SIMD (Single Instruction Multiple Data) technology plays an increasingly important role in modern CPUs. When optimizing code the first efforts should be focused on vectorization, given the potential performance improvements that this technology offers. Multithreaded applications will benefit from the multiplicative effect of using multiple cores and efficiently vectorized loops.
Intel® Advanced Vector Extensions 512 (Intel® AVX-512) ISA offers vector operations that can operate on vector registers 512 bits wide. Combined with new instructions and twice the number of registers, this represents an increase in peak FLOPS compared to Intel® Advanced Vector Extensions 2.0 (Intel® AVX2).
In this webinar, an introduction to vectorization techniques and vector performance analysis will be presented. Specific support for vector-level parallelism offered by the Intel® AVX-512 ISA will be included. Material presented in this webinar is introductory and it is intended as a basis for further exploration in vectorization techniques and software development using Intel® AVX-512.
Required Fields(*)
Alberto is a software engineer at Intel’s software and services group. His current work is focused on working with developers to let their applications take advantage of parallel architectures. Alberto has worked on software optimization and algorithm development/analysis in the energy industry for 15 years. Alberto joined Intel in 2016. He holds an MSc in Mathematics and Computer Science and an MSc in Geophysics (Colorado School of Mines, USA).