12 Fixed-Point Arithmetic Support

The MPLAB XC32 C/C++ Compiler supports fixed-point arithmetic. This, according to ISO/IEC TR 18037:2008//the N1169 draft of ISO/IEC TR 18037, the ISO C99 technical report on Embedded C. It is available at [via]: http://www.open-std.org/JTC1/SC22/WG14/www/projects#18037. or standards.iso.org/ittf/PubliclyAvailableStandards/c051126_ISO_IEC_TR_18037_2008.zip

This chapter describes the implementation-specific details of the types and operations supported by the compiler under this draft standard.

Because of the sensitivity of DSP applications to performance, application developers have historically tended to write functions in assembly. However, the XC32 compiler reduces, and may even eliminate, the need to write assembly code. This chapter describes coding styles and usage tips that can help you to obtain the best optimizations for your DSP application.

Several Microchip PIC32 MCUs feature a DSP-enhanced core with four 64-bit accumulators. The DSP-enhanced core also provides a set of new instructions and a new architectural state, with computational support for fractional data types, SIMD (Single Instruction, Multiple Data), saturation, and other operations commonly used in DSP applications.