kolibrios-fun/programs/develop/libraries/libmpg123/synth_sse_float.S
Sergey Semyonov (Serge) 2217a37e5b libmpg123 1.15.4
git-svn-id: svn://kolibrios.org@3960 a494cfbc-eb01-0410-851d-a64ba20cac60
2013-10-02 16:44:03 +00:00

242 lines
4.8 KiB
ArmAsm

/*
synth_sse_float: SSE optimized synth (float output version)
copyright 1995-2009 by the mpg123 project - free software under the terms of the LGPL 2.1
see COPYING and AUTHORS files in distribution or http://mpg123.org
initially written by Taihei Monma
*/
#include "mangle.h"
/* real *window; */
#define WINDOW %ebx
/* real *b0; */
#define B0 %edx
/* real *samples; */
#define SAMPLES %esi
/*
int synth_1to1_real_sse_asm(real *window, real *b0, real *samples, int bo1);
return value: number of clipped samples (0)
*/
#ifndef __APPLE__
.section .rodata
#else
.data
#endif
ALIGN32
ASM_NAME(scale_sse):
.long 939524096
.long 939524096
.long 939524096
.long 939524096
.text
ALIGN16
.globl ASM_NAME(synth_1to1_real_sse_asm)
ASM_NAME(synth_1to1_real_sse_asm):
pushl %ebp
movl %esp, %ebp
pushl %ebx
pushl %esi
movl 8(%ebp), WINDOW
movl 12(%ebp), B0
movl 16(%ebp), SAMPLES
movl 20(%ebp), %eax
shll $2, %eax
leal 64(WINDOW), WINDOW
subl %eax, WINDOW
movl $4, %ecx
ALIGN16
Loop_start_1:
movups (WINDOW), %xmm0
movups 16(WINDOW), %xmm1
movups 32(WINDOW), %xmm2
movups 48(WINDOW), %xmm3
movups 128(WINDOW), %xmm4
movups 144(WINDOW), %xmm5
movups 160(WINDOW), %xmm6
movups 176(WINDOW), %xmm7
mulps 0(B0), %xmm0
mulps 16(B0), %xmm1
mulps 32(B0), %xmm2
mulps 48(B0), %xmm3
mulps 64(B0), %xmm4
mulps 80(B0), %xmm5
mulps 96(B0), %xmm6
mulps 112(B0), %xmm7
addps %xmm1, %xmm0
addps %xmm3, %xmm2
addps %xmm5, %xmm4
addps %xmm7, %xmm6
addps %xmm2, %xmm0
addps %xmm6, %xmm4
movaps %xmm4, %xmm5
movaps %xmm0, %xmm4
leal 256(WINDOW), WINDOW
leal 128(B0), B0
movups (WINDOW), %xmm0
movups 16(WINDOW), %xmm1
movups 32(WINDOW), %xmm2
movups 48(WINDOW), %xmm3
movups 128(WINDOW), %xmm6
movups 144(WINDOW), %xmm7
mulps (B0), %xmm0
mulps 16(B0), %xmm1
mulps 32(B0), %xmm2
mulps 48(B0), %xmm3
mulps 64(B0), %xmm6
mulps 80(B0), %xmm7
addps %xmm1, %xmm0
addps %xmm3, %xmm2
addps %xmm7, %xmm6
movups 160(WINDOW), %xmm1
movups 176(WINDOW), %xmm3
mulps 96(B0), %xmm1
mulps 112(B0), %xmm3
addps %xmm2, %xmm0
addps %xmm3, %xmm1
addps %xmm1, %xmm6
movaps %xmm6, %xmm7
movaps %xmm0, %xmm6
leal 256(WINDOW), WINDOW
leal 128(B0), B0
movaps %xmm4, %xmm0
movaps %xmm6, %xmm1
unpcklps %xmm5, %xmm4
unpcklps %xmm7, %xmm6
unpckhps %xmm5, %xmm0
unpckhps %xmm7, %xmm1
movaps %xmm4, %xmm2
movaps %xmm0, %xmm3
movlhps %xmm6, %xmm4
movhlps %xmm2, %xmm6
movlhps %xmm1, %xmm0
movhlps %xmm3, %xmm1
subps %xmm6, %xmm4
subps %xmm1, %xmm0
addps %xmm4, %xmm0
movups (SAMPLES), %xmm1
movups 16(SAMPLES), %xmm2
mulps ASM_NAME(scale_sse), %xmm0
shufps $0xdd, %xmm2, %xmm1
movaps %xmm0, %xmm2
unpcklps %xmm1, %xmm0
unpckhps %xmm1, %xmm2
movups %xmm0, (SAMPLES)
movups %xmm2, 16(SAMPLES)
leal 32(SAMPLES), SAMPLES
decl %ecx
jnz Loop_start_1
movl $4, %ecx
ALIGN16
Loop_start_2:
movups (WINDOW), %xmm0
movups 16(WINDOW), %xmm1
movups 32(WINDOW), %xmm2
movups 48(WINDOW), %xmm3
movups 128(WINDOW), %xmm4
movups 144(WINDOW), %xmm5
movups 160(WINDOW), %xmm6
movups 176(WINDOW), %xmm7
mulps 0(B0), %xmm0
mulps 16(B0), %xmm1
mulps 32(B0), %xmm2
mulps 48(B0), %xmm3
mulps -64(B0), %xmm4
mulps -48(B0), %xmm5
mulps -32(B0), %xmm6
mulps -16(B0), %xmm7
addps %xmm1, %xmm0
addps %xmm3, %xmm2
addps %xmm5, %xmm4
addps %xmm7, %xmm6
addps %xmm2, %xmm0
addps %xmm6, %xmm4
movaps %xmm4, %xmm5
movaps %xmm0, %xmm4
leal 256(WINDOW), WINDOW
leal -128(B0), B0
movups (WINDOW), %xmm0
movups 16(WINDOW), %xmm1
movups 32(WINDOW), %xmm2
movups 48(WINDOW), %xmm3
movups 128(WINDOW), %xmm6
movups 144(WINDOW), %xmm7
mulps (B0), %xmm0
mulps 16(B0), %xmm1
mulps 32(B0), %xmm2
mulps 48(B0), %xmm3
mulps -64(B0), %xmm6
mulps -48(B0), %xmm7
addps %xmm1, %xmm0
addps %xmm3, %xmm2
addps %xmm7, %xmm6
movups 160(WINDOW), %xmm1
movups 176(WINDOW), %xmm3
mulps -32(B0), %xmm1
mulps -16(B0), %xmm3
addps %xmm2, %xmm0
addps %xmm3, %xmm1
addps %xmm1, %xmm6
movaps %xmm6, %xmm7
movaps %xmm0, %xmm6
leal 256(WINDOW), WINDOW
leal -128(B0), B0
movaps %xmm4, %xmm0
movaps %xmm6, %xmm1
unpcklps %xmm5, %xmm4
unpcklps %xmm7, %xmm6
unpckhps %xmm5, %xmm0
unpckhps %xmm7, %xmm1
movaps %xmm4, %xmm2
movaps %xmm0, %xmm3
movlhps %xmm6, %xmm4
movhlps %xmm2, %xmm6
movlhps %xmm1, %xmm0
movhlps %xmm3, %xmm1
addps %xmm6, %xmm4
addps %xmm1, %xmm0
addps %xmm4, %xmm0
movups (SAMPLES), %xmm1
movups 16(SAMPLES), %xmm2
mulps ASM_NAME(scale_sse), %xmm0
shufps $0xdd, %xmm2, %xmm1
movaps %xmm0, %xmm2
unpcklps %xmm1, %xmm0
unpckhps %xmm1, %xmm2
movups %xmm0, (SAMPLES)
movups %xmm2, 16(SAMPLES)
leal 32(SAMPLES), SAMPLES
decl %ecx
jnz Loop_start_2
xorl %eax, %eax
popl %esi
popl %ebx
movl %ebp, %esp
popl %ebp
ret
NONEXEC_STACK