NCCL tests¶

This example shows how to run distributed NCCL tests with MPI using dstack.

Running as a task¶

Here's an example of a task that runs AllReduce test on 2 nodes, each with 4 GPUs (8 processes in total).

type: task
name: nccl-tests

nodes: 2

startup_order: workers-first
stop_criteria: master-done

env:
  - NCCL_DEBUG=INFO
commands:
  - |
    if [ $DSTACK_NODE_RANK -eq 0 ]; then
      mpirun \
        --allow-run-as-root \
        --hostfile $DSTACK_MPI_HOSTFILE \
        -n $DSTACK_GPUS_NUM \
        -N $DSTACK_GPUS_PER_NODE \
        --bind-to none \
        /opt/nccl-tests/build/all_reduce_perf -b 8 -e 8G -f 2 -g 1
    else
      sleep infinity
    fi

# Uncomment if the `kubernetes` backend requires it for `/dev/infiniband` access
#privileged: true

resources:
  gpu: nvidia:1..8
  shm_size: 16GB

Default image

If you don't specify image, dstack uses its base Docker image pre-configured with uv, python, pip, essential CUDA drivers, mpirun, and NCCL tests (under /opt/nccl-tests/build).

Privileged

In some cases, the backend (e.g., kubernetes) may require privileged: true to access the high-speed interconnect (e.g., InfiniBand).

Apply a configuration¶

To run a configuration, use the dstack apply command.

$ dstack apply -f examples/clusters/nccl-tests/.dstack.yml

 #  BACKEND  REGION     INSTANCE       RESOURCES                                   SPOT  PRICE
 1  aws      us-east-1  g4dn.12xlarge  48xCPU, 192GB, 4xT4 (16GB), 100.0GB (disk)  no    $3.912
 2  aws      us-west-2  g4dn.12xlarge  48xCPU, 192GB, 4xT4 (16GB), 100.0GB (disk)  no    $3.912
 3  aws      us-east-2  g4dn.12xlarge  48xCPU, 192GB, 4xT4 (16GB), 100.0GB (disk)  no    $3.912

Submit the run nccl-tests? [y/n]: y

Source code¶

The source-code of this example can be found in examples/clusters/nccl-tests .

What's next?¶

Check dev environments, tasks, services, and fleets.