Amazon ECS now supports fractional GPU scheduling with Amazon EC2 G6f instances

This article announces that Amazon ECS now supports fractional GPU scheduling with Amazon EC2 G6f instances, enabling cost-effective GPU resource allocation for various workloads.

• Run workloads on GPU partitions as small as one-eighth of an NVIDIA L4 Tensor Core GPU with 3 GB memory
• Request fractional GPUs by setting GPU=0.125, GPU=0.25, or GPU=0.5 in ECS task definitions
• Supported on both ECS Managed Instances and ECS on EC2
• ECS Managed Instances provide automatic provisioning, scaling, patching, and lifecycle management
• Includes GPU metrics via CloudWatch Container Insights and automatic health monitoring for GPU failures
• Available in all AWS Regions where EC2 G6f instances are available

Fractional GPU scheduling enables right-sizing containers for AI inference, model experimentation, and graphics rendering while reducing infrastructure costs compared to full GPU instances.