iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[ROCm] Add tuned selective_state_update float32 config for AMD Instinct MI325X (#50007)

Browse Source 
Signed-off-by: vanshbhatia-amd <vansh.bhatia@amd.com>
This commit is contained in:
vanshbhatia-amd
2026-08-07 12:58:36 -04:00
committed by GitHub
parent 34c1cd20a5
commit e229fdbc87
1 changed files with 51 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/model_executor/layers/mamba/ops/configs/selective_state_update/headdim=64,dstate=128,device_name=AMD_Instinct_MI325X,cache_dtype=float32.json
+51
View File
@@ -0,0 +1,51 @@
{
"triton_version": "3.6.0",
"128": {
"BLOCK_SIZE_M": 8,
"num_warps": 4
},
"256": {
"BLOCK_SIZE_M": 8,
"num_warps": 4
},
"1024": {
"BLOCK_SIZE_M": 64,
"num_warps": 1
},
"2048": {
"BLOCK_SIZE_M": 16,
"num_warps": 1
},
"4096": {
"BLOCK_SIZE_M": 16,
"num_warps": 1
},
"8192": {
"BLOCK_SIZE_M": 16,
"num_warps": 1
},
"16384": {
"BLOCK_SIZE_M": 32,
"num_warps": 8
},
"32768": {
"BLOCK_SIZE_M": 64,
"num_warps": 2
},
"65536": {
"BLOCK_SIZE_M": 64,
"num_warps": 4
},
"131072": {
"BLOCK_SIZE_M": 64,
"num_warps": 1
},
"196608": {
"BLOCK_SIZE_M": 64,
"num_warps": 2
},
"262144": {
"BLOCK_SIZE_M": 64,
"num_warps": 1
}
}