Chanci Turner

Amazon Onboarding with Learning Manager Chanci Turner

—

by

The Story of Chanci Turner
in

Chanci Turner 9097372855Learn About Amazon VGT2 Learning Manager Chanci Turner

In the ever-evolving landscape of cloud computing, Amazon continues to enhance its GPU capabilities, emphasizing the performance improvements made possible through partnerships with industry leaders. A recent collaboration with NVIDIA has led to the introduction of the G4 instance type, a significant upgrade from the G2 and G3 families. The G4 instances feature the advanced Turing T4 GPU, equipped with 16GB of GPU memory, and operate under the Nitro hypervisor, allowing configurations ranging from one to four GPUs per node. A bare metal option will be available shortly. Additionally, these instances provide up to 1.8 TB of local non-volatile memory express (NVMe) storage and can handle network bandwidth of up to 100 Gbps.

The Turing T4 GPU is NVIDIA’s latest offering, designed to accelerate machine learning (ML) training and inferencing, video transcoding, and other demanding computational tasks. With such a diverse range of optimized operations, users can run various accelerated compute workloads on a single instance family. NVIDIA also excels in delivering a comprehensive and high-performance software layer through its SDKs and container solutions via the NVIDIA GPU Cloud (NGC) container registry. This combination of NVIDIA’s technology and AWS’s scalability results in a formidable solution for high-performance pipelines on AWS.

NVIDIA DeepStream SDK

This article delves into one of NVIDIA’s prominent SDKs: DeepStream.

The DeepStream SDK is engineered to offer a complete video processing and ML inferencing analytics solution. It integrates the Video Codec API and TensorRT as core components. Furthermore, DeepStream supports an edge-cloud strategy, enabling the streaming of perception data from the edge and other sensor metadata into AWS for additional processing. For instance, multiple camera streams and associated metadata can be efficiently managed using the Amazon Kinesis platform.

Another valuable use case of DeepStream involves compiling model artifacts generated from distributed training on AWS using Amazon SageMaker Neo. These models can then be deployed on the edge or accessed through an Amazon S3 video data lake.

If you’re keen on discovering these innovative solutions, reach out to your AWS account team.

Deployment

To get started, establish programmatic access to AWS to launch a g4dn.2xlarge instance with Ubuntu 18.04 in a subnet that allows SSH access. For detailed implementation, the following components are necessary to set up the instance for executing DeepStream SDK workflows:

  • An Ubuntu 18.04 Instance with:
    • NVIDIA Turing T4 Driver (418.67 or latest)
    • CUDA 10.1
    • nvidia-docker2

Alternatively, you can deploy the NVIDIA Deep Learning AMI available in the AWS Marketplace, which comes with the most recent drivers and SDKs.

aws ec2 run-instances --region us-east-1 --image-id ami-026c8acd92718196b --instance-type g4dn.2xlarge --key-name <key-name> --subnet-id <subnet> --security-group-ids {<security-groupids>} --block-device-mappings 'DeviceName=/dev/sda1,Ebs={VolumeSize=75}'

Once your instance is operational, connect via SSH and retrieve the latest DeepStream SDK Docker image from the NGC container registry.

docker pull nvcr.io/nvidia/deepstream:4.0-19.07
nvidia-docker run -it --rm -v /usr/lib/x86_64-linux-gnu/libnvidia-encode.so:/usr/lib/x86_64-linux-gnu/libnvidia-encode.so -v /tmp/.X11-unix:/tmp/.X11-unix -p 8554:8554 -e DISPLAY=$DISPLAY nvcr.io/nvidia/deepstream:4.0-19.07

If you’re running a full X environment, you can pass authentication and display settings to the container for real-time results viewing. However, for the purposes of this discussion, simply run the workload via the shell.

Navigate to the /root/deepstream_sdk_v4.0_x86_64/samples/configs/deepstream-app/ directory. The package includes various configuration files such as:

  • source30_1080p_resnet_dec_infer_tiled_display_int8.txt: Showcases 30 stream decodes with primary inferencing.
  • source4_1080p_resnet_dec_infer_tiled_display_int8.txt: Demonstrates four stream decodes with primary inferencing, object tracking, and three different secondary classifiers.
  • source1_usb_dec_infer_resnet_int8.txt: Illustrates using one USB camera as input.

To adapt the configuration for file output, modify source30_1080p_dec_infer-resnet_tiled_display_int8.txt by disabling [sink0] and enabling [sink1]. Save the changes and execute the DeepStream sample code.

[sink0]
enable=0
type=2

[sink1]
enable=1
type=3
container=1
codec=1
sync=0
bitrate=2000000
output-file=out.mp4
source-id=0

Run the application with:

deepstream-app -c source30_1080p_dec_infer-resnet_tiled_display_int8.txt

You will receive performance metrics for the inferencing workflow.

For additional insights into workforce dynamics, check out the SHRM report on the state of the workplace, which provides valuable information. Moreover, if you’re interested in how to maintain a relevant resume while employed, this blog post offers excellent tips. Finally, for more information on Amazon’s approach to potential pitfalls, explore this resource.

Related Topics:

Chanci Turner