Neuron performance

The Neuron performance pages provide comprehensive benchmarks and performance data for AWS Neuron SDK across different Trainium and Inferentia instance types. These benchmarks cover various open-source models for Natural Language Processing (NLP), Computer Vision (CV), and Recommender systems. Each benchmark includes detailed setup instructions and reproducible test configurations to help you evaluate performance for your specific use cases.

Inference performance

Inf1 Inference Performance

Comprehensive inference benchmarks for Inf1 instances across NLP, CV, and recommender models

Inf2 Inference Performance

Latest inference performance data for Inf2 instances with improved throughput and latency metrics

Trn1 Inference Performance

Inference benchmarks for Trn1 instances showcasing versatile training and inference capabilities

Training performance

Trn1 Training Performance

Training performance benchmarks for Trn1 instances with distributed training metrics and scalability data