SDRBench

SDRBench是由阿贡国家实验室等多个机构合作创建的标准压缩评估基准，包含超过10个来自不同科学领域的真实世界科学数据集。数据集涵盖气候模拟、宇宙学粒子模拟、湍流模拟等多个领域，每个数据集都提供了详细的描述、形状、数据类型和数据大小。SDRBench旨在通过集成多种先进的损失有和无损压缩器，支持研究人员和用户公平、方便地评估和比较压缩器。此外，SDRBench还总结了关键的压缩质量评估指标，如压缩/解压缩速度、压缩比、峰值信噪比等，以帮助深入理解损失压缩器。

SDRBench is a standard compression evaluation benchmark co-created by Argonne National Laboratory and multiple other institutions, comprising over 10 real-world scientific datasets from diverse scientific domains. These datasets cover multiple fields including climate simulation, cosmological particle simulation, and turbulence simulation, with each dataset providing detailed descriptions, shapes, data types, and data sizes. SDRBench aims to support researchers and users in fairly and conveniently evaluating and comparing compressors by integrating a variety of advanced lossy and lossless compressors. In addition, SDRBench also summarizes key compression quality evaluation metrics such as compression/decompression speed, compression ratio, peak signal-to-noise ratio (PSNR), etc., to facilitate in-depth understanding of lossy compressors.

搜集汇总

背景与挑战

背景概述

SDRBench是一个科学数据缩减基准测试数据集集合，涵盖气候模拟、分子动力学、天气模拟等多个科学领域，旨在为数据压缩技术的研究和评估提供标准化的参考数据集和工具链。该数据集包括详细的元数据、压缩示例和评估指标，支持用户和开发者进行有损和无损压缩算法的性能比较与优化。