Clean up allocators (#9134)

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2025-08-13 13:56:04 -07:00
parent 2f20f43026
commit 9e426466af
16 changed files with 288 additions and 295 deletions
--- a/python/sglang/srt/layers/attention/flashinfer_backend.py
+++ b/python/sglang/srt/layers/attention/flashinfer_backend.py
@@ -870,6 +870,8 @@ class FlashInferIndicesUpdaterPrefill:
        spec_info: Optional[Union[EagleDraftInput, EagleVerifyInput]],
    ):
        if use_ragged:
+            # TODO: remove this device sync, we can use forward_batch.extend_prefix_lens_cpu
+            # and forward_batch.extend_seq_lens_cpu
            paged_kernel_lens = prefix_lens
            paged_kernel_lens_sum = paged_kernel_lens.sum().item()
        else: