一种在加速神经形态硬件上进行建模的可扩展方法。
A Scalable Approach to Modeling on Accelerated Neuromorphic Hardware.
作者信息
Müller Eric, Arnold Elias, Breitwieser Oliver, Czierlinski Milena, Emmel Arne, Kaiser Jakob, Mauch Christian, Schmitt Sebastian, Spilger Philipp, Stock Raphael, Stradmann Yannik, Weis Johannes, Baumbach Andreas, Billaudelle Sebastian, Cramer Benjamin, Ebert Falk, Göltz Julian, Ilmberger Joscha, Karasenko Vitali, Kleider Mitja, Leibfried Aron, Pehle Christian, Schemmel Johannes
机构信息
Kirchhoff-Institute for Physics, Heidelberg University, Heidelberg, Germany.
Third Institute of Physics, University of Göttingen, Göttingen, Germany.
出版信息
Front Neurosci. 2022 May 18;16:884128. doi: 10.3389/fnins.2022.884128. eCollection 2022.
Neuromorphic systems open up opportunities to enlarge the explorative space for computational research. However, it is often challenging to unite efficiency and usability. This work presents the software aspects of this endeavor for the BrainScaleS-2 system, a hybrid accelerated neuromorphic hardware architecture based on physical modeling. We introduce key aspects of the BrainScaleS-2 Operating System: experiment workflow, API layering, software design, and platform operation. We present use cases to discuss and derive requirements for the software and showcase the implementation. The focus lies on novel system and software features such as multi-compartmental neurons, fast re-configuration for hardware-in-the-loop training, applications for the embedded processors, the non-spiking operation mode, interactive platform access, and sustainable hardware/software co-development. Finally, we discuss further developments in terms of hardware scale-up, system usability, and efficiency.
神经形态系统为扩大计算研究的探索空间提供了机会。然而，将效率和可用性结合起来往往具有挑战性。这项工作介绍了针对BrainScaleS-2系统在这方面的软件情况，该系统是一种基于物理建模的混合加速神经形态硬件架构。我们介绍了BrainScaleS-2操作系统的关键方面：实验工作流程、API分层、软件设计和平台操作。我们通过用例来讨论并得出对软件的要求，并展示其实现情况。重点在于新颖的系统和软件特性，如多隔室神经元、用于硬件在环训练的快速重新配置、嵌入式处理器的应用、非尖峰操作模式、交互式平台访问以及可持续的硬件/软件协同开发。最后，我们讨论了在硬件扩展、系统可用性和效率方面的进一步发展。
Zotero 插件