o
    i?                     @   sJ   d dl mZmZ d dlmZ defddZdefddZdefdd	Zd
S )   )JitSpecgen_jit_spec)envreturnc                   C      t dtjd gS )zGenerate JIT spec for concat_mla kernel.

    This kernel efficiently concatenates CKV and KPE tensors for MLA prefill attention
    
concat_mlazconcat_mla.cur   jit_envFLASHINFER_CSRC_DIR r   r   W/home/ubuntu/vllm_env/lib/python3.10/site-packages/flashinfer/jit/dsv3_optimizations.pygen_concat_mla_module   s
   r   c                   C   r   )Ndsv3_router_gemmzdsv3_router_gemm.cur   r   r   r   r   gen_dsv3_router_gemm_module   s
   r   c                	   C   s   t dtjd tjd tjd tjd tjd tjd gtjd tjd d	 tjd d
 d d	 tjd d
 d d d	 tjd d
 d d gdS )Ndsv3_fused_routingzfused_moe/noAuxTcKernels.cuz#nv_internal/cpp/common/envUtils.cppz!nv_internal/cpp/common/logger.cppz&nv_internal/cpp/common/stringUtils.cppz(nv_internal/cpp/common/tllmException.cppz%nv_internal/cpp/common/memoryUtils.cunv_internalincludetensorrt_llmcutlass_extensionskernelscutlass_kernels)extra_include_pathsr   r   r   r   r   gen_dsv3_fused_routing_module   sT   	r   N)	corer   r    r   r	   r   r   r   r   r   r   r   <module>   s
    	