Chen Chen a2daacbd71 [perf] Fix MLAPO weight disposal for KV-consumer MLA in PD-mix deploy... (#5192) ...

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

- Problem: In MLA+MLAPO, KV-consumer deployments keep
fused_qkv_a_proj/q_proj weights and quant params even though MLAPO uses
the prepacked buffers, increasing memory footprint on decode nodes.
- Fix: Conditionally drop those tensors only when
`kv_transfer_config.is_kv_consumer` to reclaim memory (consistent with
the SFA behavior #4774 ).

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: Chen Chen <0109chenchen@gmail.com>

2026-01-05 21:29:45 +08:00

..

_cann_ops_custom

[Kernel] add custom op GmmSwigluQuantWeightNzTensorList (#3804)

2025-11-28 18:06:39 +08:00

attention

[perf] Fix MLAPO weight disposal for KV-consumer MLA in PD-mix deploy... (#5192)

2026-01-05 21:29:45 +08:00

compilation

[BugFix][Fusion] Fix graph fusion failure problem (#5253)

2026-01-05 17:49:09 +08:00

core

[CI] fix lint (#5216)

2025-12-20 17:03:25 +08:00

device_allocator

[Misc]Clean up useless import from vllm (#2049)

2025-07-28 16:01:59 +08:00

distributed

[P/D]Remove mooncake kvpool unused parameter local_hostname (#5574)

2026-01-05 20:18:59 +08:00

eplb

[smoke][bugfix] moe_init_routing_v2 active_expert_range use int type (#5521)

2025-12-31 09:19:04 +08:00

kv_offload

[BugFix] Fix npu-cpu offloading interface change bug. (#5290)

2025-12-27 10:21:20 +08:00

lora

[BugFix]Fix precision issue for LoRA feature (#4141)

2025-12-19 14:22:06 +08:00

model_loader

[CI] speed up ut (#4901)

2025-12-11 18:45:43 +08:00

ops

[Bugfix] record cos and sin cache in AscendRotaryEmbedding (#5516)

2026-01-05 20:12:41 +08:00

patch

[BugFix][Fusion] Fix graph fusion failure problem (#5253)

2026-01-05 17:49:09 +08:00

quantization

[Model] Add LongCat-Flash (#3833)

2025-12-31 17:06:55 +08:00

sample

feat: implement high-performance Triton kernels for rejection sampling: optimization for rejection_random_sample_kernel (#5259)

2026-01-05 16:03:02 +08:00

spec_decode

[Feat][Spec] Optimize token index calculation in spec decode with Triton kernel (#5356)

2026-01-05 16:51:29 +08:00

worker

[bugfix] fix test_camem failed with triton-ascend (#5492)

2026-01-05 20:10:54 +08:00

xlite

[CI] add xlite e2e test (#5305)

2025-12-25 09:17:06 +08:00

__init__.py

clean up model module (#4611)

2025-12-02 17:35:47 +08:00

ascend_config.py

[Feature] Support kv nz feature for DeepSeek decode node in disagg-prefill scenario (#3072)

2025-12-31 14:24:04 +08:00

ascend_forward_context.py

[BugFix]Disable dispatch_gmm_combine_decode operator when mtp drafter model uses non-w8a8 while main model uses w8a8, or drafter model is eagle series (#5293)

2026-01-04 17:51:28 +08:00

cpu_binding.py

[main] support cpu binding (#3546)

2025-10-21 09:17:03 +08:00

envs.py

Remove VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE (#5272)

2025-12-25 11:09:56 +08:00

flash_common3_context.py

[Perf]enable prefill flashcommon3 (#4065)

2025-12-14 09:34:13 +08:00

meta_registration.py

Fix the bugs about operator registration by PyTorch Dispatcher (#2786)

2025-09-13 11:58:52 +08:00

platform.py

[Bugfix] fix pcp + eplb error (#5561)

2026-01-05 14:08:11 +08:00

profiling_config.py

Drop ascend scheduler (#4623)

2025-12-05 09:03:45 +08:00

utils.py

[BugFix]Disable dispatch_gmm_combine_decode operator when mtp drafter model uses non-w8a8 while main model uses w8a8, or drafter model is eagle series (#5293)

2026-01-04 17:51:28 +08:00