[Attention] Remove DCP indexer interleave guard and test TP1 output parity (#55802)

Signed-off-by: Stu Cao <stucao@Stus-MacBook-Pro.local>
Signed-off-by: <>
Signed-off-by: Summer Yang <girasoleyang@gmail.com>
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: Stu Cao <stucao@Stus-MacBook-Pro.local>
Co-authored-by: OpenAI Codex <codex@openai.com>
Co-authored-by: Summer Yang <girasoleyang@gmail.com>
Co-authored-by: Lucas Wilkinson <lwilkins@redhat.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
This commit is contained in:
stu-cao
2026-09-13 13:56:28 -07:00
committed by GitHub
co-authored by Stu Cao OpenAI Codex Summer Yang Lucas Wilkinson OpenAI Codex
parent 186a1e6222
commit b23433088b
@@ -771,15 +771,6 @@ class DeepseekV32IndexerMetadataBuilder(AttentionMetadataBuilder):
self.pcp_world_size = parallel_config.prefill_context_parallel_size
self.use_pcp = self.pcp_world_size > 1
self.cp_kv_cache_interleave_size = parallel_config.cp_kv_cache_interleave_size
# The DCP sparse-indexer code is parameterized by interleave size, but
# interleave > 1 is not yet validated end-to-end (gsm8k parity fails),
# so fail closed here rather than silently produce wrong output.
if self.dcp_world_size > 1 and self.cp_kv_cache_interleave_size > 1:
raise NotImplementedError(
"DCP sparse indexer currently supports only "
f"cp_kv_cache_interleave_size=1 (got "
f"{self.cp_kv_cache_interleave_size})."
)
# NOTE(Chen):an estimated max size of flattened_kv. Need to double check.
self.max_prefill_buffer_size = get_max_prefill_buffer_size(self.vllm_config)
self.num_speculative_tokens = (