AI021
CUDA Programming Guide
Optimization, Graphs, and Hardware Accelerators
Learning Objectives
- Identify computational bottlenecks in traditional CPU/GPU architectures for irregular graph data.
- Evaluate the design of domain-specific accelerators like Graphcore IPU and TPU for optimization tasks.
- Analyze spatial mapping and tiling strategies for large-scale graph optimization algorithms.
- Compare performance metrics of hardware-aware graph processing frameworks.