Fig. 2: Dendritic Hierarchical Scheduling (DHS) method significantly reduces the computational cost, i.e., computational steps in solving equations.
From: A GPU-based computational framework that bridges neuron simulation and artificial intelligence
a DHS work flow. DHS processes k deepest candidate nodes each iteration. b Illustration of calculating node depth of a compartmental model. The model is first converted to a tree structure then the depth of each node is computed. Colors indicate different depth values. c Topology analysis on different neuron models. Six neurons with distinct morphologies are shown here. For each model, the soma is selected as the root of the tree so the depth of the node increases from the soma (0) to the distal dendrites. d Illustration of performing DHS on the model in b with four threads. Candidates: nodes that can be processed. Selected candidates: nodes that are picked by DHS, i.e., the k deepest candidates. Processed nodes: nodes that have been processed before. e Parallelization strategy obtained by DHS after the process in d. Each node is assigned to one of the four parallel threads. DHS reduces the steps of serial node processing from 14 to 5 by distributing nodes to multiple threads. f Relative cost, i.e., the proportion of the computational cost of DHS to that of the serial Hines method, when applying DHS with different numbers of threads on different types of models.
