flat assembler
Message board for the users of flat assembler.

Index > Macroinstructions > No Specific Register Macros

Author
Thread Post new topic Reply to topic
aq83326



Joined: 25 Jun 2011
Posts: 21
aq83326
I'm slowly gaining on my macro-fu.

I know performance usually means speed, but below is a saving of space, putting a partially-filled 2D array (used to represent a graph) into a 1D array. Also, one could use an array of arrays and still save most of the space, but this also puts the whole array into a single contiguous block.

Looking at HLL's today, it really makes me wish early scientists had made better macro systems rather than HLL's.
Anyway, I finally realized we could even write some macros that work across platforms.
I made the code below after translating it from C#; it was like an epiphany "I don't have to use any particular register!"
With macros you have to keep track of which registers they use, so the ones below are specified as arguments, allowing for efficient register allocation. Simultaneous high level/low level just seems like a neat way to program.
Code:
; Copyright 2012 Aaron Ogden
; released under BSD license

; this file contains macros for computing indices into a half-filled array,
; for instance, if you wished to represent an edge's value as the value in a two dimensional array where the indices are the possibly connected nodes
; if the edges are bidirectional, nearly one half of a two-dimensional array would be not used or filled with duplicate data
; this code computes the index into a one-dimensional array that corresponds to such a 2D array being linearized.
; 1D index
; [0]      [1]      [2]      [3]      [4]      [5]
; 2D indices
; [0][0]   [0][1]   [0][2]   [1][1]   [1][2]   [2][2]


; it also contains a macro to compute the summation of integers from 1-N

; summation 1-N macro, parameters should be register names, result supplied and returned in N, N1 is overwritten
macro summation1N N,N1
{
      mov N1, N
   add N, 1
    mul N, N1
   shr N,1         ; equivalent to dividing by 2, safe here because N is always an even integer at this point
}

; parameters are operand names
; t1, t2, and retval are overwritten
; postcondition: X <= Y
; size is the number of nodes
; X is one node, Y is the other
; retval is for the return value, the 1D index where [X][Y] is found
; t1 and t2 are temps
macro FindHalfIndex size,X,Y,t1,t2,retval
{
    cmp X,Y                 ; compare X and Y
   jle .fhipostswap        ; skip if X is less than or equal to Y
      xchg X,Y                ; swap X & Y if X is greater than Y
     .fhipostswap:
   mov t1, size            ; t1 now holds the size
     sub t1, X               ; t1 now holds numtier
      mov t2, size            ; t2 now holds the size
     summation1N t2,retval   ; t2 now holds the summation of size
        summation1N t1,retval   ; t1 now holds the summation of numtier
     sub t2,t1               ; t2 now holds the number of elts preceding
 mov t1,Y                ; t1 now holds the Y
        sub t1,X                ; t1 now holds the offset
   mov retval,t1           ; retval now holds the offset
       add retval,t2           ; retval now holds the answer
}    
Post 15 Jun 2012, 06:04
View user's profile Send private message Visit poster's website Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  


< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on YouTube, Twitter.

Website powered by rwasa.