Abstract
In the last decade there has been a surge in the number of big science projects interested in achieving a comprehensive understanding of the functions of the brain, using Spiking Neuronal Network (SNN) simulations to aid discovery and experimentation. Such an approach increases the computational demands on SNN simulators: if natural scale brain-size simulations are to be realized, it is necessary to use parallel and distributed models of computing. Communication is recognized as the dominant part of distributed SNN simulations. As the number of computational nodes increases, the proportion of time the simulation spends in useful computing (computational efficiency) is reduced and therefore applies a limit to scalability. This work targets the three phases of communication to improve overall computational efficiency in distributed simulations: implicit synchronization, process handshake and data exchange. We introduce a connectivity-aware allocation of neurons to compute nodes by modeling the SNN as a hypergraph. Partitioning the hypergraph to reduce interprocess communication increases the sparsity of the communication graph. We propose dynamic sparse exchange as an improvement over simple point-to-point exchange on sparse communications. Results show a combined gain when using hypergraph-based allocation and dynamic sparse communication, increasing computational efficiency by up to 40.8 percentage points and reducing simulation time by up to 73%. The findings are applicable to other distributed complex system simulations in which communication is modeled as a graph network.
Highlights
IntroductionThere has been a growing scientific focus on computational neuroscience as a means to understand the brain and its functions, at large, brain-size scale (Alivisatos et al, 2012; Koch, 2012; Markram, 2012; Amunts et al, 2016; Poo et al, 2016)
The first series of strong scaling experiments are run to understand the impact of the novel neuron allocation strategy and the application of Dynamic Sparse Data Exchange (DSDE) communication patterns in Spiking Neuronal Network (SNN) simulation
The baseline neuron allocation strategy Random Balanced is a variation of random allocation that takes the number of postsynaptic connections into account to keep processes balanced
Summary
There has been a growing scientific focus on computational neuroscience as a means to understand the brain and its functions, at large, brain-size scale (Alivisatos et al, 2012; Koch, 2012; Markram, 2012; Amunts et al, 2016; Poo et al, 2016). Simulating brain-size models is computationally challenging due to the number and variety of elements involved and the high level of interconnectivity between them (Ananthanarayanan et al, 2009). The computing resources required at the brain scale far surpass the capabilities of personal computers today and will require compute power at the exascale (Markram, 2012)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
