[MoE] enable all2all fi_one_sided by default (#53311)

Signed-off-by: Artem Perevedentsev <aperevedents@nvidia.com>
This commit is contained in:
Artem Perevedentsev
2026-08-26 17:03:45 +05:00
committed by GitHub
parent 31739ceb0b
commit 2cd6c664c1
2 changed files with 7 additions and 3 deletions
@@ -815,6 +815,10 @@ class FlashInferNVLinkOneSidedManager(All2AllManagerBase):
comm_backend=CustomCommunicator(self.cpu_group),
)
# Release cached allocator blocks before FlashInfer reserves the
# symmetric MNNVL fabric workspace.
torch.accelerator.empty_cache()
self.moe_alltoall = MoeAlltoAll(
mapping=self.mapping,
max_num_tokens=self.max_num_tokens,