Level 2: JIT Compilation

Level 2 provides JIT compilation for CPU-bound loops and numerical code.

Overview

JIT Backends

Enabling Level 2

import epochly
@epochly.optimize(level=2)
def compute(data):
    result = 0
    for x in data:
        result += x ** 2
    return result

Supported Code Patterns

• Numerical Loops: Simple arithmetic in loops
• Array Operations: Element-wise array processing
• Nested Loops: Multi-dimensional iteration

Speedup Factors

Best Practices

1. Use for Large Data: GPU shines on arrays with 10M+ elements
2. Use for Heavy Tasks: Parallel execution works best with tasks >100ms each
3. Don't Force It: Epochly yields to normal behavior when it can't help