“Aggressive Performance Optimizations For 3D Graphics” by Klimovitski, Barad, Macri, Pallister and Haines

  • ©Alex Klimovitski, Haim Barad, Dean Macri, Kim Pallister, and Eric Haines

Conference:


Entry Number: 43

Title:

    Aggressive Performance Optimizations For 3D Graphics

Course Organizer(s):



Presenter(s):



Abstract:


    Prerequisites
    Basic knowledge of C++ and introductory knowledge of platform architecture.

    Topics
    Hardware: platform architecture, basic CPU architecture including SIMD extensions, memory subsystems, graphics cards. Software: data representation and organization and how it affects performance, code optimizations, memory optimizations, SIMD optimizations, high-level algorithmic optimizations.

    Description
    Through hands-on experience with actual performance measurements, analysis, and optimizations using well-established methodologies and techniques, this course imparts a working knowledge of performance optimizations for 3D graphics programs in relationship to algorithmic design, coding, and data representation. Attendees learn tool usage and optimization methodology in a series of exercises on actual, platform-independent code. Approaches include: 3D algorithmic optimizations and restructuring for target platforms, performance analysis with available profiling and event-based tools, coding techniques for performance, and data representation for efficient processing on target platform/CPU architectures.

     

Overview Page:



Type: