Quickstart: run GPU and CPU workloads on SolarCompute

1. Request access

Contact hello@solarcompute.net.au or use the Request a Quote page to describe the workloads, estimated GPU/CPU hours, and any special hardware or network requirements. We'll respond with next steps and onboarding details.

2. Typical workload types

• GPU training and fine-tuning (PyTorch/TensorFlow)
• GPU inference and batch processing
• CPU batch jobs, data preprocessing, and rendering
• Containerised workloads (Docker/Kubernetes)

3. Example: containerised GPU job

Most developers run workloads as containers. Example: run a container that checks GPU availability.

docker run --gpus all --rm nvidia/cuda:12.1-base nvidia-smi

4. Example: Kubernetes pod requesting a GPU

apiVersion: v1
kind: Pod
metadata:
  name: gpu-test
spec:
  containers:
  - name: gpu-test
    image: nvidia/cuda:12.1-base
    command: ["nvidia-smi"]
    resources:
      limits:
        nvidia.com/gpu: 1

5. Best practices

• Containerise dependencies to ensure portability.
• Use resource limits and checkpoints for long runs.
• Prefer batch-friendly, fault-tolerant workloads on distributed nodes.
• For large datasets, use object storage or staged transfer; avoid repeated large transfers per job.

6. Developer support & integration

After access is granted we provide technical onboarding, machine specs, and recommended instance images. For commercial access and machine lists, see the Machines page or Request a Quote.