mirror of
https://github.com/vllm-project/vllm.git
synced 2026-08-10 07:48:09 +00:00
[Test] Fix one-sided MNNVL alltoall test workspace under-reservation (#44946)
Signed-off-by: Yongye Zhu <[email protected]>
This commit is contained in:
@@ -747,6 +747,11 @@ def _one_sided_data_worker(rank, world_size):
|
||||
top_k=experts_per_token,
|
||||
num_experts=num_experts,
|
||||
hidden_size=hidden_size,
|
||||
# Account for the fp8 block-scale payload (a1q_scale: hidden//16 bytes
|
||||
# per token) that is dispatched alongside the nvfp4 hidden states.
|
||||
# Without this the dispatch region is under-reserved and the combine
|
||||
# payload overflows the per-rank workspace.
|
||||
dispatch_scale_bytes_per_token=hidden_size // 16,
|
||||
)
|
||||
assert manager.initialized
|
||||
assert manager.moe_alltoall is not None
|
||||
|
||||
Reference in New Issue
Block a user