CUDA-Q on Amazon Braket (Workshop Studio)

README.md

GPU インスタンス

1. Workshop Studio で起動している既存の Amazon Braket (Amazon SageMaker AI) Notebook ml.t3.2xlarge を停止 (Stop) する。
2. Amazon SageMaker AI サービスの画面で、インスタンスタイプを ml.p3.2xlarge に変更する。https://us-east-1.console.aws.amazon.com/sagemaker/home?region=us-east-1#/notebook-instances
3. その Notebook インスタンスを起動 (Start) する。
4. Jupyter の kernel config vi ~/.local/share/jupyter/kernels/docker_cudaq/kernel.json に3行追記。 https://gist.github.com/hariby/47c2a32755969f4fb3c6235d718390a7#file-kernel-json-L8-L10

Notebook

• cudaq_target.ipynb
  • n = 26 まで for n in range(10, 27):

optional

sudo vi /etc/docker/daemon.json で以下の default-runtime を追記して、

,
    "default-runtime": {
        "nvidia": {
            "args": [],
            "path": "nvidia-container-runtime"
        }
    }

ビルド ./custom-install-script.sh ws-event-55c13874-08f-us-east-1 1a98a208-4616-4f90-bed0-4735061132c5

CUDA-Q on Braket

https://github.com/amazon-braket/amazon-braket-examples/tree/main/examples/nvidia_cuda_q

Container image

(amazon-braket/amazon-braket-examples#738 で修正済み)

• ECR image name に prefix (amazon-braket-) つける
• リージョンを us-east-1 に変える !container/container_build_and_push.sh amazon-braket-cudaq-job us-east-1

QPU

IQM Garnet

• Availability window: Wednesday, 07:30:00 - 15:59:59 UTC = Wednesday, 16:30:00 - 24:59:59 JST
• eu-north-1 cf. QPU Region mapping https://docs.aws.amazon.com/braket/latest/developerguide/braket-devices.html

kernel.json

{
 "argv": [
  "/usr/bin/docker",
  "run",
  "--network=host",
  "-v",
  "{connection_file}:/connection-spec",
  "--gpus", "all",
  "--env", "NVIDIA_VISIBLE_DEVICES=all",
  "--env", "NVIDIA_DRIVER_CAPABILITIES=compute,utility",
  "--mount",
  "type=bind,source=/home/ec2-user/braket_workshops,target=/home/cudaq/braket_workshops",
  "--mount",
  "type=bind,source=/home/ec2-user/amazon-braket-examples/examples,target=/home/cudaq/braket_examples",
  "cudaq-jupyter-kernel",
  "python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "/connection-spec"
 ],
 "display_name": "docker_cudaq",
 "language": "python"
}