Amazon Elastic Container Service (ECS) expanded capability with support for fractional GPU scheduling on Amazon EC2 G6f instances. This allows users to allocate GPU resources in fractions (as small as one-eighth of an NVIDIA L4 Tensor Core GPU with 3 GB memory), optimizing resource usage for workloads like small-model AI inference and graphics rendering that don't require full GPUs. The feature can be configured directly in ECS task definitions and supports both ECS Managed Instances and ECS on EC2.
ECS Managed Instances automate provisioning, scaling, and health monitoring for GPU workloads, including failure detection and replacement of unhealthy instances. The fractional GPU feature is available across all AWS regions offering G6f instances, aiming to reduce infrastructure costs by tailoring GPU allocation to workload needs rather than over-provisioning full GPUs.
Source & further reading
Amazon ECS now supports fractional GPU scheduling with Amazon EC2 G6f instances ↗Source published 2026-08-06. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.