[Performance][Fix] update nvfp4 code to support renorm routing (#28569)

Signed-off-by: jiahanc <173873397+jiahanc@users.noreply.github.com> Co-authored-by: Michael Goin <mgoin64@gmail.com>
2025-11-16 18:02:42 -08:00
parent 80b6080ddc
commit 561253b37f
2 changed files with 15 additions and 8 deletions
--- a/vllm/model_executor/layers/quantization/utils/flashinfer_utils.py
+++ b/vllm/model_executor/layers/quantization/utils/flashinfer_utils.py
@@ -291,5 +291,8 @@ def get_flashinfer_moe_backend() -> FlashinferMoeBackend:

 def is_flashinfer_supporting_global_sf(backend: FlashinferMoeBackend | None) -> bool:
    # TODO(shuw@nvidia): Update when new backends are added.
-    backends_supporting_global_sf = (FlashinferMoeBackend.CUTLASS,)
+    backends_supporting_global_sf = (
+        FlashinferMoeBackend.CUTLASS,
+        FlashinferMoeBackend.TENSORRT_LLM,
+    )
    return backend in backends_supporting_global_sf