AWS dlami medium security documentation change
Summary
Added support for P6e-GB200 instances, updated component versions (NVIDIA driver, CUDA stack, EFA installer), and expanded technical specifications
Security assessment
The update includes a NVIDIA driver version upgrade (570.133.20 to 570.158.01) which likely contains security patches. The documentation now explicitly states CUDA version requirements and network interface configurations for new P6e-GB200 instances, which helps users implement secure configurations. The security relevance comes from the driver update (common vector for security fixes) and improved configuration guidance.
Diff
diff --git a/dlami/latest/devguide/aws-deep-learning-arm64-base-gpu-ami-amazon-linux-2023.md b/dlami/latest/devguide/aws-deep-learning-arm64-base-gpu-ami-amazon-linux-2023.md index 62fdb4f20..254ea0236 100644 --- a//dlami/latest/devguide/aws-deep-learning-arm64-base-gpu-ami-amazon-linux-2023.md +++ b//dlami/latest/devguide/aws-deep-learning-arm64-base-gpu-ami-amazon-linux-2023.md @@ -5 +5 @@ -AMI name formatSupported EC2 instancesAMI contentsNoticesRelease Date: 2025-04-24Release Date: 2025-04-22Release Date: 2025-04-04Release Date: 2025-03-03Release Date: 2025-02-14 +AMI name formatSupported EC2 instancesAMI contentsNoticesRelease Date: 2025-07-04Release Date: 2025-04-24Release Date: 2025-04-22Release Date: 2025-04-04Release Date: 2025-03-03Release Date: 2025-02-14 @@ -20 +20 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * G5g + * G5g, P6e-GB200(CUDA>=12.8 is supported on P6e-GB200) @@ -33 +33 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **Linux Kernel:** 6.12 + * **Latest available version is installed for the following packages:** @@ -35 +35,19 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **NVIDIA Driver** : 570.133.20 + * **Linux Kernel: 6**.**12** + + * **FSx Lustre** + + * **Docker** + + * **AWS CLI v2** at /usr/bin/aws + + * **NVIDIA DCGM** + + * **Nvidia container toolkit** : + + * Version command: nvidia-container-cli -V + + * **Nvidia-docker2** : + + * Version command: nvidia-docker version + + * **NVIDIA Driver** : 570.158.01 @@ -51 +69 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * For CUDA directory of 12.8, compiled NCCL Version 2.26.2+CUDA12.8 + * For CUDA directory of 12.8, compiled NCCL Version 2.27.5+CUDA12.8 @@ -65 +83 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **AWS CLI v2** at /usr/local/bin/aws + * **EFA installer** : 1.42.0 @@ -67 +85 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **EBS volume type** : gp3 + * **Nvidia GDRCopy:** 2.5.1 @@ -69 +87 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **Nvidia container toolkit** : 1.17.4 + * **AWS OFI NCCL plugin** comes with EFA installer @@ -71 +89,3 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * Version command: nvidia-container-cli -V + * Paths /opt/amazon/ofi-nccl/lib and /opt/amazon/ofi-nccl/efa are added to LD_LIBRARY_PATH. + + * **AWS CLI v2** at /usr/local/bin/aws @@ -73 +93 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - * **Docker:** 25.0.5 + * **EBS volume type** : gp3 @@ -79 +99,5 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm - aws ssm get-parameter --name/aws/service/deeplearning/ami/arm64/base-oss-nvidia-driver-gpu-amazon-linux-2023/latest/ami-id --region us-east-1 --query "Parameter.Value" --output text + SSM_PARAMETER=base-oss-nvidia-driver-gpu-amazon-linux-2023/latest/ami-id \ + aws ssm get-parameter --region us-east-1 \ + --name /aws/service/deeplearning/ami/arm64/$SSM_PARAMETER \ + --query "Parameter.Value" \ + --output text @@ -87,0 +112,29 @@ For help getting started, see [Getting started with DLAMI](./getting-started.htm +###### P6e-GB200 instances + +P6e-GB200 instances contain 17 network interface cards, and can be launched using the following AWS CLI command: + + + aws ec2 run-instances --region $REGION \ + --instance-type $INSTANCETYPE \ + --image-id $AMI --key-name $KEYNAME \ + --iam-instance-profile "Name=dlami-builder" \ + --tag-specifications "ResourceType=instance,Tags=[{Key=Name,Value=$TAG}]" \ + --network-interfaces \ + "NetworkCardIndex=0,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=interface" \ + "NetworkCardIndex=1,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=2,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=3,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=4,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=5,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=6,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=7,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=8,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=9,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=10,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=11,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=12,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=13,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=14,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=15,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" \ + "NetworkCardIndex=16,DeviceIndex=0,Groups=$SG,SubnetId=$SUBNET,InterfaceType=efa-only" + @@ -113,0 +167,17 @@ These AMIs Components of this AMI like CUDA versions may be removed and changed +## Release Date: 2025-07-04 + +**AMI name** : Deep Learning ARM64 Base OSS Nvidia Driver GPU AMI (Amazon Linux 2023) 20250704 + +### Updated + + * Added support to EC2 instance P6e-GB200. Please note that CUDA>=12.8 is supported on P6e-GB200 + + * Add EFA 1.42.0 + + * Upgraded Nvidia driver from version 570.133.20 to 570.158.01 + + * Upgraded CUDA 12.8 stack with NCCL 2.27.5 + + + +