Polar Complex Number Subtraction

Validation and Benchmark Tool: PolSubBV140.zip

Table of results: CmplxPolSubBenchmark120.xls (Old, needs an update!)

Description

The objective is to build the fastest PolSub routines.

There are three overloaded functions differing in the input and output datatype.

function PolSub(const X, Y : TPolarComplexE) : TPolarComplexE; overload;
function PolSub(const X, Y : TPolarComplexD) : TPolarComplexD; overload;
function PolSub(const X, Y : TPolarComplexS) : TPolarComplexS; overload;

Type definition

TRegtangularComplexS = packed record
 RealPart, ImaginaryPart : Single;
end;

TRegtangularComplexD = packed record
 RealPart, ImaginaryPart : Double;
end;

TRegtangularComplexE = packed record
 RealPart : Extended;
 Pad1 : Cardinal; //4 byte
 Pad2 : SmallInt; //2 byte
 ImaginaryPart : Extended;
 Pad3 : Cardinal; //4 byte
 Pad4 : SmallInt; //2 byte
end;

TPolarComplexS = packed record
 Amplitude, Phase : Single;
end;

TPolarComplexD = packed record
 Amplitude, Phase : Double;
end;

TPolarComplexE = packed record
 Amplitude : Extended;
 Pad1 : Cardinal; //4 byte
 Pad2 : SmallInt; //2 byte
 Phase : Extended;
 Pad3 : Cardinal; //4 byte
 Pad4 : SmallInt; //2 byte
end;