The Performance Analysis Of Arm Neon Technology For Mobile Platforms

Mobile platforms are required to execute efficient applications that satisfy the limited environmental factors of embedded systems. The ARM NEON technology based SIMD is a media processing architecture that adds instructions targeted primarily at audio, video, and 3-D graphics processing. The ARM NEON technology was designed for programming applications efficiently on Mobile platforms.In this paper, the NEON SIMD technology that supported in ARM Cortex-A8 processor is tested and evaluated. Especially, Auto-vectorization which is one of methods to use NEON SIMD technology is used for experiments on several open source applications. Program execution image size, execution time and power consumption which are important in Embedded System are measured to compare binary image size, execution time per data size and CPU Cycle Count. Performance evaluation results show that three performance factors, real time, power consumption and program size are important to impact the total performance when the NEON technology is adapted of mobile platforms. However, it is uncommon for experimental result to improve all of performance factors. In other words, the performance evaluation results show that it should be estimated by considering all of three performance factors.