GeNN (GPU enhanced Neuronal Network) is a cool bit of software for mapping neural models to GPUs (via CUDA), and they're showing that modern GPUs are as good or even better than (some) neuromorphic hardware. This'd be a great set of benchmarks for comparison to nengo (and to show how to do non-NEF things in nengo)...
https://genn-team.github.io/genn/documentation/4/html/d9/d61/Examples.html
Here's a comparison of an 80,000 LIF-neuron model
https://www.frontiersin.org/articles/10.3389/fnins.2018.00941/full
https://www.frontiersin.org/articles/10.3389/fnins.2018.00291/full
And here's some classifiers along with some detailed speed benchmarking. It also separates out initialization time and shows some neat advantages for doing some of your initialization inside the GPU itself....
https://www.frontiersin.org/articles/10.3389/fnins.2015.00491/full
GeNN (GPU enhanced Neuronal Network) is a cool bit of software for mapping neural models to GPUs (via CUDA), and they're showing that modern GPUs are as good or even better than (some) neuromorphic hardware. This'd be a great set of benchmarks for comparison to nengo (and to show how to do non-NEF things in nengo)...
https://genn-team.github.io/genn/documentation/4/html/d9/d61/Examples.html
Here's a comparison of an 80,000 LIF-neuron model
https://www.frontiersin.org/articles/10.3389/fnins.2018.00941/full
https://www.frontiersin.org/articles/10.3389/fnins.2018.00291/full
And here's some classifiers along with some detailed speed benchmarking. It also separates out initialization time and shows some neat advantages for doing some of your initialization inside the GPU itself....
https://www.frontiersin.org/articles/10.3389/fnins.2015.00491/full