EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

EPLB: prefer to use physical experts in the same node (#9849)

Browse Source
This commit is contained in:
Yingchun Lai
2025-09-22 15:34:30 +08:00
committed by GitHub
parent f05805515b
commit 9d7e82a0ab
1 changed files with 29 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
python/sglang/srt/eplb/expert_location.py
View File

@@ -231,6 +231,7 @@ class ExpertLocationMetadata:
logical_to_all_physical_map_num_valid=logical_to_all_physical_map_num_valid,
logical_to_rank_dispatch_physical_map=(
compute_logical_to_rank_dispatch_physical_map(
server_args=server_args,
logical_to_all_physical_map=logical_to_all_physical_map,
num_gpus=ep_size,
num_physical_experts=num_physical_experts,
@@ -340,6 +341,7 @@ def _pad_nested_array(arr, pad_value):
# TODO optimize performance (rewrite and/or run in separate process with overlap)
def compute_logical_to_rank_dispatch_physical_map(
server_args: ServerArgs,
logical_to_all_physical_map: torch.Tensor,
num_gpus: int,
num_physical_experts: int,
@@ -348,7 +350,9 @@ def compute_logical_to_rank_dispatch_physical_map(
):
r = random.Random(seed)
num_local_physical_experts = num_physical_experts // num_gpus
num_local_gpu_physical_experts = num_physical_experts // num_gpus
num_gpus_per_node = server_args.ep_size // server_args.nnodes
num_local_node_physical_experts = num_local_gpu_physical_experts * num_gpus_per_node
num_layers, num_logical_experts, _ = logical_to_all_physical_map.shape
dtype = logical_to_all_physical_map.dtype
@@ -372,13 +376,28 @@ def compute_logical_to_rank_dispatch_physical_map(
physical_expert_id
for physical_expert_id in candidate_physical_expert_ids
if _compute_gpu_id_of_physical_expert(
physical_expert_id, num_local_physical_experts
physical_expert_id, num_local_gpu_physical_experts
)
== gpu_id
]
if len(same_gpu_physical_expert_ids) > 0:
# 1. Prefer same-GPU experts
output_partial[gpu_id] = same_gpu_physical_expert_ids[0]
else:
# 2. Otherwise, prefer same-node experts
node_id = gpu_id // num_gpus_per_node
same_node_physical_expert_ids = [
physical_expert_id
for physical_expert_id in candidate_physical_expert_ids
if _compute_node_id_of_physical_expert(
physical_expert_id, num_local_node_physical_experts
)
== node_id
]
if len(same_node_physical_expert_ids) > 0:
output_partial[gpu_id] = same_node_physical_expert_ids[0]
# 3. Fill remaining slots with fair random choices
num_remain = torch.sum(output_partial == -1).item()
output_partial[output_partial == -1] = torch.tensor(
_fair_choices(candidate_physical_expert_ids, k=num_remain, r=r),
@@ -404,9 +423,15 @@ def _logical_to_all_physical_raw(
def _compute_gpu_id_of_physical_expert(
physical_expert_id: int, num_local_physical_experts: int
physical_expert_id: int, num_local_gpu_physical_experts: int
) -> int:
return physical_expert_id // num_local_physical_experts
return physical_expert_id // num_local_gpu_physical_experts
def _compute_node_id_of_physical_expert(
physical_expert_id: int, num_local_host_physical_experts: int
) -> int:
return physical_expert_id // num_local_host_physical_experts
def _fair_choices(arr: List, k: int, r: random.Random) -> List: