AWS First Steps
LanguageEnglish
New feature

Amazon SageMaker AI now supports instance preference lists for training and processing jobs

Amazon SageMaker AI introduces instance preference lists for training and processing jobs, allowing users to specify prioritized instance types so SageMaker automatically runs jobs on the first available configuration, reducing wait times during high-demand periods.

Today, Amazon SageMaker AI introduced instance priority lists for training and processing jobs, simplifying how customers find compute capacity for their workloads. Previously, users had to specify a single instance type, leading to unpredictable wait times during peak gpu demand and requiring complex retry logic. Now, by providing a priority list of acceptable instance types, SageMaker runs jobs on the first available configuration, potentially accelerating job start times. Users specify instance type and count priorities when submitting jobs, such as prioritizing two ml.g6.48xlarge instances or four ml.g5.48xlarge instances. SageMaker processes the list and starts jobs with the first available capacity. Capacity sources can also be configured within the same job submission, using either on-demand or reserved SageMaker flexible training plans. This feature simplifies compute acquisition during high-demand periods and reduces manual retries, available today across all AWS regions via the SageMaker CLI, API, SDK, and console UI. For more details, see the documentation or launch blog.

Why it matters

This update benefits users performing AI model training or data processing by enabling more efficient allocation of compute resources required for these tasks.

Read the original AWS announcement

Related articles and terms

GPU-Based Machine Learning Training with AWS Batch - Cost-Efficient Large-Scale TrainingRun GPU training with your existing Docker containers, and cut costs by up to 90% using Spot Instances and checkpointing. Includes guidance on when to choose Batch over SageMaker.Amazon SageMakerAn integrated platform for building, training, and deploying machine learning modelsAWS BatchA fully managed service that automatically schedules hundreds of thousands of batch processing jobs and runs them on optimal computing resourcesBuilding an ML Platform with Amazon SageMaker - From Model Development to DeploymentFrom development in Studio to managed spot training, MLOps with Pipelines, and data drift detection with Model Monitor, this article covers how to integrate the entire ML lifecycle.Batch Computing Infrastructure - Large-Scale Parallel Processing with AWS BatchLearn how to build large-scale batch processing with AWS Batch. Covers job queue design, auto-scaling compute environments, cost optimization with Spot Instances, and building batch infrastructure ideal for scientific computing and large-scale data processing.