Brian2CUDA is an extension of the spiking neural network simulator Brian2, written in Python. It generates C++/CUDA code to run simulations on NVIDIA GPUs.
For support, please use the Brian forum. If you think you found a bug in Brian2CUDA, please report it at the GitHub issue tracker.
For installation and usage instructions, check out the Brian2CUDA documentation. For information on general Brian2 usage, check out the Brian2 documentation.
You can install Brian2CUDA via pip:
python -m pip install brian2cudaThis will install a compatible version of Brian2 as dependency. For installation requirements and GPU configuration, check out the Brian2CUDA documentation.
Use your Brian2 code (see Brian2 documentation) and modify the imports to:
# Standard Brian2 import
from brian2 import *
# Enable GPU usage via Brian2CUDA
import brian2cuda
set_device("cuda_standalone")See Brian2's standalone code generation for more options for the set_device call.
If you use this software in a published article, please cite our Brian2CUDA publication:
Alevi, D, Stimberg, M, Sprekeler, H, Obermayer, K, Augustin, M. “Brian2CUDA: flexible and efficient simulation of spiking neural network models on GPUs” Frontiers in Neuroinformatics (2022). doi: 10.3389/fninf.2022.883700.
Brian2CUDA is free software licensed under the GNU General Public License v3 (GPLv3).
To run the test suite on Google Collab (no installation or GPU required), click on the badge below:
