File:Apple WWDC 2002 Session 102 - Performance Optimization with Velocity Engine

Apple offers supercomputing performance to both consumers and professionals in its entire line of desktop computers with Velocity Engine and the PowerPC G4 processor. Processor-bound operations such as memory copies, string compares, and page clears can be offloaded to the vector engines of the G4. This session provides an overview of the Velocity Engine vector libraries, including recently optimized double-precision FFTs, MDCT, and BLAST. Apple's numerics engineers will discuss techniques for general vectorization and for identifying code that can be vectorized. CHUD is also demonstrated.

Presenters: - Ali Sazegari, Vector and Numerics Group - Ralph Brunner, Core Graphics - Robert Murley, Vector and Numerics Group - Bill Kincaid, iTunes Group - Nathan Slingerland and Sanjay Patel, Architecture Performance Group