Skip to content

README.md

Latest commit

 

History

History
167 lines (120 loc) · 3.8 KB

README.md

File metadata and controls

167 lines (120 loc) · 3.8 KB
Raw

InfiniRT

InfiniRT is the runtime library of the InfiniCore project. It provides the public runtime entry for supported accelerator backends, starting with device management and memory operations and growing with higher-level runtime features.

The recommended public entry is:

#include <infini/rt.h>

Full user documentation is available in docs/README.md.

Build and Install

Configure, build, and install InfiniRT with CMake:

cmake -S . -B build -DCMAKE_INSTALL_PREFIX=/path/to/infini-rt-prefix
cmake --build build -j
cmake --install build

infini-rt-prefix means the installation prefix used by CMake. After installation, headers and libraries are placed under this prefix, for example:

/path/to/infini-rt-prefix/include
/path/to/infini-rt-prefix/lib

Build Options

Common options include:

-DWITH_CPU=ON
-DWITH_NVIDIA=ON
-DINFINI_RT_BUILD_TESTING=ON

Example CPU build:

cmake -S . -B build \
  -DCMAKE_INSTALL_PREFIX=/path/to/infini-rt-prefix \
  -DWITH_CPU=ON \
  -DINFINI_RT_BUILD_TESTING=ON
cmake --build build -j
ctest --test-dir build --output-on-failure
cmake --install build

Example NVIDIA build:

cmake -S . -B build \
  -DCMAKE_INSTALL_PREFIX=/path/to/infini-rt-prefix \
  -DWITH_NVIDIA=ON \
  -DINFINI_RT_BUILD_TESTING=ON
cmake --build build -j
ctest --test-dir build --output-on-failure
cmake --install build

Basic Usage

#include <cstddef>

#include <infini/rt.h>

int main() {
  infini::rt::runtime::SetDevice(0);

  constexpr std::size_t size = 1024;
  void* ptr = nullptr;

  infini::rt::runtime::Malloc(&ptr, size);
  infini::rt::runtime::Memset(ptr, 0, size);
  infini::rt::runtime::Free(ptr);

  return 0;
}

The CUDA Runtime API-aligned layer lives under infini::rt::runtime. The top-level infini::rt::set_runtime_device_type and infini::rt::runtime_device_type APIs select which enabled backend receives those runtime calls. A GPU backend is selected initially when one is enabled; otherwise CPU is selected.

constexpr std::size_t size = 1024;
void* ptr = nullptr;

infini::rt::set_runtime_device_type(infini::rt::Device::Type::kCpu);
infini::rt::runtime::Malloc(&ptr, size);
infini::rt::runtime::Free(ptr);

infini::rt::set_runtime_device_type(infini::rt::Device::Type::kNvidia);
infini::rt::runtime::Malloc(&ptr, size);
infini::rt::runtime::Free(ptr);

Use infini::rt::runtime::Runtime<infini::rt::Device::Type::kCpu> when CPU runtime calls are needed explicitly in a build that also enables an accelerator backend.

Using Installed InfiniRT From Another Project

Downstream projects should consume the installed CMake package instead of depending on the InfiniRT source tree:

find_package(InfiniRT 0.1 CONFIG REQUIRED)

add_executable(app main.cc)
target_link_libraries(app PRIVATE InfiniRT::infinirt)

Point CMake at the installation prefix when configuring the consumer:

cmake -S . -B build \
  -DCMAKE_PREFIX_PATH=/path/to/infini-rt-prefix
cmake --build build -j

The imported target provides the installed include directory, library, C++17 requirement, and required backend link dependencies. InfiniRT_ENABLED_BACKENDS and the InfiniRT_WITH_<BACKEND> variables report the backends compiled into the installed library.

See examples/consumer_cmake for a minimal CMake consumer project.

Tests

Enable C++ tests with:

-DINFINI_RT_BUILD_TESTING=ON

Then run:

ctest --test-dir build --output-on-failure

Formatting

C++ code should be formatted with clang-format.

Python scripts should be checked and formatted with Ruff:

ruff format --check .
ruff check .

Contributing

Please follow CONTRIBUTING.md for code style, commit conventions, PR workflow, development commands, and troubleshooting.