Welcome to the NEST GPU documentation!ΒΆ
NEST GPU is a GPU library for the simulation of large-scale spiking neural networks. It is written in CUDA-C++ and supports multi-GPU simulations through MPI. Originally developed under the name NeuronGPU, the code joined the NEST Initiative e.V. and follows the concepts and practices already established by the CPU-based NEST Simulator. The Python interface allows users to define neurons, connections, and synapse properties using commands familiar from PyNEST. Under the hood, the design mirrors NEST Simulator where applicable, while enabling efficient simulations on the largest GPU-powered computing systems.
To get started with NEST GPU, install it on your system or browse the latest publications:
Golosio et al. (2026) Fig. 4
Large-scale spiking neural network simulations using up to thousands of GPUs
Golosio et al. (2023) Fig. A3d-f
Statistical match of network activity with NEST CPU (here for a cortical microcircuit model)
Golosio et al. (2026) Fig. 1
Point-to-point and collective communication using MPI for GPU clusters
Golosio et al. (2026) Fig. 3
Fast onboard network construction and simulation of multi-area model