Getting Started

 

To be informed about OpenCPI events (releases, updates, etc.), you can subscribe to the OpenCPI email list for announcements and discussions by subscribing to it. To do this see the email lists page.

To learn about OpenCPI, beyond what is on the Overview page, without looking at or using any code, you can reference the documentation summarized here.

Supported development systems (which can of course also execute OpenCPI applications), include CentOS6 and CentOS7 Linux distributions as well as MacOS Sierra and Mojave.  If developing for FPGAs, your development system must support tools such as Xilinx ISE and ISim, Xilinx Vivado and Xsim, Altera Quartus, or Modelsim.  These are not supported on MacOS.  Bringing up OpenCPI on other Linux distributions has been done, but is not yet covered by the documented installation procedures.

Target systems (for execution) are composed of one or more supported execution platforms, with the host platform being a software-based platform (Linux or MacOS), and other connected platforms (e.g. FPGA cards) acting as additional processors in the system.  The ZedBoard system (see zedboard.org) is the least expensive supported target system that includes both a GPP host platform (dual-core ARM) and an FPGA platform, all on a single Zynq SoC chip.  This system can execute OpenCPI applications consisting of RCC workers for the ARM cores and HDL workers executing on the FPGA part of the SoC.  Target systems may also be a PCI-Express-based x86_64 system with PCI-Express-based FPGA cards plugged in.  OpenCPI currently supports both the ML605 (Xilinx Virtex6) and ALST4 (Altera Stratix4) development boards.

To install OpenCPI, which requires sudo privileges and the bash UNIX/Linux shell:

  • To install based on the source code at gitlab.com here, you can read the detailed install-from-source installation guide and follow its instructions.  These instructions include setting up the development systems from scratch (or in virtual machines), using FPGA simulators, and perhaps setting up some target embedded systems.  The basic installation steps for developing and executing on the development host can be accomplished by executing these steps:

    • Select the specific release on the OpenCPI gitlab.com repository page here.

    • Download the tar file for the release and expand it, which will create a directory with the name: opencpi-<release-tag>.

    • Enter that directory and type the command:

      • ./scripts/install-opencpi.sh

    This installation style has been tested on the CentOS6 and CentOS7 Linux distributions as well as MacOS Catalina (10.15) and some earlier versions.

To directly look at the source code for OpenCPI, the GitLab project page is here.  The OpenCPI source is licensed as described here.

To peruse or submit bug reports or issues, or to see discussion topics, see the Bugs/Issues page.