-
Notifications
You must be signed in to change notification settings - Fork 12
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
add benchmark for network topology with requiredDuringSchedulingIgnor…
…edDuringExecution (#112) Signed-off-by: Dmitry Shmulevich <[email protected]>
- Loading branch information
Showing
7 changed files
with
328 additions
and
40 deletions.
There are no files selected for viewing
67 changes: 67 additions & 0 deletions
67
resources/benchmarks/nwtopo/templates/jobset/jobset-acc.yaml
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,67 @@ | ||
# Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. | ||
# | ||
# Licensed under the Apache License, Version 2.0 (the "License"); | ||
# you may not use this file except in compliance with the License. | ||
# You may obtain a copy of the License at | ||
# | ||
# http://www.apache.org/licenses/LICENSE-2.0 | ||
# | ||
# Unless required by applicable law or agreed to in writing, software | ||
# distributed under the License is distributed on an "AS IS" BASIS, | ||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
# See the License for the specific language governing permissions and | ||
# limitations under the License. | ||
|
||
apiVersion: jobset.x-k8s.io/v1alpha2 | ||
kind: JobSet | ||
metadata: | ||
name: "{{._NAME_}}" | ||
namespace: default | ||
spec: | ||
# We want to declare our JobSet successful if workers finish. | ||
# If workers finish we should clean up the remaining replicatedJobs. | ||
successPolicy: | ||
operator: All | ||
targetReplicatedJobs: | ||
- workers | ||
replicatedJobs: | ||
- name: workers | ||
replicas: 1 | ||
template: | ||
spec: | ||
backoffLimit: 0 | ||
completions: {{.replicas}} | ||
parallelism: {{.replicas}} | ||
completionMode: NonIndexed | ||
template: | ||
metadata: | ||
labels: | ||
app: {{._NAME_}} | ||
annotations: | ||
pod-complete.stage.kwok.x-k8s.io/delay: "{{.ttl}}" | ||
pod-complete.stage.kwok.x-k8s.io/jitter-delay: "{{.ttl}}" | ||
spec: | ||
schedulerName: default-scheduler | ||
affinity: | ||
podAffinity: | ||
requiredDuringSchedulingIgnoredDuringExecution: | ||
- labelSelector: | ||
matchExpressions: | ||
- key: app | ||
operator: In | ||
values: | ||
- {{._NAME_}} | ||
topologyKey: network.topology.kubernetes.io/accelerator | ||
containers: | ||
- name: test | ||
image: ubuntu | ||
imagePullPolicy: IfNotPresent | ||
resources: | ||
limits: | ||
cpu: 100m | ||
memory: 250M | ||
nvidia.com/gpu: "8" | ||
requests: | ||
cpu: 100m | ||
memory: 250M | ||
nvidia.com/gpu: "8" |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
23 changes: 23 additions & 0 deletions
23
resources/benchmarks/nwtopo/workflows/config-jobset-acc.yaml
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,23 @@ | ||
# Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. | ||
# | ||
# Licensed under the Apache License, Version 2.0 (the "License"); | ||
# you may not use this file except in compliance with the License. | ||
# You may obtain a copy of the License at | ||
# | ||
# http://www.apache.org/licenses/LICENSE-2.0 | ||
# | ||
# Unless required by applicable law or agreed to in writing, software | ||
# distributed under the License is distributed on an "AS IS" BASIS, | ||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
# See the License for the specific language governing permissions and | ||
# limitations under the License. | ||
|
||
name: config-jobset | ||
tasks: | ||
- id: register | ||
type: RegisterObj | ||
params: | ||
template: "resources/benchmarks/nwtopo/templates/jobset/jobset-acc.yaml" | ||
nameFormat: "jobset{{._ENUM_}}" | ||
podNameFormat: "{{._NAME_}}-workers-[0-9]+-.+" | ||
podCount: "{{.replicas}}" |
145 changes: 145 additions & 0 deletions
145
resources/benchmarks/nwtopo/workflows/config-nodes-acc.yaml
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,145 @@ | ||
# Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved. | ||
# | ||
# Licensed under the Apache License, Version 2.0 (the "License"); | ||
# you may not use this file except in compliance with the License. | ||
# You may obtain a copy of the License at | ||
# | ||
# http://www.apache.org/licenses/LICENSE-2.0 | ||
# | ||
# Unless required by applicable law or agreed to in writing, software | ||
# distributed under the License is distributed on an "AS IS" BASIS, | ||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
# See the License for the specific language governing permissions and | ||
# limitations under the License. | ||
|
||
name: config-nw-topo-nodes | ||
description: | | ||
Create a 12-nodes cluster with a tree-like network topology | ||
and mark 5 nodes as busy: | ||
__________ sw31 __________ | ||
/ | \ | ||
sw21 sw22 sw23 | ||
/ \ / \ / \ | ||
sw11 sw12 sw13 sw14 sw15 sw16 | ||
/\ /\ /\ /\ /\ /\ | ||
n1 n2 n3 n4 n5 n6 n7 n8 n9 n10 n11 n12 | ||
|___|___|___|___|___| |___|___|___|___|___| | ||
| | | ||
nvl1 nvl2 | ||
Then deploy a 3-replicas job. The optimal nodes from the | ||
network topology perspective for this job are nodes n5, n7, n8. | ||
tasks: | ||
- id: configure | ||
type: Configure | ||
params: | ||
nodes: | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n1 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw11 | ||
network.topology.kubernetes.io/spine: sw21 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n2 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw11 | ||
network.topology.kubernetes.io/spine: sw21 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n3 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw12 | ||
network.topology.kubernetes.io/spine: sw21 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n4 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw12 | ||
network.topology.kubernetes.io/spine: sw21 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n5 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw13 | ||
network.topology.kubernetes.io/spine: sw22 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n6 | ||
network.topology.kubernetes.io/accelerator: nvl1 | ||
network.topology.kubernetes.io/block: sw13 | ||
network.topology.kubernetes.io/spine: sw22 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n7 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw14 | ||
network.topology.kubernetes.io/spine: sw22 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n8 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw14 | ||
network.topology.kubernetes.io/spine: sw22 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n9 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw15 | ||
network.topology.kubernetes.io/spine: sw23 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n10 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw15 | ||
network.topology.kubernetes.io/spine: sw23 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n11 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw16 | ||
network.topology.kubernetes.io/spine: sw23 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
- type: dgxa100.80g | ||
count: 1 | ||
labels: | ||
node-id: n12 | ||
network.topology.kubernetes.io/accelerator: nvl2 | ||
network.topology.kubernetes.io/block: sw16 | ||
network.topology.kubernetes.io/spine: sw23 | ||
network.topology.kubernetes.io/datacenter: sw31 | ||
nvidia.com/gpu.count: "8" | ||
timeout: 5m |
Oops, something went wrong.