Software/hardware design techniques to facilitate technology scaling for memory and data storage devices in data centers
Loading...
Authors
Li, Yin
Issue Date
2018-08
Type
Electronic thesis
Thesis
Thesis
Language
ENG
Keywords
Electrical engineering
Alternative Title
Abstract
To further investigate how to maximize the encoding throughput when data storage systems are protected by both distributed erasure coding (recover data copies from remote redundancy copies) and proposed local erasure coding strategy together. This thesis presents an algorithm that could effectively alleviate Galois Field computation complexity by leveraging the characteristics of erasure encoding/decoding process and the GF calculation acceleration with single instruction, multiple data (SIMD) instructions. A detailed mathematical proof is given and experiments are carried out on a general-propose PC which supports SIMD instructions to evaluate speed performance. The results show that the encoding throughput could increase by more than 50%.
Description
August 2018
School of Engineering
School of Engineering
Full Citation
Publisher
Rensselaer Polytechnic Institute, Troy, NY