Cao Yi cb4c7de856 [Perf] Optimize MTP execution by reordering state update operation (#6844) ...

## Summary
- Move `_update_states_after_model_execute` call from after main model
sampling to after draft model execution
- This reordering reduces pipeline bubbles between main model and draft
model execution
- No accuracy impact - the state update operation is independent of
draft token proposal

## Performance Impact
Reduces idle time between main model and draft model execution stages,
improving overall MTP (Multi-Token Prediction) performance.
- vLLM version: v0.15.0
- vLLM main:
83b47f67b1

---------

Signed-off-by: SlightwindSec <slightwindsec@gmail.com>
Co-authored-by: wanghuanjun2113 <wanghuanjun2113@gmail.com>

2026-03-09 15:55:27 +08:00

..

v2

[Feature] adapt to uva buffer and main2main (#6657)

2026-02-12 10:36:31 +08:00

__init__.py

[Misc][V0 Deprecation] Remove Cache Engine Used for V0 Worker (#1878)

2025-07-19 09:42:32 +08:00

block_table.py

[Lint]Style: Convert vllm-ascend/ to ruff format(Batch #7) (#6023)

2026-02-06 14:56:53 +08:00

model_runner_v1.py

[Perf] Optimize MTP execution by reordering state update operation (#6844)

2026-03-09 15:55:27 +08:00

npu_input_batch.py

[Lint]Style: Convert vllm-ascend/ to ruff format(Batch #10) (#6173)

2026-02-06 15:35:06 +08:00

pcp_utils.py

[feat]ds3.2 pcp support mtp and chunkprefill (#6917)

2026-03-03 19:03:50 +08:00

worker.py

[P/D][KVPool]Mooncake Layerwise Connector supports kv_pool (#7032)

2026-03-09 10:49:04 +08:00