Skip to content

Latest commit

 

History

History
157 lines (112 loc) · 4.63 KB

basic_setup.md

File metadata and controls

157 lines (112 loc) · 4.63 KB

Go Back

Basic Setup

This page shows the basic setup for using OpenRAM to generate an SRAM.

Table of Contents

  1. Dependencies
  2. Anaconda
  3. Docker
  4. Environment
  5. Sky130 Setup

Dependencies

In general, the OpenRAM compiler has very few dependencies:

  • Git
  • Make
  • Python 3.5 or higher
  • Various Python packages (pip install -r requirements.txt)
  • Anaconda

Anaconda

We use Anaconda package manager to install the tools used by OpenRAM. This way, you don't have to worry about updating/installing these tools. OpenRAM installs Anaconda silently in the background (without affecting any existing Anaconda setup you have).

You don't have to manually activate/deactivate the Anaconda environment. OpenRAM automatically manages this before and after running the tools.

OpenRAM uses Anaconda by default, but you can turn this feature off by setting use_conda = False in your config file. Then, OpenRAM will use the tools you have installed on your system.

You can also tell OpenRAM where Anaconda should be installed or which Anaconda setup it should use. You can set the $CONDA_HOME variable like this:

export CONDA_HOME="/path/to/conda/setup"

Note: If you want to install Anaconda without running OpenRAM (for example to run unit tests, which do not install Anaconda), you can run:

./install_conda.sh

Note: You can uninstall OpenRAM's Anaconda installation by simply deleting the folder Anaconda is installed to. You can run:

rm -rf miniconda

Note: You can change a tool's version with the following commands:

source ./miniconda/bin/activate
conda uninstall <tool>
conda install -y -c vlsida-eda <tool>=<version>

Docker (deprecated, use Anaconda instead)

We have a docker setup to run OpenRAM. To use this, you should run:

cd OpenRAM/docker
make build

This must be run once and will take a while to build all the tools. If you have the OpenRAM library installed, you can also run the docker setup from the package installation directory.

Environment

If you haven't installed the OpenRAM library or you want to use a different OpenRAM installation, you can set two environment variables:

  • OPENRAM_HOME should point to the compiler source directory.
  • OPENRAM_TECH should point to one or more root technology directories (colon separated).

If you have the library installed and OPENRAM_HOME set, the library will use the installation on the OPENRAM_HOME path.

See Python library for details.

If you don't have the library, you should also add OPENRAM_HOME to your PYTHONPATH. This is not needed if you have the library.

You can add these environment variables to your .bashrc:

export OPENRAM_HOME="$HOME/OpenRAM/compiler"
export OPENRAM_TECH="$HOME/OpenRAM/technology"
export PYTHONPATH=$OPENRAM_HOME

Note that if you want symbols to resolve in your editor, you may also want to add the specific technology directory that you use and any custom technology modules as well. For example:

export PYTHONPATH="$OPENRAM_HOME:$OPENRAM_TECH/sky130:$OPENRAM_TECH/sky130/custom"

We include the tech files necessary for SCMOS SCN4M_SUBM, FreePDK45. The SCMOS spice models, however, are generic and should be replaced with foundry models. You may get the entire FreePDK45 PDK here.

Sky130 Setup

To install Sky130, you can run:

cd $HOME/OpenRAM
make sky130-pdk

This will use volare to get the PDK.

Note: If you don't have Magic installed, you need to install and activate the conda environment before running this command. You can run:

./install_conda.sh
source miniconda/bin/activate

Then you must also install the Sky130 SRAM build space with the appropriate cell views into the OpenRAM technology directory by running:

cd $HOME/OpenRAM
make sky130-install

You can also run these from the package installation directory if you have the OpenRAM library.

GF180 Setup

OpenRAM currently does not support gf180mcu for SRAM generation. However ROM generation for gf180mcu is supported as an experimental feature.

It is not necessary to install the gf180mcu PDK, as all necessary files are already in the git repository under technology/gf180mcu/.

If you still want to install the PDK, you can run make gf180mcu-pdk.