Sunny-bot1 930f7b781c [Optimization] Put get_block_shape_and_split_kv_block in cuda graph for append attention backend (#4443) ...

* get block in cuda graph

* fix sot

2025-10-17 10:59:56 +08:00

..

template_instantiation

…

append_attention_c4_impl.cuh

…

append_attention_c8_impl.cuh

…

append_attention_c16_impl.cuh

…

append_attention_func.cuh

…

append_attention_kernel.h

…

autogen_template_instantiation.py

…

decode_attention_func.cuh

…

decoder_write_cache_with_rope_impl.cuh

…

decoder_write_cache_with_rope_kernel.cu

…

decoder_write_cache_with_rope_kernel.h

…

encoder_write_cache_with_rope_impl.cuh

…

encoder_write_cache_with_rope_kernel.h

…

get_block_shape_and_split_kv_block.cu

[Optimization] Put get_block_shape_and_split_kv_block in cuda graph for append attention backend (#4443)

2025-10-17 10:59:56 +08:00

gqa_rope_write_cache.cu

…

mem_util.cuh

…

mla_cache_kernel.cu

…

mla_cache_kernel.cuh

…

mma_tensor_op.cuh

…

multi_head_latent_attention_kernel.h

…

multiquery_attention_c4_impl.cuh

…

multiquery_attention_c4_kernel.h

…

multiquery_attention_c8_impl.cuh

…

multiquery_attention_c8_kernel.h

…

multiquery_attention_c16_impl.cuh

…

multiquery_attention_c16_kernel.h

…

multiquery_decoder_attention_impl.cuh

…

multiquery_decoder_attention_kernel.h

…

pre_cache_len_concat.cu

…

speculate_write_cache_with_rope_impl.cuh

…

speculate_write_cache_with_rope_kernel.cu

…

speculate_write_cache_with_rope_kernel.h

…

template_config.json

…

utils.cuh

…