Mixed-Precision AMG method for Many Core Accelerators
Mixed-Precision AMG method for Many Core Accelerators
复制标题
适用于多核加速器的混合精度 AMG 方法
DOI：
10.1145/2642769.2642794
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
Teruo Tanaka
中科院分区：
文献类型：
--
作者：
Yuki Sumiyoshi;A. Fujii;Akira Nukada;Teruo Tanaka
gpu的单精度和双精度性能之间存在很大的差距。单精度运算速度是gpu上双精度运算速度的两倍以上。然而，单精度算法不能达到足够的精度。通过使用单精度和双精度算法的组合，可以保持精度并快速执行计算。在代数多网格（AMG）方法中引入了一种混合精度方法，并对其在不同gpu上的性能进行了评价。结果表明，对于各向异性问题，当尺寸为1003时，混合精度AMG的性能约为双精度AMG的1.9倍。
There is a large gap between single- and double-precision performances on GPUs. Single-precision arithmetic is more than twice as fast as double-precision arithmetic on GPUs. However, single-precision arithmetic cannot achieve sufficient accuracy. By using a combination of single- and double-precision arithmetics, accuracy is maintained and calculations are performed quickly. We introduce a mixed precision method to the algebraic multigrid (AMG) method and evaluate the performance on various GPUs. The results indicate that, for anisotropic problems, the performance of the mixed-precision AMG is approximately 1.9 times greater than the double-precision AMG at a size of 1003.
影响因子：
2.1
作者：
Stone JE;Gohara D;Shi G
通讯作者：
Shi G