ASM Community

I would like to work with arrays (math arrays), example: „¡„Ÿ „Ÿ„¢ „¡„Ÿ „Ÿ„¢ „  11 12 13 „  „  11 12 13 „  A = „  21 22 23 „  B = „  21 22 23 „  „  31 32 33 „  „  31 32 33 „  „¤„Ÿ „Ÿ„£ „¤„Ÿ „Ÿ„£ C = A x B What's one good way? Thanks all.

This is the topic of my current tutorial, its almost finished, but seeing as you want some help now I'll post some code. I don't have time at the moment but when I post the tutorial tomorrow there'll be a full explanation. I assume you want this matrix maths for your 3D program. Heres a procedure that will multiply 4*4 matrices, easily modified to 3*3 if thats what you need. Note this be not be the best way. Also it assumes the matrices are stored as follows

       A  B  C  D        [00] [16] [32] [48]
       E  F  G  H   - \  [04] [20] [36] [52]
       I  J  K  L   - /  [08] [24] [40] [56]
       M  N  O  P        [12] [28] [44] [60]

MulMatrix Proc pMatrix1:DWORD, pMatrix2:DWORD
push edi
push esi
mov edi, pMatrix1
mov esi, pMatrix2
mov ebx, 4	
NxtRow:
	mov ecx, 4
	NxtCol:	
		fld  real4 ptr 
		fmul real4 ptr 
		fld  real4 ptr [16]
		fmul real4 ptr [4]
		fadd
		fld  real4 ptr [32]
		fmul real4 ptr [8]
		fadd
		fld  real4 ptr [48]
		fmul real4 ptr [12]
		fadd		
		add edi, 16
	dec ecx
	jnz NxtCol

	fstp real4 ptr [48]
	fstp real4 ptr [32]
	fstp real4 ptr [16]
	fstp real4 ptr 
	sub edi, 64
	add esi, 4
dec ebx
jnz NxtRow
pop esi
pop edi
ret
MulMatrix EndP

This code premultiplies Matrix2 by Matrix1 over writing Matrix2 While linear code would be quicker, a loop is easier to modify. Besides with registers as counters and branch prediction on CPUs this won't slow at all. This message was edited by Zadkiel, on 4/13/2001 1:45:49 PM