Skip to content

Latest commit

 

History

History
189 lines (128 loc) · 5.68 KB

CHANGES.rst

File metadata and controls

189 lines (128 loc) · 5.68 KB
Raw

Release History

3.0.1 (unreleased)

3.0.0 (November 16, 2023)

Compatible with Nengo 3.1.0

Added

  • The benchmarks in the examples folder now have nicer command-line interfaces. Use the --help flag with any benchmark to learn more about the options. (#187)
  • The new examples/benchmark_backends.py script makes it easier to compare between different backends on any of the benchmarks. (#187)

Changed

  • Sparse matrix multiplication is now faster in many cases by using the ELLPACK matrix format. It uses more memory for some sparse matrices, though; for matrices where it would result in a large increase in memory usage, we fall back on the old CSR format. To force a particular format, set the NENGO_OCL_SPMV_ALGORITHM environment variable to either "ELLPACK" or "CSR". (#188)
  • Made NengoOCL available under the GPLv2 license. (#191)

Removed

  • Dropped support for Python 3.5. (#187)

2.1.0 (Nov 23, 2020)

Compatible with Nengo 3.1.0

Added

  • Added remove_zero_incs and remove_unmodified_resets simplifications for the operator list. These are enabled by default, and remove unnecessary operators (e.g. that are multiplying by zero and adding that to a signal). This increases both build speed and run speed. These simplifications can be disabled by modifying nengo_ocl.operators.simplifications. (#183)

Changed

  • Added support for Nengo 3.1.0, and retired support for Nengo 3.0.0. (#180)
  • Changes to improve benchmarks, including comparing between benchmarks. (#182)

2.0.0 (Sept 4, 2020)

Compatible with Nengo 3.0.0

Added

  • Sparse transforms are now supported. (#176)
  • Added Simulator.clear_probes method to clear probe data stored in memory. (#179)

Changed

  • Now requires Python >= 3.5. (#172)
  • Now supports Nengo 3.0.0. Note that support for previous Nengo versions has been dropped. (#172)
  • Convolution transforms are now supported. The previous code supporting Conv2d and Pool2d processes (from NengoExtras) has been removed. (#172)

1.4.0 (July 4, 2018)

Improvements

  • Supports recent Nengo versions, up to 2.8.0.
  • Supports the new SpikingRectifiedLinear neuron type.

1.3.0 (October 6, 2017)

Improvements

  • Supports recent Nengo versions, up to 2.6.0.

Bugfixes

  • Fixed an issue in which stochastic processes would not be fully reset on simulator reset.
  • Fixed an issue in which building a model multiple times could result in old probe data persisting.

1.2.0 (February 23, 2017)

Improvements

  • Supports all Nengo versions from 2.1.2 to 2.3.1.
  • nengo_ocl.Simulator is no longer a subclass of nengo.Simulator, reducing the chances that Nengo OCL will be affected by changes in Nengo.

1.1.0 (November 30, 2016)

Features

  • Added support for RectifiedLinear and Sigmoid neuron types.
  • Added support for arbitrary Process subclasses. Unlike the processes that are explicitly supported like WhiteSignal, these processes may not fully utilize the OpenCL device.
  • Added support for applying synaptic filters to matrices, which is commonly done when probing connection weights.

Improvements

  • Supports all Nengo versions from 2.1.2 to 2.3.0.
  • The LIF model is now more accurate, and matches the implementation in Nengo (see Nengo#975).
  • Several operators have been optimized and should now run faster.

Bugfixes

  • Fixed compatibility issues with Python 3, and certain versions of NumPy and Nengo.

1.0.0 (June 6, 2016)

Release in support of Nengo 2.1.0. Since Nengo no longer supports Python 2.6, we now support Python 2.7+ and 3.3+.

Features

  • Added support for Process class and subclasses, new in Nengo in 2.1.0. We specifically support the WhiteNoise, WhiteSignal, and PresentInput processes. We also support the Conv2d and Pool2d processes in nengo_extras.
  • LinearFilter is now fully supported, allowing for general synapses.

Improvements

  • The Numpy simulator in this project (sim_npy) has been phased out and combined with the OCL simulator (sim_ocl). It is now called Simulator and resides in simulator.py.
  • Operator scheduling (i.e. the planner) is much faster. We still have only one planner (greedy_planner), which now resides in planners.py.
  • Many small speed improvements, including a number of cases where data was needlessly copied off the device to check sizes, dtypes, etc.

Documentation

  • Updated examples to use up-to-date Nengo syntax.

0.1.0 (June 8, 2015)

Initial release of Nengo OpenCL! Supports Nengo 2.0.x on Python 2.6+ and 3.3+.