# Difference between revisions of "Fast signed multiply"

m (cat) |
|||

Line 1: | Line 1: | ||

The following code multiplies two 8-bit signed integers, of range (-64..+64). The output is a signed 16-bit result, in range (-4096..4096). It uses a 256 byte lokup table, but is quite fast. | The following code multiplies two 8-bit signed integers, of range (-64..+64). The output is a signed 16-bit result, in range (-4096..4096). It uses a 256 byte lokup table, but is quite fast. | ||

− | <source lang=" | + | <source lang="6502"> |

;Fast table driven signed multiply routine | ;Fast table driven signed multiply routine | ||

;It multiplies Y with A, and relies on the mathematical fact (a+b)^2-(a-b)^2 = 4*a*b | ;It multiplies Y with A, and relies on the mathematical fact (a+b)^2-(a-b)^2 = 4*a*b |

## Revision as of 07:23, 25 August 2013

The following code multiplies two 8-bit signed integers, of range (-64..+64). The output is a signed 16-bit result, in range (-4096..4096). It uses a 256 byte lokup table, but is quite fast.

<source lang="6502">

- Fast table driven signed multiply routine
- It multiplies Y with A, and relies on the mathematical fact (a+b)^2-(a-b)^2 = 4*a*b
- Also both inputs and outputs are SIGNED and are never supposed to be outside of the range (-64; +64)
- Return with YA = 4*A*Y (result can be shifted as necessary)

Multiply

sty Temp

pha clc adc Temp ;Add two multiplicands together bpl + eor #$ff clc adc #$01 ;If result is negative, force it to be positive + asl A tax ;The square of a negative number is equal to its pos counterpart anyways

pla sec sbc Temp ;Compute difference of two multiplicands bpl + eor #$ff ;Again, force the result to be positive clc adc #$01 + asl A tay

lda SquareTbl,X sec sbc SquareTbl,Y ;Compute (a+y)^2-(a-y)^2 pha lda SquareTbl+1,X sbc SquareTbl+1,Y

tay pla

rts

- This is a square table for the multiply routine
- Note that this macro is for WLA-DX, but can be adapted
- For most assemblers

SquareTbl .def i = 0 .rept 128 .dw i^2 ;Create a square table with 128 word entries .redef i i+1 .endr .ends </source>