1. Scope of this document
This document contains the release notes for the Poplar SDK 3.3.0 for Graphcore’s IPU product family. The following software deliverables are covered by this document:
The PopTorch library provides a set of extensions for PyTorch to enable it to run on the Graphcore IPU hardware.
- PopTorch Geometric
The PopTorch Geometric library provides a set of extensions for PyTorch Geometric, enabling Graph Neural Network models to be trained, evaluated and used on Graphcore IPU hardware.
An implementation of the TensorFlow framework for the Graphcore IPU.
An implementation of the Keras library for the Graphcore IPU.
- IPU TensorFlow Addons
A collection of Graphcore IPU-specific features for the TensorFlow framework.
- TensorFlow Serving
An implementation of the TensorFlow Serving application for the Graphcore IPU.
PopXL is an experimental PopART Python package that provides greater flexibility for model construction than is possible using the standard PopART API.
The Poplar Advanced Run Time is a flexible ONNX-compatible runtime supporting both training and inference.
- Poplar Libraries
The PopLibs library provides a range of higher-level functions commonly used in machine learning applications.
A graph programming framework for the IPU.
The Graphcore Communication Library enables high-performance scale-out for IPU systems.
Poplar distributed configuration library (PopDist) is a library for configuring and coordinating distributed execution of (large-scale) machine learning applications.
PopRun is a command line utility to launch distributed applications on Graphcore Pod systems.
The PopVision analysis library (libpva) allows programmatic analysis of the IPU profiling information used by the PopVision Graph Analyser.
The PopVision trace instrumentation library (libpvti) provides functions to control the capture of profiling information for the host-code of your IPU application which can be explored with the PopVision System Analyser or using libpva.
The Poplar Exchange Format (PopEF) is a file format for exporting and importing models to run on the IPU, and a library for managing those files.
- Model Runtime
A library built on the Poplar runtime to enable loading and running models stored in the Poplar Exchange Format (PopEF) on the IPU.
- Poplar Triton Backend
A backend for the Triton Inference Server that supports models exported to PopEF files.
- Driver and utilities
Driver and associated utilities needed by the Graphcore IPU.