Controls the use of special floating point instructions to improve performance.

This pragma is compatible with the following platform targets:

68K

PowerPC

NEC V800

Intel x86

MIPS

Embedded 68K

This pragma applies to PowerPC programming only.

If this pragma is on, the compiler uses such PowerPC instructions as FMADD, FMSUB, and FNMAD to speed up floating-point computations. However, certain computations give unexpected results when this pragma is on. For example:

register double A, B, C, D, Y, Z;
register double T1, T2;

A = C = 2.0e23;
B = D = 3.0e23;

Y = (A * B) - (C * D);
printf("Y = %f\n", Y);
/* prints 2126770058756096187563369299968.000000 */

 

T1 = (A * B);
T2 = (C * D);
Z = T1 - T2;
printf("Z = %f\n", Z);  /* prints 0.000000 */

When this option is off, Y and Z have the same value.

This pragma corresponds to the Use FMADD & FMSUB option in the PowerPC Processor settings panel. To check whether this option is on, use __option (fp_contract), described in "Checking Options."