Fzenke Auryn Versions Save

Auryn: A fast simulator for spiking neural networks with synaptic plasticity

v0.8.3

8 months ago

Fixes several bugs. Adds ConcatGroup and several other improvements.

v0.8.3rc

9 months ago

Release candidate for Auryn v0.8.3 maintenance release with reworked SyncBuffer, added ConcatGroup, and improvements (FileInputGroup, Python tools, etc).

v0.8.2m

5 years ago

Auryn v0.8.2 is an overdue maintenance release which includes various fixes and minor improvements.

Main improvements where made on:

  • Code compatibility with non x86 architectures
  • Python3 support for the Auryn Python tools
  • Improvements to modular synapse/neuron models
  • Added an example of a 2-compartment model (NaudGroup)

This closes #33

v0.8.1

6 years ago

Multiple updates and minor extensions to AurynVector class

v0.8.0

7 years ago

Adds complex synapse support and OOP-based vectorization throughout.

v0.8.0-beta3

7 years ago

Bugfixes and change of name of Python tools analysis library.

v0.8.0-beta2

7 years ago

Fixes memory alignment problem and some other bugs. Moreover, merges some features from testing branch:

  • Renames global variable holding the simulation timestep dt to auryn_timestep
  • Adds master seeding mechanisms in System
  • Introduces abstract base class Trace for all synaptic traces
  • Trace is a derived class of AurynStateVector

v0.8.0-beta

7 years ago

Adds

  • Boost unit tests
  • Python analysis code for binary output files (see tools/python)
  • New abstract base class Device for Monitor

Important interface changes

  • Concerning the global variables of MPI environment: mpienv and mpicommunicator (instead of env and communicator) now defined in auryn_global.h and initialized via auryn_init() function -- see examples

v0.8.0-alpha

7 years ago

Pre-release of Auryn 0.8.0

This release comes with numerous upgrades, performance improvements and bug fixes.

Performance history from Auryn 0.7.0 to Auryn 0.8.0:

auryn_v0 8_benchmark_results

Main changes

  • Complete refactor of SyncBuffer for parallel performance
  • Complete refactor of AurynStateVector as a template class
  • Refactor of IFGroup and AIFGroup to new AurynVector interface
  • Changes memory alignment of AurynVector data for SIMD optimizations
  • Improved handling of ComplexMatrix states and
  • Performance optimizations in ComplexMatrix
  • Performance optimizations in SparseConnection
  • Performance optimizations in Connection
  • Performance optimizations in AurynVector
  • API changes in Connection
  • Adds groupstate variables to SpikingGroup
  • Adds AllToAllConnection
  • Adds IafPscExpGroup
  • Adds CubaIFGroup
  • Adds Izhikevich neuron model als IzhikevichGroup
  • Adds BinaryStateMonitor and the decoder tool "aubs" (a better name might be in order though)

v0.7.0

8 years ago