Skip to content

Latest commit

 

History

History
31 lines (20 loc) · 2.08 KB

LOCAL_INSTALL.md

File metadata and controls

31 lines (20 loc) · 2.08 KB
Raw

NOTE: It is far more complex to set up OpenLane without a Docker container. If you are a novice user, it is recommended to use the Docker container.

Using OpenLane without Docker

At its core, OpenLane is a set of scripts working with a set of tools. If you'd like to avoid using a Docker container, you can, but you will have to set up all of the tools required by OpenLane on your computer. We do provide a script to assist you with that.

Base Requirements

  • Python 3.6+ with PIP
  • pyyaml: python3 -m pip install pyyaml

Tool Library

You can run python3 ./ol_install.py --list-tools for a list. There are at least a dozen tools to install here. Luckily, you don't have to install them all one-by-one: There is an installation script that installs most of them.

You can invoke python3 ./ol_install.py. This tool copies the skeleton and installs all the tools to a directory of your choice, which is by default, /opt/openlane. Furthermore, if you are on CentOS 7, macOS or Ubuntu 20.04, the installer will offer to install all the required apt or yum packages for you.

On macOS, it may be prudent to invoke it as SKIP_TOOLS=drcu:cugr python3 ./ol_install.py instead, as these tools are a nightmare to compile on macOS.

The tools will all be installed with /opt/openlane as a prefix. You'll find all the repos in /opt/openlane/build/repos and a list of versions in /opt/openlane/build/versions.

DO NOTE: We expect you to bring your own OpenROAD. This installer will make no attempt to install OpenROAD.

After the installer is done, you can invoke sh /opt/openlane/openlane <args> to use OpenLane, where args are the same arguments you'd pass on to flow.tcl.

More about how this works

OpenLane can work as a skeleton with this file structure:

  • configuration/
  • scripts/
  • flow.tcl
  • report_generation_wrapper.py

You can copy them into any folder, then invoke tclsh /path/to/flow.tcl and go to town. The scripts are pretty light on requirements too: You only need Python 3.6+, Perl 5 and Tclsh. Unfortunately, OpenLane cannot accomplish much without its library of open source EDA tooling.