[CppIndia] SIMD and Auto-Vectorization by Mahendra Garodi

In this session, Mahendra talked about different level of parallelism available in modern computer systems. He went into the details of SIMD(Single Instruction Multiple Data) and talked about different methods like using Assembly, C SIMD Intrinsics, Wrapper Classes to achieve better performance using SIMD. He talked in detail about how modern compilers can do the source code analysis and generate SIMD code called as Auto-Vectorization. At the end of the session, he showed live demo on compiler explorer to analyse the code to check whether the compiler is able to vectorize it or not. #simd #vectorization #cppindia Speaker Profile: Name: Mahendra Garodi LinkedIn: https://www.linkedin.com/in/garodimahendra/ Twitter: https://twitter.com/garodimahendra String View Session at CppIndia: https://youtu.be/xJ7Y5zdBi8U Slides: https://github.com/CppIndia-UserGroup/CppIndia-SessionDocuments/tree/master/2021/05-22 Index for quick Access: 00:00 Introduction 04:00 Classes of Parallelism 07:00 SIMD - Architecture Overview 16:52 SIMD - Assembly 19:46 SIMD - C Intrinsics 22:05 SIMD - Explicit Vectorization Libraries 26:53 SIMD - Auto-Vectorization 48:24 Live Code Analysis for Auto-Vectorization 01:07:36 Q & A This video is a live recording of the online conference held on 22-May-2021. To join this community, please fill this form. * Registration form: https://forms.gle/N4TZwqPW24kY1y4WA * Webpage: www.cppindia.co.in * LinkedIn: https://www.linkedin.com/company/cppindia/ * Discord: https://discord.gg/D6a2jNh * Email: [email protected]