Uncompiled Execution

Uncompiled Execution

Web: uncompiled.html

Grok16 3.0 — dev runs at normal speed; compile is chamber-organized and ahead, not on every click.

Doctrine

data/field-exec-uncompiled-doctrine.json

Lane	Dev behavior	Speed (speed_demo @ 3s)
Python	True interpreter (pythong / CPython / gpy-16)	~0.72–0.76M ops/s
C / C++	No line-by-line interpreter — chamber compile ahead	~80–88M ops/s after plane cache
CMake	Configure + build once; bin reused	~85.7M ops/s

Queen .launch — organized field (default)

1. Inspect files — walk chamber, map depth-0 fields, trim excess
2. No compile on launch — interpreter for Python; BSP reuse for native
3. Staged plane — copy from data/bench/exec-plane/ when sources match (~0 ms)
4. Fallback — Python entry when native plane not staged
5. Compile mode — explicit only (QUEEN_LAUNCH_COMPILE=1 or Queen Files Compile)

Module: Queen/lib/queen-launch-chamber.py

Not compiling C/C++

There is no C/C++ REPL. Chamber organization presents a bin-like run:

• Reuse staged binary from data/bench/exec-plane/ (~0.2 ms copy)
• Or compile once into plane cache (434 ms – 2.7 s first run)
• Operator uses Compile mode only for release-day wave convert

Commands

# Organized field — inspect + run, no compile (default)
python3 Queen/lib/queen-launch-chamber.py run examples/speed-demo/speed-demo.launch

# Field map only
python3 Queen/lib/queen-launch-chamber.py project examples/speed-demo/speed-demo.launch

# Compile mode — singular plane wave-convert
QUEEN_LAUNCH_COMPILE=1 python3 Queen/lib/queen-launch-chamber.py run examples/speed-demo/speed-demo.launch

See Speed Bench for versioned compile + exec numbers.