Fixed-point calculations

Represent decimals with a documented scale:

fixed-point-scale.gpc

define SCALE = 1000;

int gain = 1250; // 1.250
int input;
int output;

main {
    input = get_ival(PS5_RX);
    output = (input * gain) / SCALE;
    set_val(PS5_RX, bounded_axis(output));
}

function bounded_axis(value) {
    if(value < -100) return -100;
    if(value > 100) return 100;
    return value;
}

Guidelines:

• Multiply before divide to preserve fractional precision.

• Check multiplication for 32-bit overflow.

• State whether negative division/truncation is acceptable.

• Clamp at the boundary where a narrower device range is required.

• Do not repeatedly rescale the same value through several stages.