Skip to content

Commit ca0551d

Browse files
authored
fix(recipe): bump NCCL all-reduce bandwidth threshold to 300 Gbps (#350)
1 parent eab65d9 commit ca0551d

File tree

1 file changed

+2
-2
lines changed

1 file changed

+2
-2
lines changed

recipes/overlays/h100-eks-ubuntu-training.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -58,7 +58,7 @@ spec:
5858
- nccl-all-reduce-bw
5959
constraints:
6060
- name: nccl-all-reduce-bw
61-
value: ">= 100"
61+
value: ">= 300"
6262
conformance:
6363
checks:
6464
- platform-health
@@ -70,4 +70,4 @@ spec:
7070
- pod-autoscaling
7171
- cluster-autoscaling
7272
- robust-controller
73-
- secure-accelerator-access
73+
- secure-accelerator-access

0 commit comments

Comments
 (0)