Build Instructions¶

1. Prerequisites¶

Required: - CMake 3.16+ - C11/C++20 compiler (MSVC 19.44+, GCC 11+, Clang 14+) - Python 3.9+ (for bindings)

Optional: - CUDA 12.0+ (GPU optimizer acceleration, SNEPPX_BUILD_CUDA=ON) - Doxygen (API docs)

2. Quick Build (All Targets)¶

This project is Ninja-based. Use the CMake presets (see CMakePresets.json), which select the Ninja generator and write to per-config binary directories.

Windows (MSVC):

git clone https://github.com/ammar49-cyber/sneppx-alg.git
cd sneppx-alg
cmake --preset release
cmake --build --preset release

Linux/Mac (GCC/Clang):

git clone https://github.com/ammar49-cyber/sneppx-alg.git
cd sneppx-alg
cmake --preset release
cmake --build --preset release -j$(nproc)

The release preset configures Release + LTO + tests; debug, asan, cuda, python, and minimal presets are also available.

3. Build Individual Targets¶

Build into the active binary directory (e.g. build/release after the release preset), then reference targets:

# Core kernel library
cmake --build build/release --config Release --target neural_core_kernel

# Architecture algorithms
cmake --build build/release --config Release --target neural_architecture_layer

# Security C library
cmake --build build/release --config Release --target neural_security_c

# Security C++ obfuscation
cmake --build build/release --config Release --target neural_security_cpp

# Python bindings
cmake --build build/release --config Release --target _SNEPPX_c

# CUDA kernels (requires CUDA SDK; use the `cuda` preset)
cmake --preset cuda
cmake --build --preset cuda --target neural_cuda_kernels

4. Build Options¶

CMake Variable	Default	Description
`SNEPPX_BUILD_CUDA`	OFF	Enable CUDA kernel compilation
`SNEPPX_BUILD_PYTHON`	OFF	Build Python bindings
`SNEPPX_BUILD_TESTS`	ON	Build test executables
`SNEPPX_BUILD_EXAMPLES`	ON	Build example programs
`SNEPPX_BUILD_VULKAN`	OFF	Vulkan reference backend
`SNEPPX_BUILD_TPU`	OFF	TPU reference backend
`SNEPPX_BUILD_HTTP`	OFF	HTTP transport backend
`SNEPPX_BUILD_ZK`	OFF	Zero-knowledge backend
`SNEPPX_ENABLE_ASAN`	OFF	AddressSanitizer
`SNEPPX_ENABLE_UBSAN`	OFF	UndefinedBehaviorSanitizer

5. Debug Build¶

Use the debug preset (Ninja, Debug build type, tests+tools on):

cmake --preset debug
cmake --build --preset debug

Debug builds disable optimization and enable assertions. Use -DSNEPPX_S1_SECURE_MEMORY=OFF for faster debug iteration.

6. Python Package Build¶

# Editable install
pip install -e .

# With CUDA support
pip install -e .[cuda]

# Build wheel
pip install build
python -m build --wheel

7. Running Tests¶

Run from the binary directory of the active preset (e.g. build/release):

# C tests
cd build\release
ctest -C Release --output-on-failure

# Specific test
ctest -C Release -R test_kyber

# Sanitizer tests (Windows)
powershell -File scripts/run_sanitizers.ps1

For Python-only tests (no build required):

$env:PYTHONPATH = "bindings/python"
python -m pytest tests/python/ -v

8. Common Issues¶

Issue	Fix
`cuda_runtime.h: No such file or directory`	Build without CUDA: `-DSNEPPX_BUILD_CUDA=OFF`
`pybind11_add_module: Unknown CMake command`	Set `pybind11_DIR` or ensure FetchContent can reach GitHub
`LNK2001: unresolved external symbol`	Build `neural_core_kernel` and `neural_security_c` before `_SNEPPX_c`
MSVC "no VLAs"	Use `calloc`/`free` instead of stack-allocated VLAs
`__int128 undeclared` on MSVC	Guard with `#ifndef NO_UINT128` (see `x25519.c` pattern)
`CreateProcess failed: The system cannot find the file specified.` (MASM, Windows+Ninja)	Use the stable Ninja 1.12+ (see `scripts/` or repo root `ninja.exe`); the Kitware dev Ninja in the venv can crash. The x64 MASM path (`ml64`) is resolved automatically — no manual `-DCMAKE_ASM_MASM_COMPILER` needed.

9. Build Directories¶

The project standardizes on CMake presets (CMakePresets.json), each using the Ninja generator and writing to a per-config binary directory. Prefer the presets:

cmake --preset release → build/release
cmake --preset debug → build/debug
cmake --preset asan → build/asan
cmake --preset cuda → build/cuda
cmake --preset python → build/python
cmake --preset minimal → build/min

Avoid configuring the plain build/ directory with the presets; it is reserved for the legacy Visual Studio generator and is not Ninja-clean. See DEVELOPMENT.md for the full development workflow.