Installation

Building from a Dockerfile

1- Make sure you have Docker installed (see Docker installation guide for: MacOS, Ubuntu Linux or Windows). Also, make sure to configure Docker to use sufficient resources (e.g., enough CPU cores).

2- Download the Dockerfile locally in some directory (we create one here):

$ mkdir pFaces
$ cd pFaces
$ curl https://raw.githubusercontent.com/parallall/pFaces/master/Dockerfile -o Dockerfile

3- Build the Docker image (don't forget the DOT at the end):

$ docker build -t pfaces/latest .

4- The building process will take proximately 10 minutes. During the build, you may receive some red-colored messages. They do not indicate errors, unless you receive an explicit red-colored error message. Once done, run the container and enter the interactive shell:

$ docker run -it pfaces/latest

5- In the Docker image, we installed Oclgrind to simulate an OpenCL platform/device that utilizes all the CPU cores using threads. As we use Oclgrind inside Docker, pFaces commands MUST be preceded with the command: oclgrind. For example, to check available devices using Oclgrind and pFaces, run:

/# oclgrind pfaces -CG -l

Installation from Linux/MacOS binaries

1- Install the OpenCL drivers. You may consider the helper scripts in pfaces-sdk/build-utils.

2- Download and unpack latest release of pFace.

3- Run the install script (install.sh) in from a terminal using your preferred sh, bash or dash.

Installation from Windows binaries

1- Make sure you install the required drivers or OpenCL runtimes for the targeted devices.

2- Make sure you have latest MS Visual C++ libraries.

3- Download and unpack latest release of pFace.

4- Run the install script (install.bat).