Neuron Custom C++ Operators (archived)#

Warning

This feature overview is archived and is no longer maintained. Neuron Custom C++ operators apply only to Inf2 and Trn1. For custom kernels on Trn2 and Trn3, see the Neuron Kernel Interface (NKI). For the existing custom operators documentation, see Neuron Custom C++ Operators [Beta] (archived).

Neuron Custom C++ Operators enable developers to write C++ Custom Operators (“CustomOps”) that run on NeuronCores. This enables developers to extend operator support beyond what is officially supported by Neuron.

Developers can use standard PyTorch custom operators programming interfaces to leverage Neuron Custom C++ Operators feature. This makes it easy to migrate CPU Custom Operators to Neuron, and implement new beta operators, all without any intimate knowledge of the NeuronCore hardware.

For more details see Neuron Custom C++ Operators [Beta] (archived)

This document is relevant for: Inf2, Trn1