biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Use aiter triton fused_add_rmsnorm_pad for gpt-oss (#30976)

Browse Source 
Signed-off-by: Rohan138 <rohanpotdar138@gmail.com>
This commit is contained in:
Rohan Potdar
2026-01-28 14:47:47 -06:00
committed by GitHub
parent 3e440786af
commit 59bcc5b6f2
9 changed files with 327 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/compile/test_fusion.py
View File

@@ -410,7 +410,7 @@ def test_aiter_fusion_rmsnorm_quant(
)
with vllm.config.set_current_vllm_config(vllm_config), monkeypatch.context() as m:
from vllm.compilation.rocm_aiter_fusion import RocmAiterRMSNormFusionPass
from vllm.compilation.rocm_aiter_fusion import RocmAiterRMSNormQuantFusionPass
m.setenv("VLLM_ROCM_USE_AITER", "1")
@@ -420,7 +420,7 @@ def test_aiter_fusion_rmsnorm_quant(
torch.set_default_dtype(dtype)
torch.manual_seed(1)
fusion_pass = RocmAiterRMSNormFusionPass(vllm_config)
fusion_pass = RocmAiterRMSNormQuantFusionPass(vllm_config)
model = TestModel(
hidden_size=hidden_size,