A Software Framework for Neuromorphic Computing

An Interactive Streamlit App version of a notebook tutorial The Jupyter Notebook that the app is based on

Overview

Lava is an open source SW framework to develop applications for neuromorphic hardware architectures. It provides developers with the abstractions and tools to develop distributed and massively parallel applications. These applications can be deployed to heterogeneous system architectures containing conventional processors as well as neuromorphic chips that exploit event-based message passing for communication. The Lava framework comprises high-level libraries for deep learning, constrained optimization, and others for productive algorithm development. It also includes tools to map those algorithms to different types of hardware architectures.

Today Lava supports conventional CPUs and Intel's Loihi architecture, but its compiler and runtime are open to extension for other architectures.

To learn more about the Lava Software Framework, please refer to the detailed documentation at http://lava-nc.org/.

The Lava framework is licensed with permissive open source BSD 3 licensing to highly encourage community contributions. Lower level components in Lava, that map algorithms to different hardware backends, are licensed with the LGPL-2.1 license to discourage commercial proprietary forks. Specific sensitive components supporting architectures like Intel Loihi may remain proprietary to Intel and will be shared as extensions to eligible users.

Lava extension for Intel's Loihi

The Lava extension for Loihi is available for members of the Intel Neuromorphic Research Community (INRC). The extension enables execution of Lava on Intel's Loihi hardware platform.

Developers interested in using Lava with Loihi systems need to join the INRC. Loihi 1 and 2 research systems are currently not available commercially. Once a member of the INRC, developers will gain access to cloud-hosted Loihi systems or may be able to obtain physical Loihi systems on a loan basis.

To join the INRC, visit http://neuromorphic.intel.com or email at [email protected].

If you are already a member of the INRC, please read how to get started with the Lava extension for Loihi. This page is only accessible to members of the INRC.

Getting started

The open-source Lava Software framework and its complementary algorithm libraries are hosted at http://github.com/lava-nc and the framework supports at minimimum CPU backends.

Note that you should install the core Lava repository lava before installing other Lava libraries such as lava-optimization or lava-dl.

Installing Lava from source

If you are interested in developing in Lava and modifying Lava source code, we recommend cloning the repository and using poetry to setup Lava. You will need to install the poetry Python package.

Open a python 3 terminal and run based on the OS you are on:

[Linux/MacOS]

cd $HOME
pip install -U pip
pip install "poetry>=1.1.13"
git clone [email protected]:lava-nc/lava.git
cd lava
git checkout v0.4.0
./utils/githook/install-hook.sh
poetry config virtualenvs.in-project true
poetry install
source .venv/bin/activate
pytest

## When running tests if you see 'OSError: [Errno 24] Too many open files'
## consider setting ulimit using `ulimit -n 4096`
## See FAQ for more info: https://github.com/lava-nc/lava/wiki/Frequently-Asked-Questions-(FAQ)#install

[Windows]

# Commands using PowerShell
cd $HOME
git clone git@github.com:lava-nc/lava.git
cd lava
git checkout v0.4.0
python3 -m venv .venv
.venv\Scripts\activate
pip install -U pip
pip install "poetry>=1.1.13"
poetry config virtualenvs.in-project true
poetry install
pytest

You should expect the following output after running the unit tests:

$ pytest
============================================== test session starts ==============================================
platform linux -- Python 3.8.10, pytest-7.0.1, pluggy-1.0.0
rootdir: /home/user/lava, configfile: pyproject.toml, testpaths: tests
plugins: cov-3.0.0
collected 205 items

tests/lava/magma/compiler/test_channel_builder.py .                                                       [  0%]
tests/lava/magma/compiler/test_compiler.py ........................                                       [ 12%]
tests/lava/magma/compiler/test_node.py ..                                                                 [ 13%]
tests/lava/magma/compiler/builder/test_channel_builder.py .                                               [ 13%]

...... pytest output ...

tests/lava/proc/sdn/test_models.py ........                                                               [ 98%]
tests/lava/proc/sdn/test_process.py ...                                                                   [100%]
=============================================== warnings summary ================================================

...... pytest output ...

src/lava/proc/lif/process.py                                                           38      0   100%
src/lava/proc/monitor/models.py                                                        27      0   100%
src/lava/proc/monitor/process.py                                                       79      0   100%
src/lava/proc/sdn/models.py                                                           159      9    94%   199-202, 225-231
src/lava/proc/sdn/process.py                                                           59      0   100%
-----------------------------------------------------------------------------------------------------------------TOTAL
                                                                                     4048    453    89%

Required test coverage of 85.0% reached. Total coverage: 88.81%
============================ 199 passed, 6 skipped, 2 warnings in 118.17s (0:01:58) =============================

[Alternative] Installing Lava via Conda

If you use the Conda package manager, you can simply install the Lava package via:

conda install lava -c conda-forge

Alternatively with intel numpy and scipy:

conda create -n lava python=3.9 -c intel
conda activate lava
conda install -n lava -c intel numpy scipy
conda install -n lava -c conda-forge lava --freeze-installed

[Alternative] Installing Lava from binaries

If you only need to install Lava as a user in your python environment, we will publish Lava releases via GitHub Releases. Please download the package and install it.

Open a Python terminal and run:

[Windows/MacOS/Linux]

python -m venv .venv
source .venv/bin/activate ## Or Windows: .venv\Scripts\activate
pip install -U pip
pip install lava-nc-0.4.0.tar.gz

Linting, testing, documentation and packaging

# Install poetry
pip install "poetry>=1.1.13"
poetry config virtualenvs.in-project true
poetry install
poetry shell

# Run linting
flakeheaven lint src/lava tests

# Run unit tests
pytest

# Create distribution
poetry build
#### Find builds at dist/

# Run Secuity Linting
bandit -r src/lava/.

#### If security linting fails run bandit directly
#### and format failures
bandit -r src/lava/. --format custom --msg-template '{abspath}:{line}: {test_id}[bandit]: {severity}: {msg}'

Refer to the tutorials directory for in-depth as well as end-to-end tutorials on how to write Lava Processes, connect them, and execute the code.

Stay in touch

To receive regular updates on the latest developments and releases of the Lava Software Framework please subscribe to our newsletter.