CMakeLists.txt

cmake_minimum_required(VERSION 3.20)
project(
  TensorCraft-HPC
  VERSION 2.0.0
  DESCRIPTION "High-Performance AI Kernels with Modern C++ & CUDA"
  LANGUAGES CXX)

# ============================================================================
# C++ Standard Configuration
# ============================================================================
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(CMAKE_CXX_EXTENSIONS OFF)

# Detect C++20/23 support
include(CheckCXXCompilerFlag)
check_cxx_compiler_flag("-std=c++20" COMPILER_SUPPORTS_CXX20)
check_cxx_compiler_flag("-std=c++23" COMPILER_SUPPORTS_CXX23)

if(COMPILER_SUPPORTS_CXX23)
  set(CMAKE_CXX_STANDARD 23)
  message(STATUS "C++23 support enabled")
elseif(COMPILER_SUPPORTS_CXX20)
  set(CMAKE_CXX_STANDARD 20)
  message(STATUS "C++20 support enabled")
else()
  message(STATUS "C++17 support enabled")
endif()

# ============================================================================
# CUDA Configuration
# ============================================================================
option(TC_ENABLE_CUDA "Enable CUDA support" ON)
set(TC_CUDA_ENABLED OFF)

if(TC_ENABLE_CUDA)
  if(NOT DEFINED CMAKE_CUDA_ARCHITECTURES OR CMAKE_CUDA_ARCHITECTURES STREQUAL
                                             "")
    set(TC_DEFAULT_CUDA_ARCHITECTURES "75;80;86;89;90")
    set(CMAKE_CUDA_ARCHITECTURES
        "${TC_DEFAULT_CUDA_ARCHITECTURES}"
        CACHE STRING "CUDA architectures")
  endif()

  include(CheckLanguage)
  check_language(CUDA)
  if(CMAKE_CUDA_COMPILER)
    set(TC_CUDA_ENABLED ON)
    enable_language(CUDA)
    find_package(CUDAToolkit REQUIRED)

    # Check CUDA version (12.0+ required for FP8 and modern features)
    if(CUDAToolkit_VERSION VERSION_LESS "12.0")
      message(
        WARNING
          "CUDA 12.0 or higher is recommended. Found: ${CUDAToolkit_VERSION}. Some features may not be available."
      )
    endif()

    message(STATUS "CUDA compiler: ${CMAKE_CUDA_COMPILER}")
    message(STATUS "CUDA version: ${CUDAToolkit_VERSION}")

    set(CMAKE_CUDA_STANDARD 17)
    set(CMAKE_CUDA_STANDARD_REQUIRED ON)

    # Feature detection based on CUDA version
    if(CUDAToolkit_VERSION VERSION_GREATER_EQUAL "12.0")
      add_compile_definitions(TC_CUDA_12=1 TC_CUDA_11=1 TC_CUDA_10=1)
      add_compile_definitions(TC_HAS_TMA=1 TC_HAS_WGMMA=1 TC_HAS_FP8=1
                              TC_HAS_BF16=1 TC_HAS_WMMA=1)
      message(STATUS "CUDA 12.x features enabled (TMA, WGMMA, FP8, BF16, WMMA)")
    elseif(CUDAToolkit_VERSION VERSION_GREATER_EQUAL "11.0")
      add_compile_definitions(TC_CUDA_11=1 TC_CUDA_10=1)
      add_compile_definitions(TC_HAS_WMMA=1 TC_HAS_BF16=1)
      message(STATUS "CUDA 11.x features enabled (WMMA, BF16)")
    else()
      add_compile_definitions(TC_CUDA_10=1)
      message(STATUS "CUDA 10.x features enabled")
    endif()
  else()
    message(
      WARNING
        "Failed to find a working CUDA compiler. Configure will continue without CUDA. To enable CUDA, set CUDAToolkit_ROOT or ensure nvcc is on PATH."
    )
  endif()
endif()

# ============================================================================
# Build Options
# ============================================================================
option(TC_BUILD_TESTS "Build unit tests" ON)
option(TC_BUILD_BENCHMARKS "Build benchmarks" ON)
option(TC_BUILD_PYTHON "Build Python bindings" ON)
option(TC_BUILD_DOCS "Build documentation" OFF)

if(NOT TC_CUDA_ENABLED)
  if(TC_BUILD_TESTS
     OR TC_BUILD_BENCHMARKS
     OR TC_BUILD_PYTHON)
    message(
      WARNING
        "CUDA is unavailable; disabling tests, benchmarks, and Python bindings."
    )
  endif()
  set(TC_BUILD_TESTS
      OFF
      CACHE BOOL "Build unit tests" FORCE)
  set(TC_BUILD_BENCHMARKS
      OFF
      CACHE BOOL "Build benchmarks" FORCE)
  set(TC_BUILD_PYTHON
      OFF
      CACHE BOOL "Build Python bindings" FORCE)
endif()

# ============================================================================
# Dependencies via FetchContent
# ============================================================================
include(FetchContent)

if(TC_BUILD_TESTS)
  FetchContent_Declare(
    googletest
    GIT_REPOSITORY https://github.com/google/googletest.git
    GIT_TAG v1.14.0)
  set(gtest_force_shared_crt
      ON
      CACHE BOOL "" FORCE)
  FetchContent_MakeAvailable(googletest)
endif()

if(TC_BUILD_BENCHMARKS)
  FetchContent_Declare(
    benchmark
    GIT_REPOSITORY https://github.com/google/benchmark.git
    GIT_TAG v1.8.3)
  set(BENCHMARK_ENABLE_TESTING
      OFF
      CACHE BOOL "" FORCE)
  FetchContent_MakeAvailable(benchmark)
endif()

if(TC_BUILD_PYTHON)
  FetchContent_Declare(
    pybind11
    GIT_REPOSITORY https://github.com/pybind/pybind11.git
    GIT_TAG v2.11.1)
  FetchContent_MakeAvailable(pybind11)
endif()

# ============================================================================
# Main Library Target (Header-Only)
# ============================================================================
add_library(tensorcraft INTERFACE)
target_include_directories(
  tensorcraft INTERFACE $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
                        $<INSTALL_INTERFACE:include>)
if(TC_CUDA_ENABLED)
  target_link_libraries(tensorcraft INTERFACE CUDA::cudart)
endif()

# Add alias for consistent naming
add_library(TensorCraft::tensorcraft ALIAS tensorcraft)

# ============================================================================
# Subdirectories
# ============================================================================
if(TC_BUILD_TESTS)
  enable_testing()
  add_subdirectory(tests)
endif()

if(TC_BUILD_BENCHMARKS)
  add_subdirectory(benchmarks)
endif()

if(TC_BUILD_PYTHON)
  add_subdirectory(src/python_ops)
endif()

# ============================================================================
# Installation
# ============================================================================
include(GNUInstallDirs)

install(
  TARGETS tensorcraft
  EXPORT TensorCraftTargets
  INCLUDES
  DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})

install(DIRECTORY include/ DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})

install(
  EXPORT TensorCraftTargets
  FILE TensorCraftTargets.cmake
  NAMESPACE TensorCraft::
  DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/TensorCraft)

# ============================================================================
# Summary
# ============================================================================
message(STATUS "")
message(STATUS "TensorCraft-HPC Configuration Summary:")
message(STATUS "  Version:        ${PROJECT_VERSION}")
message(STATUS "  C++ Standard:   ${CMAKE_CXX_STANDARD}")
if(TC_CUDA_ENABLED)
  set(TC_CUDA_VERSION_STR "${CUDAToolkit_VERSION}")
  set(TC_CUDA_ARCHS_STR "${CMAKE_CUDA_ARCHITECTURES}")
else()
  set(TC_CUDA_VERSION_STR "DISABLED")
  set(TC_CUDA_ARCHS_STR "N/A")
endif()
message(STATUS "  CUDA Enabled:   ${TC_CUDA_ENABLED}")
message(STATUS "  CUDA Version:   ${TC_CUDA_VERSION_STR}")
message(STATUS "  CUDA Archs:     ${TC_CUDA_ARCHS_STR}")
message(STATUS "  Build Tests:    ${TC_BUILD_TESTS}")
message(STATUS "  Build Bench:    ${TC_BUILD_BENCHMARKS}")
message(STATUS "  Build Python:   ${TC_BUILD_PYTHON}")
message(STATUS "")