Performance Guide

RayQuiro provides three distinct execution tiers to give you the ideal balance between instant development turnaround and production-grade execution speed.

---

1. Execution Tiers Compared

Execution Mode	Command	Startup Latency	Execution Speed	Use Case
Native Binary (LLVM)	rqio build --release	Zero (direct native binary)	Maximum (C/Rust speed)	Production releases, games, desktop apps, servers
Bytecode VM	rqio run script.rq	Instant (~5ms)	Fast (JIT-like speed)	Daily development, rapid testing, scripts
Tree-Walk Interpreter	rqio run --legacy	Instant (~2ms)	Baseline	Debugging, embedded systems, microcontrollers

---

2. LLVM Native Compilation

When you run:

rqio build main.rq --release -o app.exe

RayQuiro runs full LLVM compilation with -O3:

• Constant folding & propagation
• Inlining of small functions
• Dead code elimination (DCE)
• Loop unrolling and auto-vectorization
• Memory-to-register promotion (mem2reg)

The resulting binary executes directly on CPU registers with no virtual machine dispatch loop.

---

3. High-Performance Coding Tips

1. Cache Collection Lengths in Loops

In tight loops, cache len(arr) in a local variable:

var n = len(items);
var i = 0;
while (i < n) {
    process(items[i]);
    i = i + 1;
}

2. Pre-allocate or Reuse Arrays

Avoid repeatedly creating temporary arrays inside hot inner loops.

3. Use Native Built-ins for Bulk Operations

Built-ins like crypto.*, hash.*, regex.*, and math.* execute in compiled C/C++ without VM overhead.

4. Benchmark with time.now_ms()

Always measure using monotonic timers:

var t0 = time.now_ms();
compute_heavy_algorithm();
var delta = time.now_ms() - t0;
print("Computation time: " + str(delta) + " ms");