ARM ® NEON™ is an advanced SIMD architecture for ARM processors which implement the ARMv7-A or ARMv7-R profile. Through the NEON instruction set, a bank of sixteen 128-bit registers is made available for SIMD processing. The instruction set includes typical SIMD operations for moving data between NEON and general purpose registers as well as data processing and type conversion. Effective use of NEON through hand-coded assembly, intrinsic functions, or automatic vectorization by the compiler can lead to tremendous performance gains for multimedia applications.
Qualcomm’s SIMD/FPU co-processor is compliant with ARM’s NEON instruction set. Most of the fluid effects you see in this video are being done using this component.
We are in the process of putting together a whitepaper on NEON optimizations so check back soon. Also at Qualcomm’s Uplinq conference this year (June 30 – July 1 in San Diego) we will have a session on how to do vertex skinning using NEON optimizations, so look for GPU performance tuning sessions at the conference.