fix full graph

Signed-off-by: ganyi <ygan@amd.com>

Author: ganyi1996ppo

tests/v1/attention/test_batch_reordering.py

@@ -43,7 +43,7 @@ class ReorderTestCase:
         expected_modified=False,
     ),
     "all_prefills": ReorderTestCase(
-        requests=[(100, 100), (200, 200), (300, 300)],
+        requests=[(100, 0), (200, 0), (300, 0)],
         expected_order=[0, 1, 2],
         expected_modified=False,
     ),

vllm/config/scheduler.py

@@ -137,10 +137,6 @@ class SchedulerConfig:
     structured outputs, speculative decoding, and pipeline parallelism.
     """
 
-    split_prefill_from_chunk: bool = False
-    """Whether to split the prefill request into pure prefill and chunked
-    prefill in a single batch."""
-
     def compute_hash(self) -> str:
         """
         WARNING: Whenever a new field is added to this config,

vllm/platforms/rocm.py

@@ -391,11 +391,6 @@ def check_and_update_config(cls, vllm_config: "VllmConfig") -> None:
         ):
             compilation_config.custom_ops.append("+rms_norm")
 
-        if envs.VLLM_ROCM_USE_AITER and envs.VLLM_ROCM_USE_AITER_MHA:
-            # enable the request reorder if we are using AITER MHA
-            # for calculation
-            vllm_config.scheduler_config.split_prefill_from_chunk = True
-
     @classmethod
     def verify_model_arch(cls, model_arch: str) -> None:
         if model_arch in _ROCM_UNSUPPORTED_MODELS: