Skip to content

Commit ff3618e

Browse files
sirutBuasaisirutBuasai
committed
revert release notes
Signed-off-by: sirutBuasai <sirutbuasai27@outlook.com>
1 parent 765a451 commit ff3618e

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

47 files changed

+1760
-0
lines changed

‎docs/releasenotes/base/cuda-12.8-ec2.md‎

Lines changed: 46 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,46 @@
1+
# AWS Deep Learning Base Containers for EC2, ECS, EKS (CUDA 12.8)
2+
3+
[AWS Deep Learning Containers (DLCs)](https://aws.amazon.com/machine-learning/containers/) now support Base images that serve as a foundational layer
4+
to build the machine learning environment on EC2, ECS and EKS, with Ubuntu 24.04.
5+
6+
These Base DLCs package the essential deep learning components and dependencies without being tied to a specific framework implementation, providing
7+
users the flexibility to customize the DLCs with their preferred frameworks.
8+
9+
## Release Notes
10+
11+
- Development Tools: Includes curl, build-essential, cmake, and git
12+
- Python Environment: Python 3.12 with AWS CLI, boto3, and requests pre-installed
13+
- GPU Support: CUDA 12.8.1 with cuda-compat for backward compatibility
14+
- Neural Network Libraries: cuDNN 9.8.0.87 for deep neural network operations
15+
- Distributed Training: NCCL 2.26.2-1 for multi-GPU and multi-node communication
16+
- Network Performance: EFA 1.40.0 for low-latency network communications
17+
18+
## Security Advisory
19+
20+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
21+
22+
## Python Support
23+
24+
Python 3.12 is supported.
25+
26+
## GPU Instance Type Support
27+
28+
- CUDA 12.8
29+
- cuDNN 9.8.0.87
30+
- NCCL 2.26.2-1
31+
32+
## Example URL
33+
34+
```
35+
763104351884.dkr.ecr.us-east-1.amazonaws.com/base:12.8.1-gpu-py312-cu128-ubuntu24.04-ec2
36+
```
37+
38+
## Build and Test
39+
40+
- Built on: c5.18xlarge
41+
- Tested on: p4d.24xlarge, p4de.24xlarge, p5.48xlarge
42+
- Tested with: [openclip](https://github.com/mlfoundations/open_clip), [nccl-tests](https://github.com/NVIDIA/nccl-tests)
43+
44+
## Known Issues
45+
46+
No known issues so far.

‎docs/releasenotes/base/cuda-12.9-ec2.md‎

Lines changed: 46 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,46 @@
1+
# AWS Deep Learning Base Containers for EC2, ECS, EKS (CUDA 12.9)
2+
3+
[AWS Deep Learning Containers (DLCs)](https://aws.amazon.com/machine-learning/containers/) now support Base images that serve as a foundational layer
4+
to build the machine learning environment on EC2, ECS and EKS, with Ubuntu 22.04.
5+
6+
These Base DLCs package the essential deep learning components and dependencies without being tied to a specific framework implementation, providing
7+
users the flexibility to customize the DLCs with their preferred frameworks.
8+
9+
## Release Notes
10+
11+
- Development Tools: Includes curl, build-essential, cmake, and git
12+
- Python Environment: Python 3.12 with AWS CLI, boto3, and requests pre-installed
13+
- GPU Support: CUDA 12.9.1 with cuda-compat for backward compatibility
14+
- Neural Network Libraries: cuDNN 9.10.2.21 for deep neural network operations
15+
- Distributed Training: NCCL 2.27.3-1 for multi-GPU and multi-node communication
16+
- Network Performance: EFA 1.43.1 for low-latency network communications
17+
18+
## Security Advisory
19+
20+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
21+
22+
## Python Support
23+
24+
Python 3.12 is supported.
25+
26+
## GPU Instance Type Support
27+
28+
- CUDA 12.9
29+
- cuDNN 9.10.2.21
30+
- NCCL 2.27.3-1
31+
32+
## Example URL
33+
34+
```
35+
763104351884.dkr.ecr.us-east-1.amazonaws.com/base:12.9.1-gpu-py312-cu129-ubuntu22.04-ec2
36+
```
37+
38+
## Build and Test
39+
40+
- Built on: c5.18xlarge
41+
- Tested on: p4d.24xlarge, p4de.24xlarge, p5.48xlarge
42+
- Tested with: [openclip](https://github.com/mlfoundations/open_clip), [nccl-tests](https://github.com/NVIDIA/nccl-tests)
43+
44+
## Known Issues
45+
46+
No known issues so far.

‎docs/releasenotes/base/cuda-13.0-ec2.md‎

Lines changed: 46 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,46 @@
1+
# AWS Deep Learning Base Containers for EC2, ECS, EKS (CUDA 13.0)
2+
3+
[AWS Deep Learning Containers (DLCs)](https://aws.amazon.com/machine-learning/containers/) now support Base images that serve as a foundational layer
4+
to build the machine learning environment on EC2, ECS and EKS, with Ubuntu 22.04.
5+
6+
These Base DLCs package the essential deep learning components and dependencies without being tied to a specific framework implementation, providing
7+
users the flexibility to customize the DLCs with their preferred frameworks.
8+
9+
## Release Notes
10+
11+
- Development Tools: Includes curl, build-essential, cmake, and git
12+
- Python Environment: Python 3.12 with AWS CLI, boto3, and requests pre-installed
13+
- GPU Support: CUDA 13.0.0 with cuda-compat for backward compatibility
14+
- Neural Network Libraries: cuDNN 9.13.0.50 for deep neural network operations
15+
- Distributed Training: NCCL 2.27.7-1 for multi-GPU and multi-node communication
16+
- Network Performance: EFA 1.44.0 for low-latency network communications
17+
18+
## Security Advisory
19+
20+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
21+
22+
## Python Support
23+
24+
Python 3.12 is supported.
25+
26+
## GPU Instance Type Support
27+
28+
- CUDA 13.0
29+
- cuDNN 9.13.0.50
30+
- NCCL 2.27.7-1
31+
32+
## Example URL
33+
34+
```
35+
763104351884.dkr.ecr.us-east-1.amazonaws.com/base:13.0.0-gpu-py312-cu130-ubuntu22.04-ec2
36+
```
37+
38+
## Build and Test
39+
40+
- Built on: c5.18xlarge
41+
- Tested on: p4d.24xlarge, p4de.24xlarge, p5.48xlarge
42+
- Tested with: [openclip](https://github.com/mlfoundations/open_clip), [nccl-tests](https://github.com/NVIDIA/nccl-tests)
43+
44+
## Known Issues
45+
46+
No known issues so far.

‎docs/releasenotes/base/index.md‎

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
# Base Container Release Notes
2+
3+
Release notes for AWS Deep Learning Base Containers with CUDA support.
4+
5+
## CUDA 13.0
6+
7+
| Platform | Type | Link |
8+
| --- | --- | --- |
9+
| EC2, ECS, EKS | General | [Base CUDA 13.0 on EC2, ECS, EKS](cuda-13.0-ec2.md) |
10+
11+
## CUDA 12.9
12+
13+
| Platform | Type | Link |
14+
| --- | --- | --- |
15+
| EC2, ECS, EKS | General | [Base CUDA 12.9 on EC2, ECS, EKS](cuda-12.9-ec2.md) |
16+
17+
## CUDA 12.8
18+
19+
| Platform | Type | Link |
20+
| --- | --- | --- |
21+
| EC2, ECS, EKS | General | [Base CUDA 12.8 on EC2, ECS, EKS](cuda-12.8-ec2.md) |

‎docs/releasenotes/pytorch/2.4-graviton-ec2-ecs-eks.md‎

Lines changed: 73 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,73 @@
1+
# AWS Deep Learning Containers for PyTorch 2.4 Graviton on EC2, ECS, and EKS
2+
3+
[AWS Deep Learning Containers](https://aws.amazon.com/machine-learning/containers/) (DLC) for Amazon Elastic Kubernetes Service (EKS), Amazon Elastic
4+
Compute Cloud (EC2), and Amazon Elastic Container Service (ECS) are now available for the [Graviton](https://aws.amazon.com/ec2/graviton/) instance
5+
type with support for PyTorch 2.4.
6+
7+
This release includes container images for inference on CPU and GPU, optimized for performance and scale on AWS. The CPU image has been tested with
8+
each of the EC2, ECS, and EKS services, while the GPU image only supports EC2 (see the table below). The GPU image provides stable versions of NVIDIA
9+
CUDA, cuDNN, NCCL, and other components. All software components in these images are scanned for security vulnerabilities and updated or patched in
10+
accordance with AWS Security best practices.
11+
12+
| | EC2 | ECS | EKS |
13+
| --- | --- | --- | --- |
14+
| Graviton CPU | Supported | Supported | Supported |
15+
| Graviton with NVIDIA GPU | Supported | Not Supported | Not Supported |
16+
17+
## Release Notes
18+
19+
- Introduced containers for PyTorch 2.4 for inference supporting EC2, ECS, and EKS on Graviton instances. For details about this release, check out
20+
our GitHub release tags: [for CPU](https://github.com/aws/deep-learning-containers/releases/tag/v1.0-pt-graviton-ec2-2.4.0-inf-cpu-py311) and
21+
[for GPU](https://github.com/aws/deep-learning-containers/releases/tag/v1.0-pt-graviton-ec2-2.4.0-inf-gpu-py311).
22+
- TorchServe version: 0.11.1
23+
- 11/01/24: Updated TorchServe to 0.12.0 (release tags:
24+
[for CPU](https://github.com/aws/deep-learning-containers/releases/tag/v1.5-pt-graviton-ec2-2.4.0-inf-cpu-py311) and
25+
[for GPU](https://github.com/aws/deep-learning-containers/releases/tag/v1.4-pt-graviton-ec2-2.4.0-inf-gpu-py311))
26+
- The GPU image is the first ever DLC supporting Graviton (ARM64) + GPU platforms. It should be used with the
27+
[G5g instance type](https://aws.amazon.com/ec2/instance-types/g5g/), which is powered by Graviton CPUs and NVIDIA T4G Tensor Core GPUs.
28+
- Please refer to the official PyTorch 2.4 release notes [here](https://github.com/pytorch/pytorch/releases/tag/v2.4.0) for framework updates.
29+
30+
## Performance Improvements
31+
32+
These DLCs continue to deliver the best performance on Graviton CPU for BERT and RoBERTa sentiment analysis and fill mask models, making Graviton3 the
33+
most cost effective CPU platform on the AWS cloud for these models. For more information, please refer to the
34+
[Graviton PyTorch User Guide](https://github.com/aws/aws-graviton-getting-started/blob/main/machinelearning/pytorch.md).
35+
36+
## Security Advisory
37+
38+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
39+
40+
## Python 3.11 Support
41+
42+
Python 3.11 is supported in the PyTorch Graviton Inference containers.
43+
44+
## CPU Instance Type Support
45+
46+
The containers support Graviton CPU instance types supported under each of the above mentioned services.
47+
48+
## GPU Instance Type Support
49+
50+
The containers support the Graviton GPU instance type G5g and contain the following software components for GPU support:
51+
52+
- CUDA 12.4.0
53+
- cuDNN 9.1.0.70+cuda12.4
54+
- NCCL 2.20.5+cuda12.4
55+
56+
## Build and Test
57+
58+
- Built on: c6g.2xlarge
59+
- Tested on: c7g.4xlarge, c6g.4xlarge, t4g.2xlarge, r6g.2xlarge, m6g.4xlarge, g5g.4xlarge
60+
- Tested with [MNIST](http://yann.lecun.com/exdb/mnist/) and Resnet50/DenseNet datasets on EC2, ECS AMI (Amazon Linux AMI 2.0.20220822 arm64) and EKS
61+
AMI (1.25.6-20230304 arm64)
62+
63+
## Known Issues
64+
65+
- There is no official [Triton](https://github.com/triton-lang/triton) distribution for ARM64/aarch64 yet, so some torch.compile workloads will fail
66+
with:
67+
68+
```
69+
torch._dynamo.exc.BackendCompilerFailed: backend='inductor' raised:
70+
RuntimeError: Cannot find a working triton installation. More information on installing Triton can be found at https://github.com/openai/triton
71+
```
72+
73+
For latest updates, please refer to the [aws/deep-learning-containers GitHub repo](https://github.com/aws/deep-learning-containers/tags).

‎docs/releasenotes/pytorch/2.4-graviton-sagemaker.md‎

Lines changed: 52 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,52 @@
1+
# AWS Deep Learning Containers for PyTorch 2.4 Graviton on SageMaker
2+
3+
[AWS Deep Learning Containers](https://aws.amazon.com/machine-learning/containers/) (DLCs) for Amazon SageMaker are now available for the
4+
[Graviton](https://aws.amazon.com/ec2/graviton/) instance type with support for PyTorch 2.4.
5+
6+
This release includes a container image for inference on CPU, optimized for performance and scale on AWS. This Docker image was tested on SageMaker.
7+
All software components in this image are scanned for security vulnerabilities and updated or patched in accordance with AWS Security best practices.
8+
9+
Please refer to the
10+
[SageMaker Graviton blog](https://aws.amazon.com/blogs/machine-learning/run-machine-learning-inference-workloads-on-aws-graviton-based-instances-with-amazon-sagemaker/)
11+
and DLC [developer guide](https://docs.aws.amazon.com/dlami/latest/devguide/deep-learning-containers.html) to migrate the Deep Learning workloads to
12+
Graviton instances.
13+
14+
## Release Notes
15+
16+
- Introduced container for PyTorch 2.4 for inference supporting SageMaker services on Graviton instances. For details about this release, check out
17+
our GitHub [release tag](https://github.com/aws/deep-learning-containers/releases/tag/v1.0-pt-graviton-sagemaker-2.4.0-inf-cpu-py311).
18+
- TorchServe version: 0.11.1
19+
- 10/25/24: Updated TorchServe to 0.12.0
20+
([release tag](https://github.com/aws/deep-learning-containers/releases/tag/v1.3-pt-graviton-sagemaker-2.4.0-inf-cpu-py311))
21+
- Please refer to the official PyTorch 2.4 release notes [here](https://github.com/pytorch/pytorch/releases/tag/v2.4.0) for framework updates.
22+
23+
## Performance Improvements
24+
25+
These DLCs continue to deliver the best performance on Graviton for BERT and RoBERTa sentiment analysis and fill mask models, making Graviton3 the
26+
most cost effective CPU platform on the AWS cloud for these models. For more information, please refer to the
27+
[Graviton PyTorch User Guide](https://github.com/aws/aws-graviton-getting-started/blob/main/machinelearning/pytorch.md).
28+
29+
## Security Advisory
30+
31+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
32+
33+
## Python 3.11 Support
34+
35+
Python 3.11 is supported in the PyTorch Graviton Inference containers.
36+
37+
## CPU Instance Type Support
38+
39+
The containers support Graviton CPU instance types supported under SageMaker.
40+
41+
## Build and Test
42+
43+
- Built on: c6g.2xlarge
44+
- Tested on: c7g.4xlarge, c6g.4xlarge, t4g.2xlarge, r6g.2xlarge, m6g.4xlarge
45+
- Tested with [MNIST](http://yann.lecun.com/exdb/mnist/) and Resnet50/DenseNet datasets on EC2, ECS AMI (Amazon Linux AMI 2.0.20220822 arm64) and EKS
46+
AMI (1.25.6-20230304 arm64)
47+
48+
## Known Issues
49+
50+
- None
51+
52+
For latest updates, please refer to the [aws/deep-learning-containers GitHub repo](https://github.com/aws/deep-learning-containers/tags).

‎docs/releasenotes/pytorch/2.4-inference-ec2-ecs-eks.md‎

Lines changed: 51 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,51 @@
1+
# AWS Deep Learning Containers for PyTorch 2.4 Inference on EC2, ECS and EKS
2+
3+
[AWS Deep Learning Containers](https://aws.amazon.com/machine-learning/containers/) (DLCs) for Amazon Elastic Compute Cloud (EC2), Amazon Elastic
4+
Container Service (ECS), and Amazon Elastic Kubernetes Service (EKS) are now available with PyTorch 2.4 and support for CUDA 12.4 on Ubuntu 22.04.
5+
6+
This release includes container images for inference on CPU and GPU, optimized for performance and scale on AWS. These Docker images have been tested
7+
with EC2, ECS and EKS services, and provide stable versions of NVIDIA CUDA, cuDNN, Intel MKL, and other components. All software components in these
8+
images are scanned for security vulnerabilities and updated or patched in accordance with AWS Security best practices. If you are looking for a DLC to
9+
use with SageMaker, please refer to
10+
[this documentation](https://github.com/aws/deep-learning-containers/blob/master/available_images.md#general-framework-containers-ec2-ecs-eks--sm-support).
11+
12+
## Release Notes
13+
14+
- Introduced containers for PyTorch 2.4.0 for inference supporting EC2, ECS, and EKS. For details about this release, check out our GitHub
15+
[release tag](https://github.com/aws/deep-learning-containers/releases/tag/v1.0-pt-ec2-2.4.0-inf-py311).
16+
- PyTorch 2.4 offers support for python custom operator API allowing users to integrate custom kernels such as Triton kernels into torch.compile.
17+
- Please refer to the official PyTorch 2.4 release notes [here](https://github.com/pytorch/pytorch/releases/tag/v2.4.0) for the full description of
18+
updates.
19+
- Added Python 3.11 support
20+
- Added CUDA 12.4 support
21+
- Added Ubuntu 22.04 support
22+
- The Dockerfile for CPU can be found
23+
[here](https://github.com/aws/deep-learning-containers/blob/master/pytorch/inference/docker/2.4/py3/Dockerfile.cpu), and the Dockerfile for GPU can
24+
be found [here](https://github.com/aws/deep-learning-containers/blob/master/pytorch/inference/docker/2.4/py3/cu124/Dockerfile.gpu).
25+
26+
## Security Advisory
27+
28+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
29+
30+
## Python 3.11 Support
31+
32+
Python 3.11 is supported in the PyTorch Inference containers.
33+
34+
## CPU Instance Type Support
35+
36+
The containers support x86_64 CPU instance types.
37+
38+
## GPU Instance Type Support
39+
40+
The containers support GPU instance types and contain the following software components for GPU support:
41+
42+
- CUDA 12.4.1
43+
- cuDNN 9.1.0.70+cuda12.4
44+
- NCCL 2.22.3+cuda12.4
45+
46+
## Build and Test
47+
48+
- Built on: c5.18xlarge
49+
- Tested on: c5.18xlarge, g3.16xlarge, m5.16xlarge, t3.2xlarge, p3.16xlarge, p3dn.24xlarge, p4d.24xlarge, g4dn.xlarge
50+
- Tested with [MNIST](http://yann.lecun.com/exdb/mnist/) and Resnet50/ImageNet datasets on EC2, ECS AMI (Amazon Linux AMI 2.0.20221102), and EKS AMI
51+
(amazon-eks-gpu-node-1.25.16-20240307)

‎docs/releasenotes/pytorch/2.4-inference-sagemaker.md‎

Lines changed: 52 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,52 @@
1+
# AWS Deep Learning Containers for PyTorch 2.4 Inference on SageMaker
2+
3+
[AWS Deep Learning Containers](https://aws.amazon.com/machine-learning/containers/) (DLC) for Amazon SageMaker are now available with support for
4+
PyTorch 2.4 and support for CUDA 12.4 on Ubuntu 22.04.
5+
6+
This release includes container images for inference on CPU and GPU, optimized for performance and scale on AWS. These Docker images have been tested
7+
with SageMaker services, and provide stable versions of NVIDIA CUDA, cuDNN, and other components. All software components in these images are scanned
8+
for security vulnerabilities and updated or patched in accordance with AWS Security best practices.
9+
10+
## Release Notes
11+
12+
- Introduced containers for PyTorch 2.4 for inference supporting SageMaker services. For details about this release, check out our GitHub
13+
[release tag](https://github.com/aws/deep-learning-containers/releases/tag/v1.0-pt-sagemaker-2.4.0-inf-py311).
14+
- PyTorch 2.4 offers support for python custom operator API allowing users to integrate custom kernels such as Triton kernels into torch.compile.
15+
- Please refer to the official PyTorch 2.4 release notes [here](https://github.com/pytorch/pytorch/releases/tag/v2.4.0) for the full description of
16+
updates.
17+
- Added Python 3.11 support
18+
- Added CUDA 12.4 support
19+
- Added Ubuntu 22.04 support
20+
- Added TorchServe 0.11.1 support
21+
- 10/25/24: Updated TorchServe to 0.12.0
22+
([release tag](https://github.com/aws/deep-learning-containers/releases/tag/v1.1-pt-sagemaker-2.4.0-inf-py311))
23+
- The Dockerfile for CPU can be found
24+
[here](https://github.com/aws/deep-learning-containers/blob/master/pytorch/inference/docker/2.4/py3/Dockerfile.cpu), and the Dockerfile for GPU can
25+
be found [here](https://github.com/aws/deep-learning-containers/blob/master/pytorch/inference/docker/2.4/py3/cu124/Dockerfile.gpu).
26+
27+
## Security Advisory
28+
29+
AWS recommends that customers monitor critical security updates in the [AWS Security Bulletin](https://aws.amazon.com/security/security-bulletins/).
30+
31+
## Python 3.11 Support
32+
33+
Python 3.11 is supported in the PyTorch Inference containers.
34+
35+
## CPU Instance Type Support
36+
37+
The containers support x86_64 CPU instance types.
38+
39+
## GPU Instance Type Support
40+
41+
The containers support GPU instance types and contain the following software components for GPU support:
42+
43+
- CUDA 12.4.1
44+
- cuDNN 9.1.0.70+cuda12.4
45+
- NCCL 2.22.3+cuda12.4
46+
47+
## Build and Test
48+
49+
- Built on: c5.18xlarge
50+
- Tested on: c5.18xlarge, g3.16xlarge, m5.16xlarge, t3.2xlarge, p3.16xlarge, p3dn.24xlarge, p4d.24xlarge, g4dn.xlarge
51+
- Tested with [MNIST](http://yann.lecun.com/exdb/mnist/) and Resnet50/ImageNet datasets on EC2, ECS AMI (Amazon Linux AMI 2.0.20221102), and EKS AMI
52+
(amazon-eks-gpu-node-1.25.16-20240307)

0 commit comments

Comments
 (0)