apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-09-27 12:52:29 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

[Feature] Support pd ep deployment with yiyan adapter (#4029)

Browse Source 
* [Feature] Support mixed deployment with yiyan adapter in release2.2

* fix metrics

* add unit test

* add unit test

* add unit test

* Support pd ep deployment with yiyan adapter

* Support pd ep deployment with yiyan adapter

* refactor cache messager

* support scheduler v1 in PD

* suppport pd v1 + chunk prefill

* suppport pd v1 + chunk prefill

* add eplb

* support eplb

* support eplb

* support eplb

* support v1

* fix

* fix

* fix bug

* remove eplb support

* support prefix cache in P

* fix bug

* fix bug

* support one stop in V1

* fix bug

* fix ci

* fix ci

* fix

* fix

* fix

* fix

* fix

---------

Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
This commit is contained in:
chenjian
2025-09-22 16:41:38 +08:00
committed by GitHub
parent 9845f0d010
commit 918ccdb123
22 changed files with 1838 additions and 343 deletions
Download Patch File Download Diff File Expand all files Collapse all files

204
fastdeploy/model_executor/pre_and_post_process.py
View File

@@ -76,6 +76,7 @@ else:
update_inputs,
step_reschedule,
update_inputs_v1,
speculate_step_reschedule,
)
@@ -413,12 +414,11 @@ def step_cuda(
"""
if speculative_config.method is not None:
if enable_prefix_caching:
speculate_step_system_cache(
if DISABLE_RECOVER:
speculate_step_reschedule(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["step_seq_lens_decoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
@@ -444,64 +444,67 @@ def step_cuda(
speculative_config.num_speculative_tokens,
)
else:
speculate_step_paddle(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
share_inputs["accept_num"],
block_size,
enc_dec_block_num,
speculative_config.num_speculative_tokens,
)
if enable_prefix_caching:
speculate_step_system_cache(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["step_seq_lens_decoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
share_inputs["accept_num"],
block_size,
enc_dec_block_num,
speculative_config.num_speculative_tokens,
)
else:
speculate_step_paddle(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
share_inputs["accept_num"],
block_size,
enc_dec_block_num,
speculative_config.num_speculative_tokens,
)
else:
if enable_prefix_caching:
step_system_cache(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["step_seq_lens_decoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
block_size,
enc_dec_block_num,
)
elif DISABLE_RECOVER:
if DISABLE_RECOVER:
step_reschedule(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
@@ -529,32 +532,61 @@ def step_cuda(
enc_dec_block_num,
)
else:
step_paddle(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
block_size,
enc_dec_block_num,
)
if enable_prefix_caching:
step_system_cache(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["step_seq_lens_decoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
block_size,
enc_dec_block_num,
)
else:
step_paddle(
share_inputs["stop_flags"],
share_inputs["seq_lens_this_time"],
share_inputs["step_seq_lens_encoder"],
share_inputs["seq_lens_encoder"],
share_inputs["seq_lens_decoder"],
share_inputs["block_tables"],
share_inputs["encoder_block_lens"],
share_inputs["is_block_step"],
share_inputs["step_block_list"],
share_inputs["step_lens"],
share_inputs["recover_block_list"],
share_inputs["recover_lens"],
share_inputs["need_block_list"],
share_inputs["need_block_len"],
share_inputs["used_list_len"],
share_inputs["free_list"],
share_inputs["free_list_len"],
share_inputs["input_ids"],
share_inputs["pre_ids"],
share_inputs["step_idx"],
share_inputs["next_tokens"],
share_inputs["first_token_ids"],
block_size,
enc_dec_block_num,
)
def rebuild_padding(