use MLA on rocm

Signed-off-by: Hongxia Yang <hongxia.yang@amd.com>
vllm-project · Feb 2, 2025 · 3b84895 · 3b84895
1 parent 0c1cc40
commit 3b84895
Showing 1 changed file with 0 additions and 3 deletions.
diff --git a/vllm/platforms/rocm.py b/vllm/platforms/rocm.py
@@ -78,9 +78,6 @@ def get_attn_backend_cls(cls, selected_backend, head_size, dtype,
                              kv_cache_dtype, block_size, use_v1,
                              use_mla) -> str:
         if use_mla:
-            if selected_backend and selected_backend != _Backend.TRITON_MLA:
-                logger.warning(f"Cannot use {selected_backend.name} "
-                               "backend for MLA.")
             logger.info("Using Triton MLA backend.")
             return "vllm.attention.backends.triton_mla.TritonMLABackend"
         selected_backend = (_Backend.ROCM_FLASH if selected_backend