kolibrios/programs/demos/firework/trunk/firework.asm
Kirill Lipatov (Leency) 6a00fc624a compress animage graphics, webview fix
git-svn-id: svn://kolibrios.org@7754 a494cfbc-eb01-0410-851d-a64ba20cac60
2020-03-26 02:22:41 +00:00

451 lines
11 KiB
NASM

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; A firework demo
; Programmed by Yaniv LEVIATHAN
; http://yaniv.leviathanonline.com
; Converted to DexOS, By Dex
; Converted to KolibriOS, By Asper
; Optimized for KolibriOS, By Diamond
; Assemble with
; c:fasm firework.asm firework.kex
; NOTE: Needs MMX & SSE,
; optionally AVX, AVX2, AVX512
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
use32
org 0x0
db 'MENUET01' ; 8 byte id
dd 0x01 ; version
dd STARTAPP ; program start
dd I_END ; program image size
dd E_END ; required amount of memory
dd stacktop ; reserved=no extended header
dd 0, 0
include '../../../macros.inc'
SCREEN_WIDTH = 320
SCREEN_HEIGHT = 200
SIMD equ SSE
SIMD_BYTES = 8
; SSE 8
; AVX 16
; AVX2 32
; AVX512 64
assert SCREEN_WIDTH mod SIMD_BYTES = 0
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Global defines
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
NUM_PARTS = 150
X_OFFSET = 0
Y_OFFSET = 4
X_SPEED_OFFSET = 8
Y_SPEED_OFFSET = 12
COLOR_OFFSET = 16
PART_SIZE = 20
macro shade
{
local .lop
if SIMD eq SSE
mov ecx, SCREEN_WIDTH * SCREEN_HEIGHT / SIMD_BYTES
mov edi, buffer
movq mm1, qword [sub_mask]
.lop:
movq mm0, [edi]
psubusb mm0, mm1
movq [edi], mm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX
mov ecx, SCREEN_WIDTH * SCREEN_HEIGHT / SIMD_BYTES
mov edi, buffer
vmovdqa xmm1, xword [sub_mask]
.lop:
vmovdqa xmm0, [edi]
vpsubusb xmm0, xmm0, xmm1
vmovdqa [edi], xmm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX2
mov ecx, SCREEN_WIDTH * SCREEN_HEIGHT / SIMD_BYTES
mov edi, buffer
vmovdqa ymm1, yword [sub_mask]
.lop:
vmovdqa ymm0, [edi]
vpsubusb ymm0, ymm0, ymm1
vmovdqa [edi], ymm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX512
mov ecx, SCREEN_WIDTH * SCREEN_HEIGHT / SIMD_BYTES
mov edi, buffer
vmovdqa64 zmm1, zword [sub_mask]
.lop:
vmovdqa64 zmm0, [edi]
vpsubusb zmm0, zmm0, zmm1
vmovdqa64 [edi], zmm0
add edi, SIMD_BYTES
loop .lop
end if
}
macro blur_prepare
{
mov ecx, (SCREEN_WIDTH * SCREEN_HEIGHT - SCREEN_WIDTH * 2 - SIMD_BYTES*2) / SIMD_BYTES
mov edi, buffer + SCREEN_WIDTH + SIMD_BYTES
}
macro blur
{
local .lop
if SIMD eq SSE
.lop:
movq mm0, [edi]
movq mm1, [edi + 1]
movq mm2, [edi - 1]
movq mm3, mm0
movq mm4, [edi - SCREEN_WIDTH]
movq mm5, [edi + SCREEN_WIDTH]
pavgb mm0, mm1 ; mm0 = avg(cur,cur+1)
pavgb mm3, mm2 ; mm3 = avg(cur,cur-1)
pavgb mm4, mm5 ; mm4 = avg(cur+width,cur-width)
pavgb mm3, mm4 ; mm3 = avg(avg(cur,cur-1),avg(cur+width,cur-width))
pavgb mm0, mm3 ; mm0 = avg(avg(cur,cur+1),
movq [edi], mm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX
.lop:
vmovdqa xmm0, [edi]
vmovdqa xmm1, xmm0
vmovdqa xmm2, [edi - SCREEN_WIDTH]
vpavgb xmm0, xmm0, [edi + 1]
vpavgb xmm1, xmm1, [edi - 1]
vpavgb xmm2, xmm2, [edi + SCREEN_WIDTH]
vpavgb xmm1, xmm1, xmm2
vpavgb xmm0, xmm0, xmm1
vmovdqa [edi], xmm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX2
.lop:
vmovdqa ymm0, [edi]
vmovdqa ymm1, ymm0
vmovdqa ymm2, [edi - SCREEN_WIDTH]
vpavgb ymm0, ymm0, [edi + 1]
vpavgb ymm1, ymm1, [edi - 1]
vpavgb ymm2, ymm2, [edi + SCREEN_WIDTH]
vpavgb ymm1, ymm1, ymm2
vpavgb ymm0, ymm0, ymm1
vmovdqa [edi], ymm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX512
.lop:
vmovdqa64 zmm0, [edi]
vmovdqa64 zmm1, zmm0
vmovdqa64 zmm2, [edi - SCREEN_WIDTH]
vpavgb zmm0, zmm0, [edi + 1]
vpavgb zmm1, zmm1, [edi - 1]
vpavgb zmm2, zmm2, [edi + SCREEN_WIDTH]
vpavgb zmm1, zmm1, zmm2
vpavgb zmm0, zmm0, zmm1
vmovdqa64 [edi], zmm0
add edi, SIMD_BYTES
loop .lop
end if
}
macro blur_right
{
local .lop
if SIMD eq SSE
.lop:
movq mm0, [edi]
movq mm1, [edi + 1]
movq mm2, [edi + SCREEN_WIDTH]
movq mm3, [edi + SCREEN_WIDTH + 1]
pavgb mm0, mm1
pavgb mm3, mm2
pavgb mm0, mm3
movq [edi], mm0
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX
.lop:
vmovdqa xmm0, [edi]
vmovdqu xmm1, [edi + SCREEN_WIDTH + 1]
vpavgb xmm2, xmm0, [edi + 1]
vpavgb xmm3, xmm1, [edi + SCREEN_WIDTH]
vpavgb xmm4, xmm2, xmm3
vmovdqa [edi], xmm4
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX2
.lop:
vmovdqa ymm0, [edi]
vmovdqu ymm1, [edi + SCREEN_WIDTH + 1]
vpavgb ymm2, ymm0, [edi + 1]
vpavgb ymm3, ymm1, [edi + SCREEN_WIDTH]
vpavgb ymm4, ymm2, ymm3
vmovdqa [edi], ymm4
add edi, SIMD_BYTES
loop .lop
else if SIMD eq AVX512
.lop:
vmovdqa64 zmm0, [edi]
vmovdqu64 zmm1, [edi + SCREEN_WIDTH + 1]
vpavgb zmm2, zmm0, [edi + 1]
vpavgb zmm3, zmm1, [edi + SCREEN_WIDTH]
vpavgb zmm4, zmm2, zmm3
vmovdqa64 [edi], zmm4
add edi, SIMD_BYTES
loop .lop
end if
}
STARTAPP:
init_palette:
mov edi, pal
xor eax, eax
red_loop:
stosd
stosd
add eax, 0x040000
and eax, 0xFFFFFF
jnz red_loop
mov eax, 63*4 SHL 16
@@:
stosd
stosd
add ax, 0x0404
jnc @b
;zero_buffer:
mov ecx, SCREEN_WIDTH * SCREEN_HEIGHT / 4
; mov edi, buffer
xor eax, eax
rep stosd
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Main Functions
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
virtual at esp
global_x dd ?
global_y dd ?
seed dd ?
end virtual
rdtsc
push eax ; seed
push 100 * 64 ; global_y
push 160 * 64 ; global_x
jmp MAIN
red:
mcall 9, proc_info, -1
x = 100
y = 70
xsize = SCREEN_WIDTH+9
ysize = SCREEN_HEIGHT+4
areacolor = 0x54224466
mov eax, 12 ; function 12:tell os about windowdraw
mov ebx, 1 ; 1, start of draw
int 0x40
mov eax, 48
mov ebx, 4
int 0x40
lea ecx, [(y SHL 16) + ysize + eax]
xor eax, eax ; function 0 : define and draw window
mov ebx, (x SHL 16) + xsize ; [x start] *65536 + [x size]
mov edx, areacolor ; color of work area RRGGBB
mov edi, window_title
int 0x40
mov eax, 12 ; end of redraw
mov ebx, 2
int 0x40
MAIN:
test [proc_info.wnd_state], 0x04
jnz still
mov ecx, NUM_PARTS
mov ebp, particles
.advance_particles:
mov eax, [ebp + X_OFFSET]
mov ebx, [ebp + Y_OFFSET]
sar eax, 6
sar ebx, 6
cmp eax, 5
jb .new_particle
cmp eax, SCREEN_WIDTH - 5
jge .new_particle
cmp ebx, 5
jb .new_particle
cmp ebx, SCREEN_HEIGHT - 5
jl .part_ok
.new_particle:
call init_particle
jmp .advance_particles
.part_ok:
imul edi, ebx, SCREEN_WIDTH
mov dl, [ebp+COLOR_OFFSET]
mov [buffer+eax+edi], dl
mov eax, [ebp+X_SPEED_OFFSET]
add [ebp+X_OFFSET], eax
mov eax, [ebp+Y_SPEED_OFFSET]
add [ebp+Y_OFFSET], eax
rdtsc
and al, 0x7F
jnz .dont_inc_y_speed
inc dword [ebp+Y_SPEED_OFFSET]
.dont_inc_y_speed:
add ebp, PART_SIZE
loop .advance_particles
shade
; jmp .copy_buffer_to_video
blur_prepare
test dword [blur_right_flag] , 0x800000
jnz .do_blur_right
blur
rdtsc
and al, 1
jz .blur_ok
jmp .dont_blur
.do_blur_right:
blur_right
.blur_ok:
add dword [blur_right_flag], 0x1000
.dont_blur:
.copy_buffer_to_video:
mcall 48, 4
lea edx, [(5 SHL 16) + eax]
mov eax, 65
mov ebx, buffer
mov ecx, (SCREEN_WIDTH SHL 16) + SCREEN_HEIGHT
push 8
pop esi
mov edi, pal
xor ebp, ebp
int 0x40
still:
mov eax, 11 ; Test if there is an event in the queue.
int 0x40
dec eax ; redraw request ?
jz red
dec eax ; key in buffer ?
jz key
dec eax ; button in buffer ?
jz button
jmp MAIN
key:
mov eax, 2
int 0x40
; cmp ah, 1 ; Test Esc in Scan
; je close_app
cmp ah, 27 ; Test Esc in ASCII
je close_app
jmp MAIN
button:
; we have only one button, close
close_app:
mov eax, -1 ; close this program
int 0x40
init_particle:
rdtsc
and al, 0x1F
jnz .dont_re_init_globals
; init x
call rand
cdq
;xor dx, dx
mov ebx, SCREEN_WIDTH
div ebx
shl edx, 6
mov [4 + global_x], edx
; init y
call rand
cdq
;xor dx, dx
mov ebx, SCREEN_HEIGHT
div ebx
shl edx, 6
mov [4 + global_y], edx
.dont_re_init_globals:
; init x
mov eax, [4 + global_x]
mov [ebp + X_OFFSET], eax
; init y
mov eax, [4 + global_y]
mov [ebp + Y_OFFSET], eax
; init x speed
call rand
and eax, 31
sub eax, 15
;shl ax, 6
mov [ebp + X_SPEED_OFFSET], eax
; init y speed
call rand
and eax, 31
sub eax, 15
;shl ax, 6
mov [ebp + Y_SPEED_OFFSET], eax
; init color
mov [ebp + COLOR_OFFSET], dword 255
ret
rand:
mov eax, [8 + seed]
imul eax, 214013
add eax, 2531011
mov [8 + seed], eax
shr eax, 16
ret
; DATA AREA
window_title db 'Firework',0
align SIMD_BYTES
sub_mask db SIMD_BYTES dup 0x01
; x, y, x_speed, y_speed, color
particles: times NUM_PARTS dd 0, 0, 0, 0, 0
blur_right_flag dd 0
I_END:
proc_info process_information
align 16
pal rb 256 * 4
align SIMD_BYTES
buffer rb SCREEN_WIDTH * SCREEN_HEIGHT
E_END:
rd 0x200
stacktop: