BrilLayoutOptimizer

Cache-aware optimizations for Bril, enhancing performance with loop transformations and memory access analysis.

Usage

Run the optimizer on a Bril JSON file:

python main.py input.json -o output.json --cache-size  -v

Choosing Cache Size

Specify the cache size based on your system:

• L1 Cache: Use for smaller data sets (e.g., --cache-size 32768 for 32 KB)
• L2 Cache: For larger data sets (e.g., --cache-size 262144 for 256 KB)
• L3 Cache: For very large data or server workloads (e.g., --cache-size 8388608 for 8 MB)

Check cache sizes:

# Linux
lscpu | grep 'cache'

# macOS
sysctl -a | grep 'cachesize'

Running the Optimizer

# Run with typical L1 cache size
python main.py input.json -o output.json --cache-size 32768 -v

# Verify optimizations
python optimization_checker.py

Testing

Run the test suite:

pytest tests/