No CPython. No LLVM. No GCC. No runtime. No GIL.
The same .py file that takes 50ms with CPython executes in 0.305ms with PyDead-BIB — compiled directly to machine code, loaded into RAM via VirtualAlloc, and executed without touching disk.
What makes it different:
→ JIT Killer v2.0: not a JIT that learns at runtime — a compiler that already knows everything before execution. "The CPU doesn't think — it already knows."
→ UB Detection: 13+ Python undefined behaviors caught at compile time, not runtime. MutableDefaultArg, NoneDeref, TypeMismatch — before your program ever runs.
→ AVX2 SIMD auto-vectorization: list[float] × 8 → VMULPS ymm0 automatically. 8 floats per cycle.
→ GPU dispatch: CUDA + Vulkan via ctypes, RTX 3060 at 24,119 GFLOPS from compiled Python.
→ In-memory execution: pyb run compiles and executes in RAM — no .exe written to disk. The program exists for 0.305ms, then disappears.
Hello World: CPython 30MB runtime → PyDead-BIB 2KB 86/86 tests passing. Inherits ADead-BIB v8.0 IR — 354,134 lines of proven compiler infrastructure.
Built because nobody asked "why does Python need a runtime?" seriously enough to remove it.
Licensed under Techne v1.0 — free for personal, education, open source. 10% royalty for commercial use over $1M revenue.
GitHub: github.com/AndreeSalazar/PyDead-BIB Lima, Peru — Binary Is Binary