update benchmarks

Signed-off-by: Dmitry Shmulevich <dshmulevich@nvidia.com>

Author: dmitsh

resources/benchmarks/README.md

@@ -29,28 +29,16 @@ To run the benchmark test for Kueue:
 ./scripts/benchmarks/gang-scheduling/run-kueue.sh
 ```
 
-To run the benchmark test for Run:ai
-
-```bash
-./scripts/benchmarks/gang-scheduling/run-runai.sh
-```
-
 ## Scaling Benchmark Test
 
-The scaling benchmark workflow operates on 700 virtual GPU nodes with tho workflows. The first [workflow](scaling/workflows/run-test-multi.yaml) submits is a job with 700 replicas, the second [workflow](scaling/workflows/run-test-single.yaml) submits a batch of 700 single-node jobs.
+The scaling benchmark workflow operates on 700 virtual GPU nodes. The [workflow](scaling/workflows/run-test.yaml) submits a batch of 700 single-node jobs.
 
 ### Example
 
 To run the benchmark test for Volcano:
 
 ```bash
-./bin/knavigator -workflow 'resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-volcano.yaml,run-test-multi.yaml}'
-```
-
-To run the benchmark test for Run:ai
-
-```bash
-./bin/knavigator -workflow 'resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-runai.yaml,runai-test-single.yaml}'
+./scripts/benchmarks/scaling/run-volcano.sh
 ```
 
 ## Network Topology Benchmark Test

resources/benchmarks/gang-scheduling/workflows/config-yunikorn.yaml

@@ -39,6 +39,6 @@ tasks:
                 - name: sandbox
                   submitacl: '*'
                   resources:
-                    max:
-                      {memory: 36Gi, vcore: 8000m, nvidia.com/gpu: 256}
+                    guaranteed: {memory: 36Gi, vcore: 8000m, nvidia.com/gpu: 256}
+                    max: {memory: 36Gi, vcore: 8000m, nvidia.com/gpu: 256}
     timeout: 1m

resources/benchmarks/scaling/workflows/config-runai.yaml

@@ -15,7 +15,7 @@
 name: config-runai
 description: register, deploy and configure run:ai custom resources
 tasks:
-- id: register-trainingworkload
+- id: register
   type: RegisterObj
   params:
     template: "resources/benchmarks/templates/runai/trainingworkload.yaml"

resources/benchmarks/scaling/workflows/run-test-multi.yaml

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-name: test-scaling-single-node-jobs
+name: test-scaling
 description: deploy 700 single-replica jobs
 tasks:
 - id: job

…s/scaling/workflows/run-test-single.yaml …nchmarks/scaling/workflows/run-test.yamlresources/benchmarks/scaling/workflows/run-test-single.yaml renamed to resources/benchmarks/scaling/workflows/run-test.yaml resources/benchmarks/scaling/workflows/run-test-single.yaml renamed to resources/benchmarks/scaling/workflows/run-test.yaml

@@ -18,4 +18,4 @@ set -e
 
 REPO_HOME=$(readlink -f $(dirname $(readlink -f "$0"))/../../../)
 
-$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-kai.yaml,run-test-single.yaml}"
+$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-kai.yaml,run-test.yaml}"

resources/benchmarks/scaling/workflows/runai-test-multi.yaml

@@ -18,4 +18,4 @@ set -e
 
 REPO_HOME=$(readlink -f $(dirname $(readlink -f "$0"))/../../../)
 
-$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-kueue.yaml,run-test-single.yaml}"
+$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-kueue.yaml,run-test.yaml}"

resources/benchmarks/scaling/workflows/runai-test-single.yaml

@@ -18,4 +18,4 @@ set -e
 
 REPO_HOME=$(readlink -f $(dirname $(readlink -f "$0"))/../../../)
 
-$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-volcano.yaml,run-test-single.yaml}"
+$REPO_HOME/bin/knavigator -workflow "$REPO_HOME/resources/benchmarks/scaling/workflows/{config-nodes.yaml,config-volcano.yaml,run-test.yaml}"