Scribd
Upload
23 views

Computer Org and Architecture Presentation

This document discusses neuromorphic architecture and its implications as Moore's Law nears its end. It describes neuromorphic computers like True North, Neuro Grid, and Tianjic Chip which implement spiking neural networks with highly parallel and low power operations. Neuromorphic chips defy Moore's Law through physical limitations and Dennard scaling.

Uploaded by

ifranoide
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
23 views

Computer Org and Architecture Presentation

This document discusses neuromorphic architecture and its implications as Moore's Law nears its end. It describes neuromorphic computers like True North, Neuro Grid, and Tianjic Chip which implement spiking neural networks with highly parallel and low power operations. Neuromorphic chips defy Moore's Law through physical limitations and Dennard scaling.

Uploaded by

ifranoide
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 20

Neuromorphic Architecture

and its Implications in The Near End of Moor’s Law


First Look at Neuromorphic Computers
Why are they important anyway?
• Highly parallel operations
• Inherent scalability
• Event-driven computation
• extremely low power operation
• Autonomous systems
From Conventional CPUs to Complex Neural
Networks
Non-Spiking Neural Networks & Spiking
Neural Networks
Neuromorphic computers: True North
• SNN made by IBM
• Architecture consists of transistors
• 4096 SNN cores
• Distributed system memory
• Inherently, parallel, modular, distributed and fault-tolerant
Neuromorphic computers: Neuro Grid
• NeuroCores
• Shared Synaptic network across the entire board
NeuroCore chip block diagram. (Indiveri and Liu, 2015)
Neuromorphic computers: Tianjic Chip
Operation Modules of a Tianjic Chip. (Wang et al. 2021)
Unified Communication Format(Wang et al. 2021)
Adjustable Timing Schedule(Wang et al. 2021)
Execution of Neural Networks
How Neuromorphic chips defy Moor’s Law
• Physical limitations
• Denner’s scaling
References
• Wang G, Ma S, Wu Y, Pei J, Zhao R and Shi L (2021) End-to-End Implementation of Various Hybrid Neural Networks on a
Cross-Paradigm Neuromorphic Chip. Front. Neurosci. 15:615279. doi: 10.3389/fnins.2021.615279
• G. Indiveri and S. -C. Liu, "Memory and Information Processing in Neuromorphic Systems," in Proceedings of the IEEE, vol.
103, no. 8, pp. 1379-1397, Aug. 2015, doi: 10.1109/JPROC.2015.2444094.
• Nitin Rathi, Indranil Chakraborty, Adarsh Kosta, Abhronil Sengupta, Aayush Ankit, Priyadarshini Panda, and Kaushik Roy.
2023. Exploring Neuromorphic Computing Based on Spiking Neural Networks: Algorithms to Hardware. ACM Comput.
Surv. 55, 12, Article 243 (December 2023), 49 pages. https://doi.org/10.1145/3571155
• Markovic, Danijela & Mizrahi, Alice & Querlioz, Damien & Grollier, Julie. (2020). Physics for Neuromorphic Computing.
https://www.researchgate.net/publication/339840879_Physics_for_Neuromorphic_Computing
• Durmaz, Taygun & León, Gonzalo & velasco, guillermo. (2020). NEUROMORPHIC COMPUTING Concepts, actors,
applications, market and future trends (Human Brain Project). 10.13140/RG.2.2.33153.33122.

You might also like