name: Benchmarking
number: 19
type: analysis
difficulty: intermediate
estimated_hours: 6-8

description: |
  Performance measurement and analysis. Students learn to scientifically benchmark
  ML systems, identify bottlenecks, and compare optimization techniques.

learning_objectives:
  - Build performance profiling tools
  - Measure memory and compute usage
  - Compare optimization techniques
  - Create reproducible benchmarks

prerequisites:
  - Module 15: Acceleration
  - Module 16: Caching
  - Module 17: Precision
  - Module 18: Compression

skills_developed:
  - Performance profiling
  - Bottleneck identification
  - Scientific measurement
  - Benchmark design

exports:
  - tinytorch.benchmarks