Arbor — A Morphologically-Detailed Neural Network Simulation Library for Contemporary High-Performance Computing Architectures

Akar, Nora Abi; Cumming, Ben; Karakasis, Vasileios; Küsters, A. Mesquida; Klijn, Wouter; Peyser, Alexander; Yates, Stuart

doi:10.1109/empdp.2019.8671560

Cited by 69 publications

(60 citation statements)

References 13 publications

(17 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…8). These results have been qualitatively confirmed in cross-platform performance studies (Knight and Nowotny 2018;Kumbhar et al 2019a;Akar et al 2019). In this work, however, we are able to dig deeper and identify substantial differences in the hardware bottleneck profiles of individual in silico models.…”

Section: Strawman Architectures and Hardware Design Decisionssupporting

confidence: 70%

Understanding Computational Costs of Cellular-Level Brain Tissue Simulations Through Analytical Performance Models

Cremonesi

Schürmann

2020

Neuroinform

View full text Add to dashboard Cite

Computational modeling and simulation have become essential tools in the quest to better understand the brain's makeup and to decipher the causal interrelations of its components. The breadth of biochemical and biophysical processes and structures in the brain has led to the development of a large variety of model abstractions and specialized tools, often times requiring high performance computing resources for their timely execution. What has been missing so far was an in-depth analysis of the complexity of the computational kernels, hindering a systematic approach to identifying bottlenecks of algorithms and hardware. If whole brain models are to be achieved on emerging computer generations, models and simulation engines will have to be carefully co-designed for the intrinsic hardware tradeoffs. For the first time, we present a systematic exploration based on analytic performance modeling. We base our analysis on three in silico models, chosen as representative examples of the most widely employed modeling abstractions: current-based point neurons, conductance-based point neurons and conductance-based detailed neurons. We identify that the synaptic modeling formalism, i.e. current or conductance-based representation, and not the level of morphological detail, is the most significant factor in determining the properties of memory bandwidth saturation and shared-memory scaling of in silico models. Even though general purpose computing has, until now, largely been able to deliver high performance, we find that for all types of abstractions, network latency and memory bandwidth will become severe bottlenecks as the number of neurons to be simulated grows. By adapting and extending a performance modeling approach, we deliver a first characterization of the performance landscape of brain tissue simulations, allowing us to pinpoint current bottlenecks for state-of-the-art in silico models, and make projections for future hardware and software requirements. Keywords Computational models of neurons • Brain tissue simulations • Performance modeling • High performance computing This study was supported by funding to the Blue Brain Project, a research center of theÉcole polytechnique fédérale de Lausanne (EPFL), from the Swiss governments ETH Board of the Swiss Federal Institutes of Technology Electronic supplementary material The online version of this article (

show abstract

Section: Strawman Architectures and Hardware Design Decisionssupporting

confidence: 70%

Understanding Computational Costs of Cellular-Level Brain Tissue Simulations Through Analytical Performance Models

Cremonesi

Schürmann

2020

Neuroinform

View full text Add to dashboard Cite

show abstract

“…To maintain maximal compatibility, we chose the path of extracting the computational relevant parts of NEURON into a library called CoreNEURON and adapting it to exploit the computational features of modern compute architectures. This is a different path as for example taken by the Arbor (Akar et al, 2019) which started its developments from scratch. While such a fresh start has its benefits in terms of designing for future architectures from the start, we can show that the transformation approach we took immediately gives compatibility with a large number of existing NEURON models with minimal modification.…”

Section: Discussionmentioning

confidence: 99%

CoreNEURON : An Optimized Compute Engine for the NEURON Simulator

Kumbhar

Hines

Fouriaux

et al. 2019

Front. Neuroinform.

View full text Add to dashboard Cite

The NEURON simulator has been developed over the past three decades and is widely used by neuroscientists to model the electrical activity of neuronal networks. Large network simulation projects using NEURON have supercomputer allocations that individually measure in the millions of core hours. Supercomputer centers are transitioning to next generation architectures and the work accomplished per core hour for these simulations could be improved by an order of magnitude if NEURON was able to better utilize those new hardware capabilities. In order to adapt NEURON to evolving computer architectures, the compute engine of the NEURON simulator has been extracted and has been optimized as a library called CoreNEURON. This paper presents the design, implementation, and optimizations of CoreNEURON. We describe how CoreNEURON can be used as a library with NEURON and then compare performance of different network models on multiple architectures including IBM BlueGene/Q, Intel Skylake, Intel MIC and NVIDIA GPU. We show how CoreNEURON can simulate existing NEURON network models with 4–7x less memory usage and 2–7x less execution time while maintaining binary result compatibility with NEURON.

show abstract

“…the technology investigated here. In simulators for networks of many-compartment neurons, such as NEURON (Carnevale and Hines, 2006) and Arbor (Akar et al, 2019) the problem is less pronounced as the investigated networks are typically of smaller size while single neurons are represented at a great level of detail. This typically results in high per-process workload while less data needs to be communicated among MPI processes, such that the overhead caused by communication is less noticeable.…”

Section: Discussionmentioning

confidence: 99%

Efficient Communication in Distributed Simulations of Spiking Neuronal Networks With Gap Junctions

Jordan

Helias

Diesmann

et al. 2020

Front. Neuroinform.

View full text Add to dashboard Cite

Investigating the dynamics and function of large-scale spiking neuronal networks with realistic numbers of synapses is made possible today by state-of-the-art simulation code that scales to the largest contemporary supercomputers. However, simulations that involve electrical interactions, also called gap junctions, besides chemical synapses scale only poorly due to a communication scheme that collects global data on each compute node. In comparison to chemical synapses, gap junctions are far less abundant. To improve scalability we exploit this sparsity by integrating an existing framework for continuous interactions with a recently proposed directed communication scheme for spikes. Using a reference implementation in the NEST simulator we demonstrate excellent scalability of the integrated framework, accelerating large-scale simulations with gap junctions by more than an order of magnitude. This allows, for the first time, the efficient exploration of the interactions of chemical and electrical coupling in large-scale neuronal networks models with natural synapse density distributed across thousands of compute nodes.

show abstract

Arbor — A Morphologically-Detailed Neural Network Simulation Library for Contemporary High-Performance Computing Architectures

Cited by 69 publications

References 13 publications

Understanding Computational Costs of Cellular-Level Brain Tissue Simulations Through Analytical Performance Models

Understanding Computational Costs of Cellular-Level Brain Tissue Simulations Through Analytical Performance Models

CoreNEURON : An Optimized Compute Engine for the NEURON Simulator

Efficient Communication in Distributed Simulations of Spiking Neuronal Networks With Gap Junctions

Contact Info

Product

Resources

About