[MLA] Simplification to batch P/D reordering (#16673)

Signed-off-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
Nick Hill
2025-04-17 16:12:09 -04:00
committed by GitHub
parent e4755f7fac
commit 0377b8310b
2 changed files with 12 additions and 16 deletions
+7 -9
View File
@@ -458,7 +458,13 @@ class GPUModelRunner(LoRAModelRunnerMixin):
if removed_req_indices:
self.input_batch.condense(removed_req_indices)
if batch_changed:
# Some attention backends (namely MLA) may want to separate requests
# based on if the attention computation will be compute-bound or
# memory-bound. This gives them a hook to do that.
batch_reordered = self.attn_metadata_builder.reorder_batch(
self.input_batch, scheduler_output)
if batch_changed or batch_reordered:
self.input_batch.refresh_sampling_metadata()
def _prepare_inputs(
@@ -471,14 +477,6 @@ class GPUModelRunner(LoRAModelRunnerMixin):
num_reqs = self.input_batch.num_reqs
assert num_reqs > 0
# Some attention backends (namely MLA) may want to separate requests
# based on if the attention computation will be compute-bound or
# memory-bound. This gives them a hook to do that.
modified_batch = self.attn_metadata_builder.reorder_batch(
self.input_batch, scheduler_output)
if modified_batch:
self.input_batch.refresh_sampling_metadata()
# OPTIMIZATION: Start copying the block table first.
# This way, we can overlap the copy with the following CPU operations.
self.input_batch.block_table.commit(num_reqs)