I think that in a physical sense you can compare computation throughput unfairly against GPUs, by comparing 1 FLO against a minimum window of time in which a pa

I think that in a physical sense you can compare computation throughput unfairly against GPUs, by comparing 1 FLO against a minimum window of time in which a particular neuron spiking or not is significant, multiplied by the number of neurons.