Posted On: Nov 30, 2021
Today, we are announcing the preview of AWS Trainium-based Amazon EC2 Trn1 instances. AWS Trainium, is the second machine learning chip built by AWS that is optimized for high-performance deep learning training.
Trn1 instances will deliver the best price performance for training deep learning models in the cloud for use cases such as natural language processing, object detection, image recognition, recommendation engines, intelligent search, and more. They support up to 16 Trainium accelerators, up to 800 Gbps of EFA networking throughput (double the networking bandwidth available in GPU-based instances), and ultra high speed intra-instance connectivity for the fastest ML training in Amazon EC2.
They are deployed in EC2 UltraClusters, which can be scaled to tens of thousands of Trainium accelerators with petabit scale, non-blocking networking. These Trn1 UltraClusters are 2.5x larger than previous generation EC2 UltraClusters and serve as a powerful supercomputer to rapidly train the most complex deep learning models.
If you are interested in Trn1 instances, you can sign up for preview by visiting our product detail page.