[Bugfix] Shard UniformTypeKVCacheSpecs block table width under DCP (#50823)

Signed-off-by: Mikhail Kostryukov <mike@triptrack.net>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
drakosha
2026-08-03 17:34:29 +00:00
committed by GitHub
co-authored by Claude Matthew Bonanni
parent 5df9999fcf
commit f0de1a604c
2 changed files with 42 additions and 0 deletions
+14
View File
@@ -855,6 +855,20 @@ class UniformTypeKVCacheSpecs(KVCacheSpec):
)
return max_num_pages * self.page_size_bytes
def max_num_blocks_per_req(self, vllm_config: VllmConfig, max_len: int) -> int:
# Metadata builders are constructed from the per-layer spec, so the base
# cdiv(max_len, block_size) would drop its DCP sharding and size the
# block table wider than those builders expect.
widths = {
spec.max_num_blocks_per_req(vllm_config, max_len)
for spec in self.kv_cache_specs.values()
}
assert len(widths) == 1, (
"All layers in the same KV cache group must need the same number "
f"of block table entries, got {sorted(widths)}."
)
return next(iter(widths))
@classmethod
def is_uniform_type(cls, kv_cache_specs: dict[str, KVCacheSpec]) -> bool:
"""