Author Topic: my routine clears ram (Read 3493 times)

ben_g · « **on:** May 11, 2011, 03:56:58 pm »

Hi,

I've got an other problem: my routine clears the ram, and I don't know why...
from 'learn ti83plus asm in 28 Days', i know that ram clears usually appear when you used the stack incorrect, but I've counted all the pushes and pops of the routine, and there were as much pushes as pops. Further is the mem for all the vars stored in the program itself, so it can't be becouse i'm editing mem that shouldn't be edited.

It's just calculating with registers and vars, but i don't know how that can clear the ram.

anyway, here's my code

Code: [Select]

3DTo2D:
  ld a, 0
  ld (valid), a
  ld bc, (xpoint)
  ld de, (xfrom)
  call SubFP
  ld (xpoint), de
  ld bc, (ypoint)
  ld de, (yfrom)
  call SubFP
  ld (ypoint), de
  ld bc, (zpoint)
  ld de, (zfrom)
  call SubFP
  ld (zpoint), de
  ld hl, (xpoint)
  ld bc, (xto)
  call MulFP
  push de			;1
  ld hl, (ypoint)
  ld bc, (yto)
  call MulFP
  pop hl			;0
  add hl, de
  push hl			;1
  ld hl, (zpoint)
  ld bc, (zto)
  call MulFP
  pop hl			;0
  add hl, de
  push hl			;1
  ld a, l
  ld c, a
  ld a, (min_range)
  ld b, a
  ld a, c
  cp b
  jp c, Invalid
  ld c, a
  ld a, (max_range)
  ld b, a
  ld a, c
  cp b
  jp nc, Invalid
  ld a, 1
  ld (valid), a
  ld hl, (xpoint)
  ld bc, (xcross)
  call MulFP
  push de			;2
  ld hl, (ypoint)
  ld bc, (ycross)
  call MulFP
  pop hl			;1
  add hl, de
  push hl			;2
  ld hl, (zpoint)
  ld bc, (zcross)
  call MulFP
  pop hl			;1
  add hl, de
  pop de			;0
  push de			;1
  call DivFP
  ld (screenx), hl
  ld hl, (xpoint)
  ld bc, (xup)
  call MulFP
  push de			;2
  LD HL, (ypoint)
  ld bc, (yup)
  call MulFP
  pop hl			;1
  add hl, de
  push hl			;2
  ld hl, (zpoint)
  ld bc, (zup)
  call MulFP
  pop hl			;1
  add hl, bc
  pop de			;0
  call DivFP
  ld (screeny), hl
  ld hl, $0100
  ld de, (screenx)
  add hl, de
  ld bc, $5F00
  call MulFP
  ld (screenx), de
  ld hl, $0100
  ld de, (screeny)
  add hl, de
  ld bc, $4100
  call MulFP
  ld (screeny), de
Invalid:
  ret

btw: the commented numbers show the amount of items put on the stack by this routine.

AngelFish · « **Reply #1 on:** May 11, 2011, 04:11:29 pm »

Where are those calls going to?

ben_g · « **Reply #2 on:** May 11, 2011, 04:25:16 pm »

Here's the full source code

AngelFish · « **Reply #3 on:** May 11, 2011, 04:48:36 pm »

I just made a few minor changes to the source so that it would assemble and WabbitEmu runs it fine. Not sure where your problem is, because I'm not getting a RAM clear.

Someone more experienced than myself might be able to identify it, though.

thepenguin77 · « **Reply #4 on:** May 11, 2011, 06:18:29 pm »

Well, the reason that the full program isn't crashing is because it never actually calls the routine in question. I just does clearGBuf, lineDraw, fastCopy and checks for quit.

I would call it myself, but I have no idea what to call.

ben_g · « **Reply #5 on:** May 12, 2011, 03:18:47 pm »

just add 'call 3DTo2D' somewere on a part that is executed. You will get a ram clear

if you add 'call InitView' somewere, it doesn't couse a ram clear, but i don't know if it's working corectely as it should work together with 3DTo2D.

have written a simple 3D engine in GML and i try to rewrite it in asm. Maybe the GML code can help:

Code: (d3dto2d_init) [Select]

// d3dto2d_init(xfrom,yfrom,zfrom,xto,yto,zto,xup,yup,zup,angle,aspect,znear,zfar)
var d;

global.d3dto2d_xfrom = argument0;
global.d3dto2d_yfrom = argument1;
global.d3dto2d_zfrom = argument2;

global.d3dto2d_xto = argument3-global.d3dto2d_xfrom;
global.d3dto2d_yto = argument4-global.d3dto2d_yfrom;
global.d3dto2d_zto = argument5-global.d3dto2d_zfrom;
d = sqrt(global.d3dto2d_xto*global.d3dto2d_xto+global.d3dto2d_yto*global.d3dto2d_yto+global.d3dto2d_zto*global.d3dto2d_zto);
global.d3dto2d_xto /= d;
global.d3dto2d_yto /= d;
global.d3dto2d_zto /= d;

global.d3dto2d_xup = argument6;
global.d3dto2d_yup = argument7;
global.d3dto2d_zup = argument8;
d = global.d3dto2d_xup*global.d3dto2d_xto+global.d3dto2d_yup*global.d3dto2d_yto+global.d3dto2d_zup*global.d3dto2d_zto;
global.d3dto2d_xup -= d*global.d3dto2d_xto;
global.d3dto2d_yup -= d*global.d3dto2d_yto;
global.d3dto2d_zup -= d*global.d3dto2d_zto;
d = tan(degtorad(argument9)/2)*sqrt(global.d3dto2d_xup*global.d3dto2d_xup+global.d3dto2d_yup*global.d3dto2d_yup+global.d3dto2d_zup*global.d3dto2d_zup);
global.d3dto2d_xup /= d;
global.d3dto2d_yup /= d;
global.d3dto2d_zup /= d;

global.d3dto2d_xcross = (global.d3dto2d_yup*global.d3dto2d_zto-global.d3dto2d_zup*global.d3dto2d_yto)/argument10;
global.d3dto2d_ycross = (global.d3dto2d_zup*global.d3dto2d_xto-global.d3dto2d_xup*global.d3dto2d_zto)/argument10;
global.d3dto2d_zcross = (global.d3dto2d_xup*global.d3dto2d_yto-global.d3dto2d_yup*global.d3dto2d_xto)/argument10;

global.d3dto2d_znear = argument11;
global.d3dto2d_zfar = argument12;

The 'aspect' variable won't be ust in the asm version.

Code: (d3dto2dpoint) [Select]

// d3dto2d_point(x,y,z)
var xx,yy,zz,d;
xx = argument0-global.d3dto2d_xfrom;
yy = argument1-global.d3dto2d_yfrom;
zz = argument2-global.d3dto2d_zfrom;
d= xx*global.d3dto2d_xto+yy*global.d3dto2d_yto+zz*global.d3dto2d_zto;
if (d<global.d3dto2d_znear or d>global.d3dto2d_zfar){ Return = 0}else{Return = 1}
global.xx = (xx*global.d3dto2d_xcross+yy*global.d3dto2d_ycross+zz*global.d3dto2d_zcross)/d;
global.yy = (xx*global.d3dto2d_xup+yy*global.d3dto2d_yup+zz*global.d3dto2d_zup)/d;

return Return;

ben_g · « **Reply #6 on:** May 16, 2011, 03:49:41 pm »

All of the calls made in the 3DTo2D part are also used in the InitView part. The number of pushes and pops are the same, and the vars are defined in the program memory, so I still don't know why it clears the ram.

Maybe my own math routines aren't really doing what they should do, so does enybody here have some code on their hard drive for 16bit 8.8 fixed point signed addition, substraction, multiplication and division?

EDIT: I double-checked all my routines, then tripple-checked them, then tested every routine on it's own, and I have found out that the code below clears the ram, but I still don't know why

Code: [Select]

3DTo2D:
  ld a, 0
  ld (valid), a
  ld bc, (xpoint)
  ld de, (xfrom)
  call SubFP
  ld (xpoint), de
  ld bc, (ypoint)
  ld de, (yfrom)
  call SubFP
  ld (ypoint), de
  ld bc, (zpoint)
  ld de, (zfrom)
  call SubFP
  ld (zpoint), de
  ld hl, (xpoint)
  ld bc, (xto)
  call MulFP
  push de			;1
  ld hl, (ypoint)
  ld bc, (yto)
  call MulFP
  pop hl			;0
  add hl, de
  push hl			;1
  ld hl, (zpoint)
  ld bc, (zto)
  call MulFP
  pop hl			;0
  add hl, de
  push hl			;1
  ld a, l
  ld c, a
  ld a, (min_range)
  ld b, a
  ld a, c
  cp b
  ret c
  ld c, a
  ld a, (max_range)
  ld b, a
  ld a, c
  cp b
  ret nc
  ld a, 1
  ld (valid), a
  ld hl, (xpoint)
  ld bc, (xcross)
  call MulFP
  push de			;2
  ld hl, (ypoint)
  ld bc, (ycross)
  call MulFP
  pop hl			;1
  add hl, de
  push hl			;2
  ld hl, (zpoint)
  ld bc, (zcross)
  call MulFP
  pop hl			;1
  add hl, de
  pop de			;0
  push de			;1
  call DivFP
  ld (screenx), hl
  ld hl, (xpoint)
  ld bc, (xup)
  call MulFP
  push de			;2
  LD HL, (ypoint)
  ld bc, (yup)
  call MulFP
  pop hl			;1
  add hl, de
  push hl			;2
  ld hl, (zpoint)
  ld bc, (zup)
  call MulFP
  pop hl			;1
  add hl, bc
  pop de			;0
  call DivFP
  ld (screeny), hl
  ld hl, $0100
  ld de, (screenx)
  add hl, de
  ld bc, $5F00
  call MulFP
  ld (screenx), de
  ld hl, $0100
  ld de, (screeny)
  add hl, de
  ld bc, $4100
  call MulFP
  ld (screeny), de
  ret

I've tested every subroutine, and they all worked normally. The stack is also used correctelly (I think), So why is it clearing my ram?

Author Topic: my routine clears ram (Read 3493 times)

ben_g

my routine clears ram

AngelFish

Re: my routine clears ram

ben_g

Re: my routine clears ram

AngelFish

Re: my routine clears ram

thepenguin77

Re: my routine clears ram

ben_g

Re: my routine clears ram

ben_g

Re: my routine clears ram