GitLab

新增 VLLM_USE_FUSED_RMS_ROPE 分支，走 fused 路径
注册 torch.ops.vllm.rms_rotary_embedding_fuse（direct_register_custom_op）
cos_sin_cache 自动转 device/dtype 并缓存，避免每次重复拷贝

for qwen3, VLLM_USE_FUSED_RMS_ROPE=1 (default)

feat: Supprot fp8 channle-wise matmul.

See merge request dcutoolkit/deeplearing/vllm!380

V0.11.0 dev kvscale

See merge request dcutoolkit/deeplearing/vllm!378

feat: support fp8-blockwise matmul impl.

See merge request dcutoolkit/deeplearing/vllm!377

feat(moe/marlin): 移除 VLLM_USE_MARLIN_W16A16_MOE，改为基于 lightop 探测自动启用并一次性缓存决策

See merge request dcutoolkit/deeplearing/vllm!376

  - 使用 get_moe_cuda_marlin_config_w16a16(status) 判断 W16A16 Marlin MoE 是否可用
  - 在 FusedMoE 初始化阶段计算并缓存 _marlin_w16a16_moe_enabled，满足条件时强制 use_nn_moe=False
  - 权重加载后按缓存结果进行一次性 Marlin pack；运行时按 packed 标记走 Marlin fast path
  - 删除 envs.py 中 VLLM_USE_MARLIN_W16A16_MOE 环境变量定义与解析逻辑

[fix]解决glm4 moe + mtp精度异常

See merge request dcutoolkit/deeplearing/vllm!374

[fix]解决gpt oss nn moe权重加载出错

See merge request dcutoolkit/deeplearing/vllm!372

[qwen3-480b] MoE(TN) configs for nmz TP=8

1、pcie 解决custom cudagraph模式需要拷贝的问题，这个需要配合dtk进行使用。

See merge request dcutoolkit/deeplearing/vllm!369

   vllm：export VLLM_CUSTOM_CACHE=1
   dtk：export HIP_KERNEL_EVENT_SYSTENFENCE=1
2、kvcache支持fp8

todo: add VLLM_USE_QUERY_QUANT to not use q quant

适配block-wise fp8接口

See merge request dcutoolkit/deeplearing/vllm!366

set VLLM_USE_PD_SPLIT=1
update moe_align_block_size

Switch default w8a8 gemm impl to blaslt.

See merge request dcutoolkit/deeplearing/vllm!365