ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper A Signal Integrity Analysis for the Design of a Computer Mainboard
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Kyong Hee Lee, Hyuk Je Kwon, Young Woo Kim, Seon Young Kim, Yoo Mi Park
Issue Date
2023-10
Citation
International Conference on Consumer Electronics (ICCE) 2023 : Asia, pp.380-382
Publisher
IEEE
Language
English
Type
Conference Paper
Abstract
As artificial intelligence technology has recently advanced, various learning methods are being developed to make training deep learning models faster. Multiple GPUs can be connected to a mainboard equipped with a multi-core, highperformance CPU to increase learning and calculation speed. Additionally, a multi-node hardware platform consisting of multiple mainboards can be utilized for parallel processing.In this paper, power integrity and signal integrity are considered when designing a mainboard that processes high-speed data signals. These show the power distributed network noise, signal impedance, crosstalk, and high-speed data signal characteristics that occur in the circuit and PCB. To manufacture a mainboard using a multi-layer PCB, we present insertion loss and eye diagram characteristic simulation results among signal integrity analysis factors. Such power integrity and signal integrity methods will be beneficial in reducing errors and predicting basic characteristics when designing and manufacturing hardware for artificial intelligence technology.
KSP Keywords
Artificial intelligence technology, Basic characteristics, Eye diagram, Hardware platform, High Speed, Learning methods, Multi-node, Multilayer PCB, Multiple GPUs, Network noise, Parallel Processing