Get started with EFA and NCCL for ML workloads on Amazon EC2 - Amazon Elastic Compute Cloud
Step 1: Prepare an EFA-enabled security groupStep 2: Launch a temporary instanceStep 3: Install Nvidia GPU drivers, Nvidia CUDA toolkit, and cuDNNStep 4: Install GDRCopyStep 5: Install the EFA softwareStep 6: Install NCCLStep 7: Install the aws-ofi-nccl pluginStep 8: Install the NCCL testsStep 9: Test your EFA and NCCL configurationStep 10: Install your machine learning applicationsStep 11: Create an EFA and NCCL-enabled AMIStep 12: Terminate the temporary instanceStep 13: Launch EFA and NCCL-enabled instances into a cluster placement groupStep 14: Enable passwordless SSH

Get started with EFA and NCCL for ML workloads on Amazon EC2

The NVIDIA Collective Communications Library (NCCL) is a library of standard collective communication routines for multiple GPUs across a single node or multiple nodes. NCCL can be used together with EFA, Libfabric, and MPI to support various machine learning workloads. For more information, see the NCCL website.

The following steps help you to get started with EFA and NCCL using a base AMI for one of the supported operating systems.

Note

  • Only the p3dn.24xlarge, p4d.24xlarge, p5.48xlarge instance types are supported.

  • Only Amazon Linux 2 and Ubuntu 20.04/22.04 base AMIs are supported.

  • Only NCCL 2.4.2 and later is supported with EFA.

  • For more information about running machine learning workloads with EFA and NCCL using an AWS Deep Learning AMIs, see Using EFA on the DLAMI in the AWS Deep Learning AMIs Developer Guide.

Step 1: Prepare an EFA-enabled security group

An EFA requires a security group that allows all inbound and outbound traffic to and from the security group itself. The following procedure creates a security group that allows all inbound and outbound traffic to and from itself, and that allows inbound SSH traffic from any IPv4 address for SSH connectivity.

Important

This security group is intended for testing purposes only. For your production environments, we recommend that you create an inbound SSH rule that allows traffic only from the IP address from which you are connecting, such as the IP address of your computer, or a range of IP addresses in your local network.

For other scenarios, see Security group rules for different use cases.

To create an EFA-enabled security group

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. In the navigation pane, choose Security Groups and then choose Create security group.

  3. In the Create security group window, do the following:

    1. For Security group name, enter a descriptive name for the security group, such as EFA-enabled security group.

    2. (Optional) For Description, enter a brief description of the security group.

    3. For VPC, select the VPC into which you intend to launch your EFA-enabled instances.

    4. Choose Create security group.

  4. Select the security group that you created, and on the Details tab, copy the Security group ID.

  5. With the security group still selected, choose Actions, Edit inbound rules, and then do the following:

    1. Choose Add rule.

    2. For Type, choose All traffic.

    3. For Source type, choose Custom and paste the security group ID that you copied into the field.

    4. Choose Add rule.

    5. For Type, choose SSH.

    6. For Source type, choose Anywhere-IPv4.

    7. Choose Save rules.

  6. With the security group still selected, choose Actions, Edit outbound rules, and then do the following:

    1. Choose Add rule.

    2. For Type, choose All traffic.

    3. For Destination type, choose Custom and paste the security group ID that you copied into the field.

    4. Choose Save rules.

Step 2: Launch a temporary instance

Launch a temporary instance that you can use to install and configure the EFA software components. You use this instance to create an EFA-enabled AMI from which you can launch your EFA-enabled instances.

To launch a temporary instance

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. In the navigation pane, choose Instances, and then choose Launch Instances to open the new launch instance wizard.

  3. (Optional) In the Name and tags section, provide a name for the instance, such as EFA-instance. The name is assigned to the instance as a resource tag (Name=EFA-instance).

  4. In the Application and OS Images section, select an AMI for one of the supported operating systems. Only Amazon Linux 2, Ubuntu 20.04, and Ubuntu 22.04 are supported.

  5. In the Instance type section, select either p3dn.24xlarge, p4d.24xlarge, or p5.48xlarge.

  6. In the Key pair section, select the key pair to use for the instance.

  7. In the Network settings section, choose Edit, and then do the following:

    1. For Subnet, choose the subnet in which to launch the instance. If you do not select a subnet, you can't enable the instance for EFA.

    2. For Firewall (security groups), choose Select existing security group, and then select the security group that you created in the previous step.

    3. Expand the Advanced network configuration section.

      For Network interface 1, select Network card index = 0, Device index = 0, and Interface type = EFA with ENA.

      (Optional) If you are using a multi-card instance type, such as p4d.24xlarge or p5.48xlarge, for each additional network interface required, choose Add network interface, for Network card index select the next unused index, and then select Device index = 1 and Interface type = EFA with ENA or EFA-only.

  8. In the Storage section, configure the volumes as needed.

    Note

    You must provision an additional 10 to 20 GiB of storage for the Nvidia CUDA Toolkit. If you do not provision enough storage, you will receive an insufficient disk space error when attempting to install the Nvidia drivers and CUDA toolkit.

  9. In the Summary panel on the right, choose Launch instance.

Step 3: Install Nvidia GPU drivers, Nvidia CUDA toolkit, and cuDNN

Amazon Linux 2
To install the Nvidia GPU drivers, Nvidia CUDA toolkit, and cuDNN

  1. To ensure that all of your software packages are up to date, perform a quick software update on your instance.

    $ sudo yum upgrade -y && sudo reboot

    After the instance has rebooted, reconnect to it.

  2. Install the utilities that are needed to install the Nvidia GPU drivers and the Nvidia CUDA toolkit.

    $ sudo yum groupinstall 'Development Tools' -y

  3. Disable the nouveau open source drivers.

    1. Install the required utilities and the kernel headers package for the version of the kernel that you are currently running.

      $ sudo yum install -y wget kernel-devel-$(uname -r) kernel-headers-$(uname -r)

    2. Add nouveau to the /etc/modprobe.d/blacklist.conf deny list file.

      $ cat << EOF | sudo tee --append /etc/modprobe.d/blacklist.conf
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv
EOF

    3. Append GRUB_CMDLINE_LINUX="rdblacklist=nouveau" to the grub file and rebuild the Grub configuration.

      $ echo 'GRUB_CMDLINE_LINUX="rdblacklist=nouveau"' | sudo tee -a /etc/default/grub \
&& sudo grub2-mkconfig -o /boot/grub2/grub.cfg

  4. Reboot the instance and reconnect to it.

  5. Prepare the required repositories

    1. Install the EPEL repository for DKMS and enable any optional repos for your Linux distribution.

      $ sudo yum install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm

    2. Install the CUDA repository public GPG key.

      $ distribution='rhel7'

    3. Set up the CUDA network repository and update the repository cache.

      $ ARCH=$( /bin/arch ) \
&& sudo yum-config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/$distribution/${ARCH}/cuda-$distribution.repo \
&& sudo yum clean expire-cache

    4. (Kernel version 5.10 only) Perform these steps only if you are using Amazon Linux 2 with kernel version 5.10. If you are using Amazon Linux 2 with kernel version 4.12, skip these steps. To check your kernel version, run uname -r.

      1. Create the Nvidia driver configuration file named /etc/dkms/nvidia.conf.

        $ sudo mkdir -p /etc/dkms \
&& echo "MAKE[0]=\"'make' -j2 module SYSSRC=\${kernel_source_dir} IGNORE_XEN_PRESENCE=1 IGNORE_PREEMPT_RT_PRESENCE=1 IGNORE_CC_MISMATCH=1 CC=/usr/bin/gcc10-gcc\"" | sudo tee /etc/dkms/nvidia.conf

      2. (p4d.24xlarge and p5.48xlarge only) Copy the Nvidia driver configuration file.

        $ sudo cp /etc/dkms/nvidia.conf /etc/dkms/nvidia-open.conf

  6. Install the Nvidia GPU drivers, NVIDIA CUDA toolkit, and cuDNN.

    • p3dn.24xlarge

      $ sudo yum clean all \
&& sudo yum -y install kmod-nvidia-latest-dkms nvidia-driver-latest-dkms \
&& sudo yum -y install cuda-drivers-fabricmanager cuda libcudnn8-devel

    • p4d.24xlarge and p5.48xlarge

      $ sudo yum clean all \
&& sudo yum -y install kmod-nvidia-open-dkms nvidia-driver-latest-dkms \
&& sudo yum -y install cuda-drivers-fabricmanager cuda libcudnn8-devel

  7. Reboot the instance and reconnect to it.

  8. (p4d.24xlarge and p5.48xlarge only) Start the Nvidia Fabric Manager service, and ensure that it starts automatically when the instance starts. Nvidia Fabric Manager is required for NV Switch Management.

    $ sudo systemctl enable nvidia-fabricmanager && sudo systemctl start nvidia-fabricmanager

  9. Ensure that the CUDA paths are set each time that the instance starts.

    • For bash shells, add the following statements to /home/username/.bashrc and /home/username/.bash_profile. 

      export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH

    • For tcsh shells, add the following statements to /home/username/.cshrc.

      setenv PATH=/usr/local/cuda/bin:$PATH
setenv LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH

  10. To confirm that the Nvidia GPU drivers are functional, run the following command.

    $ nvidia-smi -q | head

    The command should return information about the Nvidia GPUs, Nvidia GPU drivers, and Nvidia CUDA toolkit.

Ubuntu 20.04/22.04
To install the Nvidia GPU drivers, Nvidia CUDA toolkit, and cuDNN

  1. To ensure that all of your software packages are up to date, perform a quick software update on your instance.

    $ sudo apt-get update && sudo apt-get upgrade -y

  2. Install the utilities that are needed to install the Nvidia GPU drivers and the Nvidia CUDA toolkit.

    $ sudo apt-get update && sudo apt-get install build-essential -y

  3. To use the Nvidia GPU driver, you must first disable the nouveau open source drivers.

    1. Install the required utilities and the kernel headers package for the version of the kernel that you are currently running.

      $ sudo apt-get install -y gcc make linux-headers-$(uname -r)

    2. Add nouveau to the /etc/modprobe.d/blacklist.conf deny list file.

      $ cat << EOF | sudo tee --append /etc/modprobe.d/blacklist.conf
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv
EOF

    3. Open /etc/default/grub using your preferred text editor and add the following. 

      GRUB_CMDLINE_LINUX="rdblacklist=nouveau"

    4. Rebuild the Grub configuration.

      $ sudo update-grub

  4. Reboot the instance and reconnect to it.

  5. Add the CUDA repository and install the Nvidia GPU drivers, NVIDIA CUDA toolkit, and cuDNN.

    • p3dn.24xlarge

      $ sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/7fa2af80.pub \
&& wget -O /tmp/deeplearning.deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/nvidia-machine-learning-repo-ubuntu2004_1.0.0-1_amd64.deb \
&& sudo dpkg -i /tmp/deeplearning.deb \
&& wget -O /tmp/cuda.pin https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin \
&& sudo mv /tmp/cuda.pin /etc/apt/preferences.d/cuda-repository-pin-600 \
&& sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub \
&& sudo add-apt-repository 'deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /' \
&& sudo apt update \
&& sudo apt install nvidia-dkms-535 \
&& sudo apt install -o Dpkg::Options::='--force-overwrite' cuda-drivers-535 cuda-toolkit-12-3 libcudnn8 libcudnn8-dev -y

    • p4d.24xlarge and p5.48xlarge

      $ sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/7fa2af80.pub \
&& wget -O /tmp/deeplearning.deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/nvidia-machine-learning-repo-ubuntu2004_1.0.0-1_amd64.deb \
&& sudo dpkg -i /tmp/deeplearning.deb \
&& wget -O /tmp/cuda.pin https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin \
&& sudo mv /tmp/cuda.pin /etc/apt/preferences.d/cuda-repository-pin-600 \
&& sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub \
&& sudo add-apt-repository 'deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /' \
&& sudo apt update \
&& sudo apt install nvidia-kernel-open-535 \
&& sudo apt install -o Dpkg::Options::='--force-overwrite' cuda-drivers-535 cuda-toolkit-12-3 libcudnn8 libcudnn8-dev -y

  6. Reboot the instance and reconnect to it.

  7. (p4d.24xlarge and p5.48xlarge only) Install the Nvidia Fabric Manager.

    1. You must install the version of the Nvidia Fabric Manager that matches the version of the Nvidia kernel module that you installed in the previous step.

      Run the following command to determine the version of the Nvidia kernel module.

      $ cat /proc/driver/nvidia/version | grep "Kernel Module"

      The following is example output.

      NVRM version: NVIDIA UNIX x86_64 Kernel Module  450.42.01  Tue Jun 15 21:26:37 UTC 2021

      In the example above, major version 450 of the kernel module was installed. This means that you need to install Nvidia Fabric Manager version 450.

    2. Install the Nvidia Fabric Manager. Run the following command and specify the major version identified in the previous step.

      $ sudo apt install -o Dpkg::Options::='--force-overwrite' nvidia-fabricmanager-major_version_number

      For example, if major version 450 of the kernel module was installed, use the following command to install the matching version of Nvidia Fabric Manager.

      $ sudo apt install -o Dpkg::Options::='--force-overwrite' nvidia-fabricmanager-450

    3. Start the service, and ensure that it starts automatically when the instance starts. Nvidia Fabric Manager is required for NV Switch Management.

      $ sudo systemctl start nvidia-fabricmanager && sudo systemctl enable nvidia-fabricmanager

  8. Ensure that the CUDA paths are set each time that the instance starts.

    • For bash shells, add the following statements to /home/username/.bashrc and /home/username/.bash_profile. 

      export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH

    • For tcsh shells, add the following statements to /home/username/.cshrc.

      setenv PATH=/usr/local/cuda/bin:$PATH
setenv LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64:$LD_LIBRARY_PATH

  9. To confirm that the Nvidia GPU drivers are functional, run the following command.

    $ nvidia-smi -q | head

    The command should return information about the Nvidia GPUs, Nvidia GPU drivers, and Nvidia CUDA toolkit.

Step 4: Install GDRCopy

Install GDRCopy to improve the performance of Libfabric. For more information about GDRCopy, see the GDRCopy repository.

Amazon Linux 2
To install GDRCopy

  1. Install the required dependencies.

    $ sudo yum -y install dkms rpm-build make check check-devel subunit subunit-devel

  2. Download and extract the GDRCopy package.

    $ wget https://github.com/NVIDIA/gdrcopy/archive/refs/tags/v2.4.tar.gz \
&& tar xf v2.4.tar.gz ; cd gdrcopy-2.4/packages

  3. Build the GDRCopy RPM package.

    $ CUDA=/usr/local/cuda ./build-rpm-packages.sh

  4. Install the GDRCopy RPM package.

    $ sudo rpm -Uvh gdrcopy-kmod-2.4-1dkms.noarch*.rpm \
&& sudo rpm -Uvh gdrcopy-2.4-1.x86_64*.rpm \
&& sudo rpm -Uvh gdrcopy-devel-2.4-1.noarch*.rpm
Ubuntu 20.04/22.04
To install GDRCopy

  1. Install the required dependencies.

    $ sudo apt -y install build-essential devscripts debhelper check libsubunit-dev fakeroot pkg-config dkms

  2. Download and extract the GDRCopy package.

    $ wget https://github.com/NVIDIA/gdrcopy/archive/refs/tags/v2.4.tar.gz \
&& tar xf v2.4.tar.gz \
&& cd gdrcopy-2.4/packages

  3. Build the GDRCopy RPM package.

    $ CUDA=/usr/local/cuda ./build-deb-packages.sh

  4. Install the GDRCopy RPM package.

    $ sudo dpkg -i gdrdrv-dkms_2.4-1_amd64.*.deb \
&& sudo dpkg -i libgdrapi_2.4-1_amd64.*.deb \
&& sudo dpkg -i gdrcopy-tests_2.4-1_amd64.*.deb \
&& sudo dpkg -i gdrcopy_2.4-1_amd64.*.deb

Step 5: Install the EFA software

Install the EFA-enabled kernel, EFA drivers, Libfabric, and Open MPI stack that is required to support EFA on your temporary instance.

To install the EFA software

  1. Connect to the instance you launched. For more information, see Connect to your Linux instance using SSH.

  2. Download the EFA software installation files. The software installation files are packaged into a compressed tarball (.tar.gz) file. To download the latest stable version, use the following command.

    $ curl -O https://efa-installer.amazonaws.com/aws-efa-installer-1.35.0.tar.gz

    You can also get the latest version by replacing the version number with latest in the preceding command.

  3. (Optional) Verify the authenticity and integrity of the EFA tarball (.tar.gz) file.

    We recommend that you do this to verify the identity of the software publisher and to check that the file has not been altered or corrupted since it was published. If you do not want to verify the tarball file, skip this step.

    Note

    Alternatively, if you prefer to verify the tarball file by using an MD5 or SHA256 checksum instead, see Verify the EFA installer using a checksum.

    1. Download the public GPG key and import it into your keyring.

      $ wget https://efa-installer.amazonaws.com/aws-efa-installer.key && gpg --import aws-efa-installer.key

      The command should return a key value. Make a note of the key value, because you need it in the next step.

    2. Verify the GPG key's fingerprint. Run the following command and specify the key value from the previous step.

      $ gpg --fingerprint key_value

      The command should return a fingerprint that is identical to 4E90 91BC BB97 A96B 26B1 5E59 A054 80B1 DD2D 3CCC. If the fingerprint does not match, don't run the EFA installation script, and contact AWS Support.

    3. Download the signature file and verify the signature of the EFA tarball file.

      $ wget https://efa-installer.amazonaws.com/aws-efa-installer-1.35.0.tar.gz.sig && gpg --verify ./aws-efa-installer-1.35.0.tar.gz.sig

      The following shows example output.

      gpg: Signature made Wed 29 Jul 2020 12:50:13 AM UTC using RSA key ID DD2D3CCC
gpg: Good signature from "Amazon EC2 EFA <ec2-efa-maintainers@amazon.com>"
gpg: WARNING: This key is not certified with a trusted signature!
gpg:          There is no indication that the signature belongs to the owner.
Primary key fingerprint: 4E90 91BC BB97 A96B 26B1  5E59 A054 80B1 DD2D 3CCC

      If the result includes Good signature, and the fingerprint matches the fingerprint returned in the previous step, proceed to the next step. If not, don't run the EFA installation script, and contact AWS Support.

    Show moreShow less

  4. Extract the files from the compressed .tar.gz file and navigate into the extracted directory.

    $ tar -xf aws-efa-installer-1.35.0.tar.gz && cd aws-efa-installer

  5. Run the EFA software installation script.

    Note

    From EFA 1.30.0, both Open MPI 4 and Open MPI 5 are installed by default. Unless you need Open MPI 5, we recommend that you install only Open MPI 4. The following command installs Open MPI 4 only. If you want to install Open MPI 4 and Open MPI 5, remove --mpi=openmpi4.

    $ sudo ./efa_installer.sh -y --mpi=openmpi4

    Libfabric is installed in the /opt/amazon/efa directory, while Open MPI is installed in the /opt/amazon/openmpi directory.

  6. If the EFA installer prompts you to reboot the instance, do so and then reconnect to the instance. Otherwise, log out of the instance and then log back in to complete the installation.

  7. Confirm that the EFA software components were successfully installed.

    $ fi_info -p efa -t FI_EP_RDM

    The command should return information about the Libfabric EFA interfaces. The following example shows the command output.

    • p3dn.24xlarge with single network interface

      provider: efa
fabric: EFA-fe80::94:3dff:fe89:1b70
domain: efa_0-rdm
version: 2.0
type: FI_EP_RDM
protocol: FI_PROTO_EFA

    • p4d.24xlarge and p5.48xlarge with multiple network interfaces

      provider: efa
fabric: EFA-fe80::c6e:8fff:fef6:e7ff
domain: efa_0-rdm
version: 111.0
type: FI_EP_RDM
protocol: FI_PROTO_EFA
provider: efa
fabric: EFA-fe80::c34:3eff:feb2:3c35
domain: efa_1-rdm
version: 111.0
type: FI_EP_RDM
protocol: FI_PROTO_EFA
provider: efa
fabric: EFA-fe80::c0f:7bff:fe68:a775
domain: efa_2-rdm
version: 111.0
type: FI_EP_RDM
protocol: FI_PROTO_EFA
provider: efa
fabric: EFA-fe80::ca7:b0ff:fea6:5e99
domain: efa_3-rdm
version: 111.0
type: FI_EP_RDM
protocol: FI_PROTO_EFA

Step 6: Install NCCL

Install NCCL. For more information about NCCL, see the NCCL repository.

To install NCCL

  1. Navigate to the /opt directory.

    $ cd /opt

  2. Clone the official NCCL repository to the instance and navigate into the local cloned repository.

    $ sudo git clone https://github.com/NVIDIA/nccl.git && cd nccl

  3. Build and install NCCL and specify the CUDA installation directory.

    $ sudo make -j src.build CUDA_HOME=/usr/local/cuda

Step 7: Install the aws-ofi-nccl plugin

The aws-ofi-nccl plugin maps NCCL's connection-oriented transport APIs to Libfabric's connection-less reliable interface. This enables you to use Libfabric as a network provider while running NCCL-based applications. For more information about the aws-ofi-nccl plugin, see the aws-ofi-nccl repository.

To install the aws-ofi-nccl plugin

  1. Navigate to your home directory.

    $ cd $HOME

  2. Install the required utilities.

    • Amazon Linux 2

      $ sudo yum install hwloc-devel

    • Ubuntu

      $ sudo apt-get install libhwloc-dev

  3. Download the aws-ofi-nccl plugin files. The files are packaged into a compressed tarball (.tar.gz).

    $ wget https://github.com/aws/aws-ofi-nccl/releases/download/v1.11.0-aws/aws-ofi-nccl-1.11.0-aws.tar.gz

  4. Extract the files from the compressed .tar.gz file and navigate into the extracted directory.

    $ tar -xf aws-ofi-nccl-1.11.0-aws.tar.gz && cd aws-ofi-nccl-1.11.0-aws

  5. To generate the make files, run the configure script and specify the MPI, Libfabric, NCCL, and CUDA installation directories. 

    $ ./configure --prefix=/opt/aws-ofi-nccl --with-mpi=/opt/amazon/openmpi \
--with-libfabric=/opt/amazon/efa \
--with-cuda=/usr/local/cuda \
--enable-platform-aws

  6. Add the Open MPI directory to the PATH variable.

    $ export PATH=/opt/amazon/openmpi/bin/:$PATH

  7. Install the aws-ofi-nccl plugin.

    $ make && sudo make install

Step 8: Install the NCCL tests

Install the NCCL tests. The NCCL tests enable you to confirm that NCCL is properly installed and that it is operating as expected. For more information about the NCCL tests, see the nccl-tests repository.

To install the NCCL tests

  1. Navigate to your home directory.

    $ cd $HOME

  2. Clone the official nccl-tests repository to the instance and navigate into the local cloned repository.

    $ git clone https://github.com/NVIDIA/nccl-tests.git && cd nccl-tests

  3. Add the Libfabric directory to the LD_LIBRARY_PATH variable.

    • Amazon Linux 2

      $ export LD_LIBRARY_PATH=/opt/amazon/efa/lib64:$LD_LIBRARY_PATH

    • Ubuntu

      $ export LD_LIBRARY_PATH=/opt/amazon/efa/lib:$LD_LIBRARY_PATH

  4. Install the NCCL tests and specify the MPI, NCCL, and CUDA installation directories.

    $ make MPI=1 MPI_HOME=/opt/amazon/openmpi NCCL_HOME=/opt/nccl/build CUDA_HOME=/usr/local/cuda

Step 9: Test your EFA and NCCL configuration

Run a test to ensure that your temporary instance is properly configured for EFA and NCCL.

To test your EFA and NCCL configuration

  1. Create a host file that specifies the hosts on which to run the tests. The following command creates a host file named my-hosts that includes a reference to the instance itself.

    IMDSv2
    [ec2-user ~]$ TOKEN=`curl -X PUT "http://169.254.169.254/latest/api/token" -H "X-aws-ec2-metadata-token-ttl-seconds: 21600"` \
&& curl -H "X-aws-ec2-metadata-token: $TOKEN" -v http://169.254.169.254/latest/meta-data/local-ipv4 >> my-hosts
    IMDSv1
    [ec2-user ~]$ curl http://169.254.169.254/latest/meta-data/local-ipv4 >> my-hosts

  2. Run the test and specify the host file (--hostfile) and the number of GPUs to use (-n). The following command runs the all_reduce_perf test on 8 GPUs on the instance itself, and specifies the following environment variables.

    • FI_EFA_USE_DEVICE_RDMA=1—(p4d.24xlarge only) uses the device's RDMA functionality for one-sided and two-sided transfer.

    • NCCL_DEBUG=INFO—enables detailed debugging output. You can also specify VERSION to print only the NCCL version at the start of the test, or WARN to receive only error messages.

    For more information about the NCCL test arguments, see the NCCL Tests README in the official nccl-tests repository.

    • p3dn.24xlarge

      $ /opt/amazon/openmpi/bin/mpirun \
-x LD_LIBRARY_PATH=/opt/nccl/build/lib:/usr/local/cuda/lib64:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:$LD_LIBRARY_PATH \
-x NCCL_DEBUG=INFO \
--hostfile my-hosts -n 8 -N 8 \
--mca pml ^cm --mca btl tcp,self --mca btl_tcp_if_exclude lo,docker0 --bind-to none \
$HOME/nccl-tests/build/all_reduce_perf -b 8 -e 1G -f 2 -g 1 -c 1 -n 100

    • p4d.24xlarge and p5.48xlarge

      $ /opt/amazon/openmpi/bin/mpirun \
-x FI_EFA_USE_DEVICE_RDMA=1 \
-x LD_LIBRARY_PATH=/opt/nccl/build/lib:/usr/local/cuda/lib64:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:$LD_LIBRARY_PATH \
-x NCCL_DEBUG=INFO \
--hostfile my-hosts -n 8 -N 8 \
--mca pml ^cm --mca btl tcp,self --mca btl_tcp_if_exclude lo,docker0 --bind-to none \
$HOME/nccl-tests/build/all_reduce_perf -b 8 -e 1G -f 2 -g 1 -c 1 -n 100

  3. You can confirm that EFA is active as the underlying provider for NCCL when the NCCL_DEBUG log is printed.

    ip-192-168-2-54:14:14 [0] NCCL INFO NET/OFI Selected Provider is efa*

    The following additional information is displayed when using a p4d.24xlarge instance.

    ip-192-168-2-54:14:14 [0] NCCL INFO NET/OFI Running on P4d platform, Setting NCCL_TOPO_FILE environment variable to /home/ec2-user/install/plugin/share/aws-ofi-nccl/xml/p4d-24xl-topo.xml

Step 10: Install your machine learning applications

Install the machine learning applications on the temporary instance. The installation procedure varies depending on the specific machine learning application. For more information about installing software on your Linux instance, see Manage software on your Amazon Linux 2 instance.

Note

Refer to your machine learning application’s documentation for installation instructions.

Step 11: Create an EFA and NCCL-enabled AMI

After you have installed the required software components, you create an AMI that you can reuse to launch your EFA-enabled instances.

To create an AMI from your temporary instance

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. In the navigation pane, choose Instances.

  3. Select the temporary instance that you created and choose Actions, Image, Create image.

  4. For Create image, do the following:

    1. For Image name, enter a descriptive name for the AMI.

    2. (Optional) For Image description, enter a brief description of the purpose of the AMI.

    3. Choose Create image.

  5. In the navigation pane, choose AMIs.

  6. Locate the AMI tht you created in the list. Wait for the status to change from pending to available before continuing to the next step.

Step 12: Terminate the temporary instance

At this point, you no longer need the temporary instance that you launched. You can terminate the instance to stop incurring charges for it.

To terminate the temporary instance

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. In the navigation pane, choose Instances.

  3. Select the temporary instance that you created and then choose Actions, Instance state, Terminate instance.

  4. When prompted for confirmation, choose Terminate.

Step 13: Launch EFA and NCCL-enabled instances into a cluster placement group

Launch your EFA and NCCL-enabled instances into a cluster placement group using the EFA-enabled AMI and the EFA-enabled security group that you created earlier.

Note

  • It is not an absolute requirement to launch your EFA-enabled instances into a cluster placement group. However, we do recommend running your EFA-enabled instances in a cluster placement group as it launches the instances into a low-latency group in a single Availability Zone.

  • To ensure that capacity is available as you scale your cluster’s instances, you can create a Capacity Reservation for your cluster placement group. For more information, see Create Capacity Reservations in cluster placement groups.

New console
To launch a temporary instance

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. In the navigation pane, choose Instances, and then choose Launch Instances to open the new launch instance wizard.

  3. (Optional) In the Name and tags section, provide a name for the instance, such as EFA-instance. The name is assigned to the instance as a resource tag (Name=EFA-instance).

  4. In the Application and OS Images section, choose My AMIs, and then select the AMI that you created in the previous step.

  5. In the Instance type section, select either p3dn.24xlarge or p4d.24xlarge.

  6. In the Key pair section, select the key pair to use for the instance.

  7. In the Network settings section, choose Edit, and then do the following:

    1. For Subnet, choose the subnet in which to launch the instance. If you do not select a subnet, you can't enable the instance for EFA.

    2. For Firewall (security groups), choose Select existing security group, and then select the security group that you created in the previous step.

    3. Expand the Advanced network configuration section.

      For Network interface 1, select Network card index = 0, Device index = 0, and Interface type = EFA with ENA.

      (Optional) If you are using a multi-card instance type, such as p4d.24xlarge or p5.48xlarge, for each additional network interface required, choose Add network interface, for Network card index select the next unused index, and then select Device index = 1 and Interface type = EFA eith ENA or EFA-only.

  8. (Optional) In the Storage section, configure the volumes as needed.

  9. In the Advanced details section, for Placement group name, select the cluster placement group into which to launch the instance. If you need to create a new cluster placement group, choose Create new placement group.

  10. In the Summary panel on the right, for Number of instances, enter the number of EFA-enabled instances that you want to launch, and then choose Launch instance.

Old console
To launch your EFA and NCCL-enabled instances into a cluster placement group

  1. Open the Amazon EC2 console at https://console.aws.amazon.com/ec2/.

  2. Choose Launch Instance.

  3. On the Choose an AMI page, choose My AMIs, find the AMI that you created earlier, and then choose Select.

  4. On the Choose an Instance Type page, select p3dn.24xlarge and then choose Next: Configure Instance Details.

  5. On the Configure Instance Details page, do the following:

    1. For Number of instances, enter the number of EFA and NCCL-enabled instances that you want to launch.

    2. For Network and Subnet, select the VPC and subnet into which to launch the instances.

    3. For Placement group, select Add instance to placement group.

    4. For Placement group name, select Add to a new placement group, and then enter a descriptive name for the placement group. Then for Placement group strategy, select cluster.

    5. For EFA, choose Enable.

    6. In the Network Interfaces section, for device eth0, choose New network interface. You can optionally specify a primary IPv4 address and one or more secondary IPv4 addresses. If you are launching the instance into a subnet that has an associated IPv6 CIDR block, you can optionally specify a primary IPv6 address and one or more secondary IPv6 addresses.

    7. Choose Next: Add Storage.

  6. On the Add Storage page, specify the volumes to attach to the instances in addition to the volumes specified by the AMI (such as the root device volume). Then choose Next: Add Tags.

  7. On the Add Tags page, specify tags for the instances, such as a user-friendly name, and then choose Next: Configure Security Group.

  8. On the Configure Security Group page, for Assign a security group, select Select an existing security group, and then select the security group that you created earlier.

  9. Choose Review and Launch.

  10. On the Review Instance Launch page, review the settings, and then choose Launch to choose a key pair and to launch your instances.

Step 14: Enable passwordless SSH

To enable your applications to run across all of the instances in your cluster, you must enable passwordless SSH access from the leader node to the member nodes. The leader node is the instance from which you run your applications. The remaining instances in the cluster are the member nodes.

To enable passwordless SSH between the instances in the cluster

  1. Select one instance in the cluster as the leader node, and connect to it.

  2. Disable strictHostKeyChecking and enable ForwardAgent on the leader node. Open ~/.ssh/config using your preferred text editor and add the following.

    Host *
    ForwardAgent yes
Host *
    StrictHostKeyChecking no

  3. Generate an RSA key pair.

    $ ssh-keygen -t rsa -N "" -f ~/.ssh/id_rsa

    The key pair is created in the $HOME/.ssh/ directory.

  4. Change the permissions of the private key on the leader node.

    $ chmod 600 ~/.ssh/id_rsa
chmod 600 ~/.ssh/config

  5. Open ~/.ssh/id_rsa.pub using your preferred text editor and copy the key.

  6. For each member node in the cluster, do the following:

    1. Connect to the instance.

    2. Open ~/.ssh/authorized_keys using your preferred text editor and add the public key that you copied earlier.

  7. To test that the passwordless SSH is functioning as expected, connect to your leader node and run the following command.

    $ ssh member_node_private_ip

    You should connect to the member node without being prompted for a key or password.