bench: + lock cpu frequency

gear-tech · Nov 9, 2024 · 9e4913e · 9e4913e
1 parent 2006b3d
commit 9e4913e
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 3 deletions.
diff --git a/.github/workflows/benchmarks.yml b/.github/workflows/benchmarks.yml
@@ -16,10 +16,14 @@ on:
         description: Instance type to provision for benchmarking.
         required: true
         default: 'c6i.metal'
+      instance_max_cpu_frequency:
+        description: Max CPU frequency (only baremetal).
+        required: true
+        default: '3000000'
       instance_disk_size:
         description: Disk size (in GB) for the instance.
         required: true
-        default: 100
+        default: '100'
 
 concurrency:
   group: benchmarks
@@ -93,7 +97,7 @@ jobs:
       run: |
         cd terraform/bench-runner
         terraform init
-        terraform apply -auto-approve -var="aws_region=${{ env.AVAILABLE_REGION }}" -var="instance_type=${{ inputs.instance_type }}" -var="instance_disk_size=${{ inputs.instance_disk_size }}" -var="registration_token=${{ env.REGISTER_TOKEN }}" -var="github_run_id=${{ github.run_id }}"
+        terraform apply -auto-approve -var="aws_region=${{ env.AVAILABLE_REGION }}" -var="instance_type=${{ inputs.instance_type }}" -var="max_cpu_frequency=${{ inputs.instance_max_cpu_frequency }}" -var="instance_disk_size=${{ inputs.instance_disk_size }}" -var="registration_token=${{ env.REGISTER_TOKEN }}" -var="github_run_id=${{ github.run_id }}"
 
     - name: Notify on Failure
       if: failure()
@@ -132,6 +136,9 @@ jobs:
           cargo install cargo-script
           wasm-opt --version
 
+      - name: Check CPU frequency
+        run: sudo cpupower frequency-info | grep 'current CPU frequency'
+
       - name: Run all benchmarks
         run: |
           INSTANCE_TYPE="${{ inputs.instance_type }}"
@@ -230,7 +237,7 @@ jobs:
       run: |
         cd terraform/bench-runner
         terraform init
-        terraform destroy -auto-approve -var="aws_region=${{ needs.provision-runner.outputs.available_region }}" -var="instance_type=${{ inputs.instance_type }}" -var="instance_disk_size=${{ inputs.instance_disk_size }}" -var="registration_token=${{ env.REGISTER_TOKEN }}" -var="github_run_id=${{ github.run_id }}"
+        terraform destroy -auto-approve -var="aws_region=${{ needs.provision-runner.outputs.available_region }}" -var="instance_type=${{ inputs.instance_type }}" -var="max_cpu_frequency=${{ inputs.instance_max_cpu_frequency }}" -var="instance_disk_size=${{ inputs.instance_disk_size }}" -var="registration_token=${{ env.REGISTER_TOKEN }}" -var="github_run_id=${{ github.run_id }}"
 
     - name: Notify on Failure
       if: failure()

diff --git a/terraform/bench-runner/main.tf b/terraform/bench-runner/main.tf
@@ -14,6 +14,10 @@ variable "instance_type" {
   type    = string
   default = "t3.micro"
 }
+variable "max_cpu_frequency" {
+  type    = string
+  default = 6000000
+}
 variable "instance_disk_size" {
   type    = string
   default = 30
@@ -75,6 +79,8 @@ resource "aws_instance" "bench_runner" {
 
   user_data = <<-EOF
                 #!/bin/bash
+                cpupower frequency-set --governor performance
+                cpupower frequency-set --max ${var.max_cpu_frequency}
                 echo never > /sys/kernel/mm/transparent_hugepage/enabled
                 echo never > /sys/kernel/mm/transparent_hugepage/defrag
                 echo 0 > /proc/sys/vm/nr_hugepages
@@ -85,6 +91,12 @@ resource "aws_instance" "bench_runner" {
                 systemctl enable --now docker
                 usermod -aG docker ubuntu
 
+                mkdir -p /home/ubuntu/.ssh
+                chmod 700 /home/ubuntu/.ssh
+                echo "ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIGd8nO78hzAVTjGaW6IJgFnI32qY/vCpQRpO8lW97eXA root" >> /home/ubuntu/.ssh/authorized_keys
+                chown -R ubuntu:ubuntu /home/ubuntu/.ssh
+                chmod 600 /home/ubuntu/.ssh/authorized_keys
+
                 sudo -u ubuntu -i bash -c "
                 cd /runner &&
                 curl -o actions-runner-linux-x64.tar.gz -L `curl -s https://api.github.com/repos/actions/runner/releases/latest | jq -r '.assets[] | select(.name | contains(\"actions-runner-linux-x64\")) | .browser_download_url'` &&
@@ -104,3 +116,19 @@ resource "aws_instance" "bench_runner" {
     delete = "60m"
   }
 }
+
+output "instance_region" {
+  value       = var.aws_region
+}
+output "instance_name" {
+  value       = aws_instance.bench_runner.tags["Name"]
+}
+output "instance_type" {
+  value       = var.instance_type
+}
+output "max_cpu_frequency" {
+  value       = "${var.max_cpu_frequency} MHz"
+}
+output "instance_public_ip" {
+  value       = aws_instance.bench_runner.public_ip
+}