Skip to content

Latest commit

 

History

History
86 lines (68 loc) · 3.53 KB

INSTALL.md

File metadata and controls

86 lines (68 loc) · 3.53 KB
Raw

FlexFlow Installation

FlexFlow can be built from source code using the following instructions.

1. Download the source code

  • Clone the FlexFlow source code from the github.
git clone --recursive https://github.com/flexflow/FlexFlow.git

FlexFlow Python dependencies

  • The FlexFlow Python support requires several additional Python libraries, please check this for details. We recommend to use pip or conda to install the dependencies.

Note: all Python dependencies will be automatically installed if install the FlexFlow Python Interface using the PyPi repository (see the Installation below).

2. Build the FlexFlow

It is prefer to use the CMake building system as it will automatically build all dependencies inlcuding NCCL and Legion.

Build the FlexFlow (including C++ and Python)

The config/config.linux is an example of how to set the varibles required for CMake build. The followings are configurable parameters:

  • CUDA_DIR is used to specify the directory of CUDA. It is only required when CMake can not automatically detect the installation directory of CUDA.
  • CUDNN_DIR is only required when CUDNN is not installed in the CUDA directory.
  • FF_CUDA_ARCH is used to set the architecture of targeted GPUs, for example, the value can be 60 if the GPU architecture is Pascal. If it is not sepecified, FlexFlow is compiled for all architectures.
  • FF_USE_PYTHON is used to enable the Python support for the FlexFlow.
  • FF_USE_NCCL is used to enable the NCCL support for the FlexFlow, by default it is set to ON.
  • FF_USE_GASNET is used to enable distributed run of the FlexFlow.
  • FF_BUILD_EXAMPLES is used to enable all C++ examples.
  • FF_MAX_DIM is used to set the maximum dimension of tensors, by default it is set to 4.

More options are available in cmake, please run ccmake and search for options starting with FF.

Once the variables in the config.linux is set correctly, go to the home directory of FlexFlow, and run

mkdir build
cd build
../config/config.linux
make

3. Test the FlexFlow

1. Set the FF_HOME environment variable before running the FlexFlow. You can add the following line in ~/.bashrc.

export FF_HOME=/path/to/FlexFlow

2. Run FlexFlow Python examples

The C++ examples are in the examples/python. For example, the AlexNet can be run as:

cd python
./flexflow_python examples/python/native/alexnet.py -ll:py 1 -ll:gpu 1 -ll:fsize <size of gpu buffer> -ll:zsize <size of zero buffer>

The script of running all the Python examples is python/test.sh

3. Run FlexFlow C++ examples

The C++ examples are in the examples/cpp. For example, the AlexNet can be run as:

./alexnet -ll:gpu 1 -ll:fsize <size of gpu buffer> -ll:zsize <size of zero buffer>

Size of buffers is in MBs, e.g. for an 8GB gpu -ll:fsize 8000

4. Install the FlexFlow

1. Install the FlexFlow binary, header file and library if using CMake.

cd build
make install

2. Install the FlexFlow Python interface using pip

If install from local:

pip install .

All Python depencies will be automatically installed. The default installation is configurated with FF_USE_NCCL=OFF FF_USE_GASNET=OFFand FF_MAX_DIM=4. In order to change the configuration, please use the --global-option of the setuptools

pip install --global-option="build_ext" --global-option="CMake options" .