计算机科学
压缩比
数据压缩
压缩(物理)
有损压缩
气体压缩机
无损压缩
数据传输
炸薯条
基础(拓扑)
数据压缩比
架空(工程)
计算机硬件
并行计算
图像压缩
算法
工程类
材料科学
数学
内燃机
图像(数学)
汽车工程
复合材料
人工智能
数学分析
操作系统
机械工程
电信
图像处理
作者
G Vivekananda,P Thejaswini
标识
DOI:10.1109/indicon52576.2021.9691570
摘要
Data compression techniques in hardware are used for increasing system efficiency by effective on-chip storage and reduction in transmission time through on-chip interconnects. We exploit the inherent limitations of conventional base delta compression and propose an extension to how the base values are calculated. We propose a cost-effective compression technique called Base Delta Extended (BDE) that is best suited for time critical and energy critical applications. In BDE, the base value is not fixed, instead computed for each iteration based on the previously compressed group values. Our base calculation involves the right usage of the history of delta values and their relative variation from peer data elements. We use a small block size for compression, coupled with multiple parallel blocks to speed up the compression and decompression time. Our proposed technique achieves a fairly good compression ratio with minimum time to compress and decompress. With the help of grouping and subsequent parallel compressor blocks, we achieve compression faster. We experimentally prove the effectiveness of our technique in terms of compression ratio as well as hardware and timing overhead over a wide range of input data.
科研通智能强力驱动
Strongly Powered by AbleSci AI