[Feature] Support block scheduler v1 for FD (#2928)

* Support FD block scheduler v1

* Support FD block scheduler v1

* Support FD block scheduler v1

* Fix according to copilot review

* Fix according to review

* Remove is_dummy

* Fix bug when real_bsz=1

* Fix infer first token cost time

---------

Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>
This commit is contained in:
chenjian
2025-07-23 20:31:31 +08:00
committed by GitHub
parent ca0f71bd39
commit 85a78d695d
16 changed files with 898 additions and 40 deletions

View File

@@ -262,6 +262,8 @@ elif paddle.is_compiled_with_cuda():
"gpu_ops/stop_generation_multi_ends.cu",
"gpu_ops/stop_generation_multi_stop_seqs.cu",
"gpu_ops/set_flags.cu",
"gpu_ops/update_inputs_v1.cu",
"gpu_ops/recover_decode_task.cu",
"gpu_ops/step.cu",
"gpu_ops/step_reschedule.cu",
"gpu_ops/fused_get_rope.cu",