Memory and Performance Optimisation on the IPU
- 1. Overview
- 2. Understanding the IPU programming model
- 3. Mapping a model to an IPU system
- 4. Optimising for performance
- 5. Common memory optimisations
- 6. Debugging an out-of-memory exception
- 7. Scaling an application over multiple replicas
- 7.1. Quick guide to scaling
- 7.2. Analyse your scaling behaviour
- 7.3. Constant or slowed-down processes (Amdahl’s law)
- 7.4. Graph compilation and executable loading
- 7.5. Host-I/O optimisation
- 7.6. Batch size and gradient accumulation count
- 7.7. Memory optimization for more replicas
- 7.8. Pipeline optimization and replicated tensor sharding
- 7.9. Technical background
- 8. Trademarks & copyright