Session Name: | Simpler Better Faster Vector |
Speaker(s): | Quentin Froemke, Stan Melax, Ganesh Rao |
Company Name(s): | Intel Corporation, Intel, Intel Software Development Products |
Track / Format: | Programming |
Overview: | You have just implemented a sensational game system but it runs slower than you would like. Because of data dependencies and strict time budgets, a logical step for making video game code run faster is to get the most out of each core. This intermediate class teaches optimizations and fine-grain vector parallelism (SIMD) concepts for current and upcoming x86 CPUs. Building on this foundation, we will demonstrate the acceleration of various algorithms relevant to gaming. You'll learn best practices for resolving performance hotspots, with minimal impact to the source code and game experience. This session includes a review of the new automatic vectorization features in the upcoming Intel C++ Compiler version 12. |