GitLab

  - 在 C++ 算子中移除 query/key/residual 的 contiguous 强校验
  - 对非连续输入使用工作张量执行计算（优化路径与 fallback 路径统一）
  - 计算完成后将结果 copy_ 回原张量，保持 in-place 语义
  - 移除 qwen3 / qwen3_moe Python 前向中的 q、k.contiguous() 预处理

  launch_opt_rms_rope 增加 rot_dim=64/128 分发路径
  放宽 supports_qwen3_opt 条件，允许 rot_dim=64 进入优化分支
  qwen3 / qwen3_moe 将 q_bias、k_bias 参数统一为 q_residual、k_residual
  qwen3_moe 的自定义 op 注册增加重复注册保护"`

  - 在 vLLM _C 扩展中新增 rms_rotary_embedding_fuse（注册 op + CUDA kernel），减少对 LightOp 的硬依赖
  - 新增环境变量 VLLM_FUSED_RMS_ROPE_BACKEND=auto|vllm|lightop，auto 优先走 vLLM，缺失时回退 LightOp
  - 更新 Qwen3 / Qwen3-MoE 的 fused 路径按后端选择执行
  - 补充 tc_opt benchmark 结果解析脚本 benchmarks/tc_opt/test/parse_bench_results.py

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

add VLLM_USE_PP_SYNC to use pp sync
update qwen3 of rmsnorm

Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

Signed-off-by: Rahul Tuli <rtuli@redhat.com>

Signed-off-by: Tao He <linzhu.ht@alibaba-inc.com>

Signed-off-by: Dipika Sikka <dipikasikka1@gmail.com>

Signed-off-by: wang.yuqi <noooop@126.com>

Signed-off-by: wang.yuqi <noooop@126.com>

Signed-off-by: wang.yuqi <noooop@126.com>

Signed-off-by: Max de Bayser <mbayser@br.ibm.com>
Signed-off-by: Max de Bayser <maxdebayser@gmail.com>
Signed-off-by: 22quinn <33176974+22quinn@users.noreply.github.com>
Co-authored-by: 22quinn <33176974+22quinn@users.noreply.github.com>

Signed-off-by: simon-mo <simon.mo@hey.com>

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

Signed-off-by: evian <eviantai@u.nus.edu>
Co-authored-by: evian <eviantai@u.nus.edu>

Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>

Signed-off-by: YamPengLi <yampayne.lyp@alibaba-inc.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>