Learn more about the Graphcore Poplar® SDK and get started programming IPU systems.

Read Analyst Report on Poplar Support

Open & Extensible Poplar Libraries

Get access to 50+ optimised functions for common ML models and 750 high performance compute elements. Modify and write custom libraries. 

Straightforward Deployment

Pre-built Docker containers with Poplar SDK, Tools and frameworks images to get up and running fast.

ML Frameworks Support

Support for standard ML frameworks: TensorFlow 1 and 2, ONNX and PyTorch with PaddlePaddle coming soon. 

TensorFlow PyTorch ONNX PaddlePaddle

Standard Ecosystem Support

Ready for production with Microsoft Azure deployment, Kubernetes orchestration, Docker containers and Hyper-V virtualization & security.

Azure Docker Kubernetes Microsoft Hyper-v

Watch on-demand videos and demos.

Running PyTorch on IPU

ResNext computer vision model demo

Running TensorFlow on IPU

MCMC probability model demo

Bert-Base Training Click to Zoom
We regularly update our benchmark charts to reflect new optimisations and the latest performance features in our software. And we share public examples for all of our benchmarks on GitHub so you can run them for yourself during evaluation. Go to our benchmark page to see the results and go direct to our public examples repo. Latest benchmarks
Apr 15, 2020 Major new Graphcore Poplar® SDK 1.1. Released Read more
Apr 15, 2020 Graphcore launches new Poplar® Analysis Tool Read more
Mar 19, 2020 Graphcore makes Poplar SDK Docs publicly available Read more
Github Visit public repo
Graphcore Support Get support
Stack Overflow Ask a question
Zhihu Visit Zhihu