GeNN

GeNN accelerates large-scale numerical simulations of detailed brain circuit models by generating GPU-optimized code for execution on NVIDIA GPUs to enable efficient testing of hypotheses about brain function.

Key Features:

• Code generation: Generates optimized GPU-targeted simulation code tailored for execution on NVIDIA GPUs.
• GPU execution: Leverages NVIDIA GPUs to increase computational throughput for network simulations.
• Scalability and performance: Demonstrated up to a 200-fold speedup versus single-core CPU when simulating a network of one million conductance-based Hodgkin-Huxley neurons.
• Conductance-based Hodgkin-Huxley model support: Supports simulation of conductance-based Hodgkin-Huxley neuron models.
• Extensible model specification: Provides a flexible, extensible code-generation framework for custom neuron and synapse definitions.

Scientific Applications:

• Large-scale neuronal network simulation: Enables simulation of complex, large-scale neuronal networks.
• Hypothesis testing for brain function: Facilitates formulation and testing of hypotheses about brain function via simulated networks.
• High-performance conductance-based simulations: Supports scalable simulation of conductance-based Hodgkin-Huxley neuron networks.
• Performance benchmarking: Serves for comparing GPU-accelerated simulation performance against CPU baselines.

Methodology:

Generates GPU-optimized simulation code via code generation and executes it on NVIDIA GPUs, demonstrated on simulations of conductance-based Hodgkin-Huxley neuron networks with reported speedups.