AMD's library for high peformance machine learning primitives. MIOpen supports two programming models -
- OpenCL
- MinGW development environment
- OpenCL headers and OCL SDK
- Half at least version 1.12.0
- OpenSSL at least version 1.1.0i
- Boost at least version 1.5x.x
- You will need also the followng DLLs:
- libstdc++-6, ssl, crypto (use MinGW Installation Manager. Note 32 or 64 bit compatibility)
- MinGW with Boost Don't forget to set the env. variable
BOOST_INCLUDEDIR = ...\MinGW\include
- OCL SDK
- OpenSSL
- MinGW Installation Manager
First create a build directory:
mkdir build; cd build;
Next configure cmake. The preferred backend for MIOpen can be set using the -DMIOPEN_BACKEND
cmake variable.
cmake -DMIOPEN_BACKEND=OpenCL ..
The above assumes that OpenCL is installed in one of the standard locations. If not, then manually set these two cmake variables:
cmake -DMIOPEN_BACKEND=OpenCL -DOPENCL_LIBRARIES=<opencl-library-path> -DOPENCL_INCLUDE_DIRS=<opencl-headers-path> ..
The library can be built, from the build
directory using the 'Release' configuration:
cmake --build . --config Release
OR make
And can be installed by using the 'install' target:
cmake --build . --config Release --target install
OR make install
This will install the library to the CMAKE_INSTALL_PREFIX
path that was set.