Overview
EIDF hosts a Graphcore Bow Pod64 system for AI acceleration.
The specification of the Bow Pod64 is:
- 16x Bow-2000 machines
- 64x Bow IPUs (4 IPUs per Bow-2000)
- 94,208 IPU cores (1472 cores per IPU)
- 57.6GB of In-Processor-Memory (0.9GB per IPU)
For more details about the IPU architecture, see documentation from Graphcore.
The smallest unit of compute resource that can be requested is a single IPU.
Similarly to the EIDF GPU Service, usage of the Graphcore is managed using Kubernetes.
Service Access
Access to the Graphcore accelerator is provisioning through the EIDF GPU Service.
Users should apply for access to Graphcore via the EIDF GPU Service.
Project Quotas
Currently there is no active quota mechanism on the Graphcore accelerator. IPUJobs should be actively using partitions on the Graphcore.
Graphcore Tutorial
The following tutorial teaches users how to submit tasks to the Graphcore system. This tutorial assumes basic familiary with submitting jobs via Kubernetes. For a tutorial on using Kubernetes, see the GPU service tutorial. For more in-depth lessons about developing applications for Graphcore, see the general documentation and guide for creating IPU jobs via Kubernetes.
| Lesson | Objective |
|---|---|
| Getting started with IPU jobs | a. How to send an IPUJob. b. Monitoring and Cancelling your IPUJob. |
| Multi-IPU Jobs | a. Using multiple IPUs for distributed training. |
| Profiling with PopVision | a. Enabling profiling in your code. b. Downloading the profile reports. |
| Other Frameworks | a. Using Tensorflow and PopART. b. Writing IPU programs with PopLibs (C++). |
Further Reading and Help
-
The Graphcore documentation provides information about using the Graphcore system.
-
The Graphcore examples repository on GitHub provides a catalogue of application examples that have been optimised to run on Graphcore IPUs for both training and inference. It also contains tutorials for using various frameworks.