI

---------

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
This commit is contained in:
Thrupti Raj Lakshmana Gowda
2025-09-12 13:50:19 -05:00
committed by GitHub
parent 1894a0dbc3
commit f6ba94fb5c
12 changed files with 3256 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

90
tile_engine/ops/gemm_preshuffle/configs/default_config.json Normal file
View File

@@ -0,0 +1,90 @@
{
"tile_config": {
"tile_m": {
"values": [
128
]
},
"tile_n": {
"values": [
128
]
},
"tile_k": {
"values": [
128
]
},
"warp_m": {
"values": [
1
]
},
"warp_n": {
"values": [
4
]
},
"warp_k": {
"values": [
1
]
},
"warp_tile_m": {
"values": [
16
]
},
"warp_tile_n": {
"values": [
16
]
},
"warp_tile_k": {
"values": [
16,32
]
}
},
"trait_config": {
"pipeline": {
"values": [
"preshufflev1",
"preshufflev2"
]
},
"scheduler": {
"values": [
"interwave",
"intrawave"
]
},
"epilogue": {
"values": [
"default",
"cshuffle"
]
},
"pad_m": {
"values": [
false
]
},
"pad_n": {
"values": [
false
]
},
"pad_k": {
"values": [
false
]
},
"persistent": {
"values": [
true,
false
]
}
}
}