realliujiaxu 5d12446573 [Feat][SP] Suport SP for VL MoE models (#7044) ...

### What this PR does / why we need it?

2nd PR for https://github.com/vllm-project/vllm-ascend/issues/5712,
extend SP to VL MoE models.


### Does this PR introduce _any_ user-facing change?
remove `sp_threshold` in additional config and reuse `sp_min_token_num`
from vLLM.


### How was this patch tested?
- Model: Qwen3-VL-30B-A3B, 
- TP4 DP2
- 100 reqs
- max concurrency 1

| Seq length | Mean TTFT (ms) main | Mean TTFT (ms) this PR |
|------------|---------------------|------------------------|
| 4k         | 429.40               | 323.3                  |
| 16k        | 1297.01              | 911.74                |

- vLLM version: v0.16.0
- vLLM main:
4034c3d32e

---------

Signed-off-by: realliujiaxu <realliujiaxu@163.com>

2026-03-24 17:16:00 +08:00

..

spec_decode

[Test][Feature] Add e2e test for QuaRot model with eagle3 (#7128)

2026-03-16 15:35:55 +08:00

test_aclgraph_capture_replay.py

Revert "[CI] fix skiped e2e test when upgrade vllm version (#6654)" (#7166)

2026-03-11 23:03:15 +08:00

test_data_parallel.py

Main2main upgrade vllm to 0318 commit (#7412)

2026-03-19 17:17:36 +08:00

test_disaggregated_encoder.py

[CI] Recover pd disaggregated encoder test case that been incorrectly skipped (#7505)

2026-03-23 21:41:28 +08:00

test_expert_parallel.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_external_launcher.py

[CI][Misc] Use offline mode for model downloads (#7179)

2026-03-13 08:52:24 +08:00

test_full_graph_mode.py

fixed fia pad logic in graph mode. (#7144)

2026-03-12 14:50:54 +08:00

test_ilama_lora_tp2.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_llama32_lora_tp2.py

[Bugfix][LoRA] Fix the issue when enable LoRA + tp + fully_sharded_loras (#6650)

2026-03-11 15:43:15 +08:00

test_offline_inference_distributed.py

[Bugfix] Fix multi-instance serving OOM on single card (#7427)

2026-03-23 14:22:59 +08:00

test_offline_weight_load.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_prefix_caching.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_quantization.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_qwen3_moe_routing_replay.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_qwen3_moe.py

[EPLB][Bugfix] Set parallel_config.enable_eplb to true to load redundant experts (#7470)

2026-03-20 15:22:55 +08:00

test_qwen3_performance.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_sequence_parallelism_moe.py

[Feat][SP] Suport SP for VL MoE models (#7044)

2026-03-24 17:16:00 +08:00

test_shared_expert_dp.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_single_request_aclgraph.py

[Lint]Style: Convert test/ to ruff format(Batch #1) (#6738)

2026-03-10 09:52:50 +08:00

test_sp_pass.py

[Feat][SP] Suport SP for VL MoE models (#7044)

2026-03-24 17:16:00 +08:00