GeNN 5.3.0

Release Notes for GeNN 5.3.0

This release adds a number of significant new features to GeNN as well as including a number of bug fixes that have been identified since the 5.2.0 release. We also proud to say that GeNN is now part of the Open Neuromorphic community and is an official target for the NESTML modelling language.

User Side Changes

• Added support for CUDA Array Interface for interoperability with other Python libraries (#675)
• Expose CurrentSource target_var to PyGeNN (#676)
• Add NVTX profiling support to CUDA backend (#680)
• Add support for building generated code using NMake as well as MSBuild build systems on windows (#700, #705)
• Extended the timing system to record the time spent in each custom update group running host code from custom connectivity updates (#701)

Bug fixes

• Fixed several issues working with sparse connectivity in PyGeNN (#694)
• Fixed several bugs with the transpiler (#679, #696, #702)
• Fixes RNG initialisation issues in networks with very simple neuron models which use an RNG (#685)
• Fixed several issues with building GeNN on Mac OS X (#707)
• Fixed crashes when using CUDA backend on latest Visual Studio 2022 builds (#716)

Optimisations

• The code generator now marks relevant pointers with a backend-specific restrict keyword (#690)

New Contributors

Massive thanks to some new contributors to GeNN:

• @JRV7903 made their first contribution in #675
• @clinssen made their first contribution in #676
• @neural-loop made their first contribution in #699
• @Agrim-P777 made their first contribution in #705

Special thanks to @JRV7903 and @Agrim-P777 who have just completed very successful Google Summer of Code projects with us, working on Conda packaging and ISPC support respectively.