Run GitHub Actions on AWS p4d.24xlarge

Why this fits

• 96 vCPU and 1152 GiB RAM — sized for workloads the default GitHub-hosted runners can't comfortably host, all on your own AWS account.
• Multi-GPU training-style integration tests — 8× A100 in one node.
• NVIDIA A100 40GB for PyTorch / JAX and other CUDA workloads, with drivers + CUDA pre-baked in the runner image.

.cirun.yml

.cirun.yml

1runners:
2  - name: aws-p4d-24xlarge
3    cloud: aws
4    instance_type: p4d.24xlarge
5    # Use AWS Deep Learning AMI GPU PyTorch on Ubuntu 22.04, or the
6    # Cirun-published NVIDIA AMI.
7    machine_image: ami-04823729c75214919
8    labels:
9      - cirun-aws-p4d-24xlarge

Drop this in your repo root. The first workflow that requests the runner label spins this configuration up on your cloud account.