Delivering video content with fractional GPUs in containers on Amazon EKS
Containers Blog
This article discusses how to build a video encoding pipeline on AWS that uses fractional GPUs in containers running on Amazon EKS. By splitting GPUs into fractions, multiple encoding jobs can share a single GPU concurrently, improving resource utilization and lowering costs.
Specifically, the article covers:
- Configuring GPU time-slicing in Amazon EKS using the NVIDIA device plugin
- A comparison of video encoding density on different GPU instance types (G4dn, G5, G5g)
- Horizontal node auto-scaling strategies for video encoding workloads, using tools like Karpenter and Bottlerocket
- Techniques for faster scaling, such as pre-caching container images and over-provisioning nodes
- Conclusion highlighting the benefits of the proposed architecture for efficient and scalable video encoding
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
Oct 28
2025
2025
Extending GPU Fractionalization and Orchestration to the edge with NVIDIA Run:ai and Amazon EKS
Jul 29
2025
2025
Announcing general availability of Amazon EC2 G6f instances with fractional GPUs
Sep 16
2025
2025
Amazon AppStream 2.0 adds support for fractional GPU instances
Nov 21
2025
2025
Amazon CloudWatch Container Insights adds Sub-Minute GPU Metrics for Amazon EKS
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.