This is a simple Python interface to lfk-mp-benchmark

Usage

import lfkbenchmark
benchmark = lfkbenchmark.Benchmark()
benchmark.console_run_benchmark()

INSTALL

Binary wheels

pip install lfkbenchmark 

Through Github Actions

Just Fork the repository on github and you should be able to get Github Actions to build your library as is. Just commit your changes and you should get a new build.

git commit -m "commit message [build wheel]"

From Source

Make sure to either compile the lfk-mp-benchmark library in the root of this repo or just pass the path to the library for compilation.

git --recurse-submodules clone `https://github.com/waverian/py-lfk-benchmark`
cd py_lfk_mp_benchmark

isolate your build env with a Virtual Environment

python3 venv venv-lfk
source venv-lfk/bin/acivate

Install Dependencies

pip install cython wheel auditwheel cmake

on macOS replace auditwheel with delocate

Compile lfk-mp-benchmark

Using script

cd path/to/py-lfk-mp-benchmark
sh tools/build_osx_wheel.sh

Manual instructions

Let's build lfk-mp-benchmark submodule.

mkdir -p lfk-mp-benchmark/build
cd lfk-mp-benchmark/build
CMAKE_OSX_ARCHITECTURES="arm64;x86_64" cmake .. -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=On -DBUILD_CONSOLE_APP=Off
cmake --build .

Check if compiled lib shows both architectures.

file lfk_benchmark/liblfk-benchmark.dylib

On linux your lib would be similarly named .so

Should give a output similar to::

lfk_benchmark/liblfk-benchmark.dylib: Mach-O universal binary with 2 architectures: [x86_64:Mach-O 64-bit dynamically linked shared library x86_64Mach-O 64-bit dynamically linked shared library x86_64] [arm64:Mach-O 64-bit dynamically linked shared library arm64Mach-O 64-bit dynamically linked shared library arm64]
lfk_benchmark/liblfk-benchmark.dylib (for architecture x86_64):	Mach-O 64-bit dynamically linked shared library x86_64
lfk_benchmark/liblfk-benchmark.dylib (for architecture arm64):	Mach-O 64-bit dynamically linked shared library arm64

On linux your lib probably would be built for your local architecture unless you are using a cross compile toolchain.

Let's build our sources for py-lfk-benchmark

python setup.py build_ext -t build/

Now Let's build our wheel.

pip wheel . -w build/wheels

This wheel by default does not include the liblfk-benchmark.so to include this we need delocate package.

DYLD_LIBRARY_PATH=lfk-mp-benchmark/build/lfk_benchmark delocate-wheel -w fixed_wheels -v build/wheels/path/to/your/wheel.universal2.whl

on linux simply replace DYLD_LIBRARY_PATH with LD_LIBRARY_PATH.

The wheel in dir fixed_wheels should now be a univeral wheel that should be able dynamically load the benchmark lib. Let's test this.

pip install fixed_wheels/lfkbenchmark-0.1.0.dev0-cp39-cp39-macosx_10_9_universal2.whl
python -c "import lfkbenchmark"