Application Notes

Performance

General

• Performance Tuning
• Mixed precision and performance-accuracy tuning
• Parallel Execution using NEURONCORE_GROUP_SIZES

PyTorch Neuron

• Running inference on variable input shapes with bucketing
• Data Parallel Inference on Torch Neuron

Apache MXNet Neuron

• Flexible Execution Group (FlexEG) in Neuron-MXNet

Models

• Inferentia Model Architecture Fit

Neuron Components

• Introducing Neuron Runtime 2.x (libnrt.so)