format ELF64 executable 3 entry _start ;------------------------------------------------------------------------------- ; NAME: SMOOTHSTEP ; IN: xmm0 edge0 ; IN: xmm1 edge1 ; IN: xmm2 x ; OUT: xmm0 smoothstep(edge0, edge1, x) like in GLSL ;------------------------------------------------------------------------------- macro SMOOTHSTEP { subps xmm2,xmm0 ; x - edge0 subps xmm1,xmm0 ; edge1 - edge0 divps xmm2,xmm1 minps xmm2,dqword [g_1_0] maxps xmm2,dqword [g_0_0] movaps xmm0,dqword [g_3_0] movaps xmm1,xmm2 addps xmm1,xmm1 subps xmm0,xmm1 mulps xmm2,xmm2 mulps xmm0,xmm2 } ;------------------------------------------------------------------------------- segment readable executable ;------------------------------------------------------------------------------- ; NAME: sin ; IN: xmm0 | w z y x | ; OUT: xmm0 | sin(w) sin(z) sin(y) sin(x) | ;------------------------------------------------------------------------------- align 16 sin: movaps xmm7,xmm0 andps xmm0,dqword [g_inv_sign_mask] andps xmm7,dqword [g_sign_mask] mulps xmm0,dqword [g_2_div_pi] pxor xmm3,xmm3 movdqa xmm5,dqword [g_1] movaps xmm4,dqword [g_1_0] cvttps2dq xmm2,xmm0 pand xmm5,xmm2 pcmpeqd xmm5,xmm3 cvtdq2ps xmm6,xmm2 pand xmm2,dqword [g_2] pslld xmm2,30 subps xmm0,xmm6 minps xmm0,xmm4 subps xmm4,xmm0 andps xmm0,xmm5 andnps xmm5,xmm4 orps xmm0,xmm5 movaps xmm1,xmm0 mulps xmm0,xmm0 xorps xmm2,xmm7 orps xmm1,xmm2 movaps xmm7,xmm0 mulps xmm0,dqword [g_sincos_p3] addps xmm0,dqword [g_sincos_p2] mulps xmm0,xmm7 addps xmm0,dqword [g_sincos_p1] mulps xmm0,xmm7 addps xmm0,dqword [g_sincos_p0] mulps xmm0,xmm1 ret ;------------------------------------------------------------------------------- ; NAME: atanr2 ; IN: xmm0 | w0 z0 y0 x0 | ; IN: xmm1 | w1 z1 y1 x1 | ; OUT: xmm0 | atan2(w0,1/w1) atan2(z0,1/z1) atan2(y0,1/y1) atan2(x0,1/x1) | ;------------------------------------------------------------------------------- align 16 atanr2: movaps xmm2,dqword [g_sign_mask] xorps xmm3,xmm3 movaps xmm5,dqword [g_1_0] andps xmm2,xmm0 mulps xmm0,xmm1 orps xmm2,dqword [g_pi] cmpleps xmm3,xmm1 movaps xmm6,dqword [g_m1_0] rcpps xmm4,xmm0 cmpltps xmm5,xmm0 cmpnleps xmm6,xmm0 movaps xmm1,dqword [g_atan_s0] orps xmm5,xmm6 movaps xmm8,xmm2 movaps xmm9,xmm3 andps xmm4,xmm5 movaps xmm2,dqword [g_atan_t0] movaps xmm7,xmm5 andnps xmm5,xmm0 movaps xmm3,dqword [g_atan_s1] orps xmm4,xmm5 movaps xmm0,xmm4 movaps xmm6,dqword [g_atan_t1] mulps xmm4,xmm4 addps xmm1,xmm4 movaps xmm5,dqword [g_atan_s2] rcpps xmm1,xmm1 mulps xmm1,xmm2 movaps xmm2,dqword [g_atan_t2] addps xmm3,xmm4 addps xmm1,xmm3 movaps xmm3,dqword [g_atan_s3] rcpps xmm1,xmm1 mulps xmm1,xmm6 movaps xmm6,dqword [g_atan_t3] addps xmm5,xmm4 addps xmm1,xmm5 movaps xmm5,dqword [g_sign_mask] rcpps xmm1,xmm1 mulps xmm1,xmm2 addps xmm3,xmm4 movaps xmm4,dqword [g_0_5pi] mulps xmm6,xmm0 addps xmm1,xmm3 andps xmm0,xmm5 rcpps xmm1,xmm1 movaps xmm3,xmm9 mulps xmm1,xmm6 orps xmm0,xmm4 subps xmm0,xmm1 movaps xmm2,xmm8 andps xmm0,xmm7 andnps xmm7,xmm1 orps xmm0,xmm7 movaps xmm1,xmm0 andps xmm0,xmm3 addps xmm1,xmm2 andnps xmm3,xmm1 orps xmm0,xmm3 ret ;------------------------------------------------------------------------------- ; NAME: ComputeColors ; IN: rdi pointer to 4 image pixels ; IN: xmm0 | x3 x2 x1 x0 | normalized pixel x coordinates ; IN: xmm1 | y3 y2 y1 y0 | normalized pixel y coordinates ; OUT: xmm0 | ? b g r | pixel color ;------------------------------------------------------------------------------- align 16 ComputeColors: ; r = sqrt(x * x + y * y) movaps xmm2,xmm0 mulps xmm2,xmm0 movaps xmm3,xmm1 mulps xmm3,xmm1 addps xmm2,xmm3 sqrtps xmm15,xmm2 ; xmm15 = r ; a = atan2(y, x) = atanr2(y, 1/x) movaps xmm2,xmm0 movaps xmm0,xmm1 movaps xmm1,xmm2 rcpps xmm1,xmm1 call atanr2 movaps xmm14,xmm0 ; xmm14 = a ; s = 0.5 + 0.5 * sin(3.0 * a) mulps xmm0,dqword [g_3_0] movaps xmm12,xmm0 ; xmm12 = 3.0 * a call sin movaps xmm1,dqword [g_0_5] mulps xmm0,xmm1 addps xmm0,xmm1 movaps xmm13,xmm0 ; xmm13 = s ; g = sin(0.5 * pi + 3.0 * a) movaps xmm0,xmm12 addps xmm0,dqword [g_0_5pi] call sin ; xmm0 = g ; d = 0.3 + 0.6 * sqrt(s) + 0.15 * g * g mulps xmm0,xmm0 mulps xmm0,dqword [g_0_15] sqrtps xmm13,xmm13 mulps xmm13,dqword [g_0_6] addps xmm0,xmm13 addps xmm0,dqword [g_0_3] ; xmm0 = d ; h = r / d divps xmm15,xmm0 ; xmm15 = h ; f = 1.0 - smoothstep(0.95, 1.0, h) movaps xmm0,dqword [g_0_95] ; edge0 movaps xmm1,dqword [g_1_0] ; edge1 movaps xmm2,xmm15 ; x SMOOTHSTEP movaps xmm14,dqword [g_1_0] subps xmm14,xmm0 ; xmm14 = f ; movaps xmm0,xmm12 call sin movaps xmm2,xmm0 movaps xmm0,xmm15 mulps xmm0,dqword [g_0_05] addps xmm0,dqword [g_0_95] movaps xmm1,dqword [g_1_0] SMOOTHSTEP movaps xmm1,dqword [g_1_0] movaps xmm2,xmm1 subps xmm1,xmm15 mulps xmm1,dqword [g_0_5] mulps xmm0,xmm1 subps xmm2,xmm0 mulps xmm15,xmm2 ; movaps xmm6,dqword [g_1_0] movaps xmm7,dqword [g_c0] mulps xmm7,xmm15 addps xmm7,dqword [g_c1] movaps xmm0,xmm14 shufps xmm0,xmm0,00000000b ; xmm0 = | f f f f | movaps xmm1,xmm6 subps xmm1,xmm0 ; xmm1 = | 1-f 1-f 1-f 1-f | mulps xmm0,xmm7 mulps xmm1,xmm6 addps xmm0,xmm1 movaps [rdi+0],xmm0 movaps xmm0,xmm14 shufps xmm0,xmm0,01010101b ; xmm0 = | f f f f | movaps xmm1,xmm6 subps xmm1,xmm0 ; xmm1 = | 1-f 1-f 1-f 1-f | mulps xmm0,xmm7 mulps xmm1,xmm6 addps xmm0,xmm1 movaps [rdi+16],xmm0 movaps xmm0,xmm14 shufps xmm0,xmm0,10101010b ; xmm0 = | f f f f | movaps xmm1,xmm6 subps xmm1,xmm0 ; xmm1 = | 1-f 1-f 1-f 1-f | mulps xmm0,xmm7 mulps xmm1,xmm6 addps xmm0,xmm1 movaps [rdi+32],xmm0 movaps xmm0,xmm14 shufps xmm0,xmm0,11111111b ; xmm0 = | f f f f | movaps xmm1,xmm6 subps xmm1,xmm0 ; xmm1 = | 1-f 1-f 1-f 1-f | mulps xmm0,xmm7 mulps xmm1,xmm6 addps xmm0,xmm1 movaps [rdi+48],xmm0 ret ;------------------------------------------------------------------------------- ; NAME: AllocateMemory ; IN: rdi size in bytes ; OUT: rax pointer to allocated memory ;------------------------------------------------------------------------------- align 16 AllocateMemory: mov eax,9 ; sys_mmap mov rsi,rdi ; length xor edi,edi ; addr mov edx,0x1+0x2 ; PROT_READ | PROT_WRITE mov r10d,0x02+0x20 ; MAP_PRIVATE | MAP_ANONYMOUS mov r8,-1 ; fd xor r9d,r9d ; offset syscall ret ;------------------------------------------------------------------------------- ; NAME: main ; DESC: Program main function. ;------------------------------------------------------------------------------- main: imgptr equ rbp-8 imgptr_float equ rbp-16 x equ rbp-32 y equ rbp-48 push rbp mov rbp,rsp sub rsp,64 ; alloc memory mov edi,IMG_W*IMG_H*4 call AllocateMemory mov [imgptr],rax mov edi,IMG_W*IMG_H*16 call AllocateMemory mov [imgptr_float],rax mov rbx,rax ;; ;; Compute image ;; ; begin loops pxor xmm0,xmm0 movdqa [y],xmm0 .LoopY: movdqa xmm0,dqword [g_3_2_1_0] movdqa [x],xmm0 .LoopX: mov rdi,rbx ; compute normalized x coordinate [-1.0 , 1.0] cvtdq2ps xmm0,[x] divps xmm0,dqword [g_img_w] subps xmm0,dqword [g_0_5] addps xmm0,xmm0 ; compute normalized y coordinate [-1.0 , 1.0] cvtdq2ps xmm1,[y] divps xmm1,dqword [g_img_h] subps xmm1,dqword [g_0_5] addps xmm1,xmm1 call ComputeColors ; advance pixel pointer add rbx,64 ; continue .LoopX movdqa xmm0,[x] paddd xmm0,dqword [g_4] movdqa [x],xmm0 cmp dword [x+12],IMG_W jb .LoopX ; continue .LoopY movdqa xmm0,[y] paddd xmm0,dqword [g_1] movdqa [y],xmm0 cmp dword [y],IMG_H jb .LoopY ;; ;; Convert image ;; mov rsi,[imgptr_float] mov rdi,[imgptr] mov ecx,IMG_W*IMG_H .Convert: movaps xmm0,[rsi] maxps xmm0,dqword [g_0_0] minps xmm0,dqword [g_1_0] mulps xmm0,dqword [g_255_0] cvttps2dq xmm0,xmm0 pshufb xmm0,dqword [g_img_conv_mask] movd eax,xmm0 or eax,0xff000000 mov [rdi],eax add rsi,16 add rdi,4 sub ecx,1 jnz .Convert ;; ;; Save image ;; ; create TGA file mov eax,85 mov rdi,g_tga_name mov esi,110000000b syscall mov rbx,rax ; write header mov eax,1 mov rdi,rbx mov rsi,g_tga_head mov edx,18 syscall ; write pixel data mov eax,1 mov rdi,rbx mov rsi,[imgptr] mov edx,IMG_W*IMG_H*4 syscall mov rsp,rbp pop rbp ret restore imgptr,imgptr_float,x,y ;------------------------------------------------------------------------------- ; NAME: _start ; DESC: Program entry point. ;------------------------------------------------------------------------------- _start: call main ; terminate process mov eax,60 xor edi,edi syscall ;------------------------------------------------------------------------------- segment readable ;------------------------------------------------------------------------------- align 1 g_tga_name db 'clover.tga',0 g_tga_head db 0,0,2,9 dup 0 db (IMG_W and 0x00ff),(IMG_W and 0xff00) shr 8 db (IMG_H and 0x00ff),(IMG_H and 0xff00) shr 8,32,0 align 16 g_img_w dd 4 dup 800.0 g_img_h dd 4 dup 800.0 IMG_W=800 IMG_H=800 g_c0 dd 0.4,0.3,0.0,0.0 g_c1 dd 0.0,0.2,0.0,0.0 g_3_0 dd 4 dup 3.0 g_0_3 dd 4 dup 0.3 g_0_6 dd 4 dup 0.6 g_0_15 dd 4 dup 0.15 g_0_95 dd 4 dup 0.95 g_0_05 dd 4 dup 0.05 g_img_conv_mask db 8,4,0,12,12 dup 0x80 g_0_0 dd 4 dup 0.0 g_0_5 dd 4 dup 0.5 g_1_0 dd 4 dup 1.0 g_m1_0 dd 4 dup -1.0 g_255_0 dd 4 dup 255.0 g_1 dd 4 dup 1 g_2 dd 4 dup 2 g_4 dd 4 dup 4 g_3_2_1_0 dd 0,1,2,3 g_sign_mask dd 4 dup 0x80000000 g_inv_sign_mask dd 4 dup not 0x80000000 g_0_5pi dd 4 dup 1.57079633 g_2_div_pi dd 4 dup 0.636619772 g_pi dd 4 dup 3.1415926535897 g_sincos_p0 dd 4 dup 1.5707963267948963959 g_sincos_p1 dd 4 dup -0.64596409750621907082 g_sincos_p2 dd 4 dup 0.07969262624561800806 g_sincos_p3 dd 4 dup -0.00468175413106023168 g_atan_t0 dd 4 dup -0.091646118527267623468 g_atan_t1 dd 4 dup -1.3956945682312098640 g_atan_t2 dd 4 dup -94.393926122725531747 g_atan_t3 dd 4 dup 12.888383034157279340 g_atan_s0 dd 4 dup 1.2797564625607904396 g_atan_s1 dd 4 dup 2.1972168858277355914 g_atan_s2 dd 4 dup 6.8193064729268275701 g_atan_s3 dd 4 dup 28.205206687035841409 ;-------------------------------------------------------------------------------