As presented at the conference, many users from different fields are deeply involved in the KEEP program – from traditional high-performance computing areas, such as oceanographic simulation, fluid mechanics computing, astronomical data processing and oil exploration, to new emerging fields such as HPDA – a combination of high-performance computing, big data, and intelligent computing, etc.
Several KEEP program user experiences show that KNL is not only the most powerful many-core computer processor, it is also programming-friendly. Additionally, the basic performance reaches at least twice that of a dual-socket E5-2680 v4 CPU even without code level deep optimization. There is even more room to improve the performance when code-level deep optimization is performed specifically for the structural characteristics, or for the application algorithms of the KNL.
KNL will continue to greatly improve the application of high-performance computing, deep learning, intelligent computing and big data applications, and a growing number of future applications will apply KNL to achieve better performance. The large number of realistic and reliable HPC application optimization experiences gathered via the KEEP program will undoubtedly contribute to the acceleration of the broader application and popularization of KNL.
To view the original version on PR Newswire, visit:http://www.prnewswire.com/news-releases/keep-program-proves-to-boost-application-performance-for-knl-300365694.html
SOURCE Inspur Electronic Information Industry Co., Ltd