Skip to content

F
ffmpeg.wasm-core
Commit 711c8ee7 authored by Michael Niedermayer's avatar Michael Niedermayer
Browse files
Options

Merge commit '4a7af92c' 

* commit '4a7af92c':
  sbrdsp: Unroll and use integer operations
  sbrdsp: Unroll sbr_autocorrelate_c
  x86: sbrdsp: Implement SSE2 qmf_deint_bfly

Conflicts:
	libavcodec/sbrdsp.c
	libavcodec/x86/sbrdsp.asm
	libavcodec/x86/sbrdsp_init.c
Merged-by: 's avatarMichael Niedermayer <michaelni@gmx.at>
parents f18d2dff 4a7af92c
Hide whitespace changes
Inline Side-by-side
Showing with 65 additions and 65 deletions
libavcodec/sbrdsp.c
View file @ 711c8ee7
......@@ -52,41 +52,41 @@ static float sbr_sum_square_c(float (*x)[2], int n)
static void sbr_neg_odd_64_c(float *x)
{
union av_intfloat32 *xi = (union av_intfloat32*)x;
union av_intfloat32 *xi = (union av_intfloat32*) x;
int i;
for (i = 1; i < 64; i += 4)
{
xi[i+0].i ^= 1U<<31;
xi[i+2].i ^= 1U<<31;
for (i = 1; i < 64; i += 4) {
xi[i + 0].i ^= 1U << 31;
xi[i + 2].i ^= 1U << 31;
}
}
static void sbr_qmf_pre_shuffle_c(float *z)
{
union av_intfloat32 *zi = (union av_intfloat32*)z;
union av_intfloat32 *zi = (union av_intfloat32*) z;
int k;
zi[64].i = zi[0].i;
zi[65].i = zi[1].i;
for (k = 1; k < 31; k+=2) {
zi[64+2*k+0].i = zi[64 - k].i ^ (1U<<31);
zi[64+2*k+1].i = zi[ k + 1].i;
zi[64+2*k+2].i = zi[63 - k].i ^ (1U<<31);
zi[64+2*k+3].i = zi[ k + 2].i;
for (k = 1; k < 31; k += 2) {
zi[64 + 2 * k + 0].i = zi[64 - k].i ^ (1U << 31);
zi[64 + 2 * k + 1].i = zi[ k + 1].i;
zi[64 + 2 * k + 2].i = zi[63 - k].i ^ (1U << 31);
zi[64 + 2 * k + 3].i = zi[ k + 2].i;
}
zi[64+2*31+0].i = zi[64 - 31].i ^ (1U<<31);
zi[64+2*31+1].i = zi[31 + 1].i;
zi[64 + 2 * 31 + 0].i = zi[64 - 31].i ^ (1U << 31);
zi[64 + 2 * 31 + 1].i = zi[31 + 1].i;
}
static void sbr_qmf_post_shuffle_c(float W[32][2], const float *z)
{
const union av_intfloat32 *zi = (const union av_intfloat32*)z;
union av_intfloat32 *Wi = (union av_intfloat32*)W;
const union av_intfloat32 *zi = (const union av_intfloat32*) z;
union av_intfloat32 *Wi = (union av_intfloat32*) W;
int k;
for (k = 0; k < 32; k+=2) {
Wi[2*k+0].i = zi[63-k].i ^ (1U<<31);
Wi[2*k+1].i = zi[k+0].i;
Wi[2*k+2].i = zi[62-k].i ^ (1U<<31);
Wi[2*k+3].i = zi[k+1].i;
for (k = 0; k < 32; k += 2) {
Wi[2 * k + 0].i = zi[63 - k].i ^ (1U << 31);
Wi[2 * k + 1].i = zi[ k + 0].i;
Wi[2 * k + 2].i = zi[62 - k].i ^ (1U << 31);
Wi[2 * k + 3].i = zi[ k + 1].i;
}
}
......@@ -96,8 +96,8 @@ static void sbr_qmf_deint_neg_c(float *v, const float *src)
union av_intfloat32 *vi = (union av_intfloat32*)v;
int i;
for (i = 0; i < 32; i++) {
vi[ i].i = si[63 - 2*i ].i;
vi[63 - i].i = si[63 - 2*i - 1].i ^ (1U<<31);
vi[ i].i = si[63 - 2 * i ].i;
vi[63 - i].i = si[63 - 2 * i - 1].i ^ (1U << 31);
}
}
......@@ -139,32 +139,32 @@ static av_always_inline void autocorrelate(const float x[40][2],
static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
{
#if 0
// This code is slower because it multiplies memory accesses.
// It is left as eucational purpose and because it may offer
// a better reference for writing arch-specific dsp functions.
/* This code is slower because it multiplies memory accesses.
* It is left for educational purposes and because it may offer
* a better reference for writing arch-specific DSP functions. */
autocorrelate(x, phi, 0);
autocorrelate(x, phi, 1);
autocorrelate(x, phi, 2);
#else
float real_sum2 = x[ 0][0] * x[ 2][0] + x[ 0][1] * x[ 2][1];
float imag_sum2 = x[ 0][0] * x[ 2][1] - x[ 0][1] * x[ 2][0];
float real_sum1 = 0.f, imag_sum1 = 0.f, real_sum0 = 0.0f;
float real_sum2 = x[0][0] * x[2][0] + x[0][1] * x[2][1];
float imag_sum2 = x[0][0] * x[2][1] - x[0][1] * x[2][0];
float real_sum1 = 0.0f, imag_sum1 = 0.0f, real_sum0 = 0.0f;
int i;
for (i = 1; i < 38; i++) {
real_sum0 += x[i][0] * x[i ][0] + x[i][1] * x[i ][1];
real_sum1 += x[i][0] * x[i+1][0] + x[i][1] * x[i+1][1];
imag_sum1 += x[i][0] * x[i+1][1] - x[i][1] * x[i+1][0];
real_sum2 += x[i][0] * x[i+2][0] + x[i][1] * x[i+2][1];
imag_sum2 += x[i][0] * x[i+2][1] - x[i][1] * x[i+2][0];
real_sum0 += x[i][0] * x[i ][0] + x[i][1] * x[i ][1];
real_sum1 += x[i][0] * x[i + 1][0] + x[i][1] * x[i + 1][1];
imag_sum1 += x[i][0] * x[i + 1][1] - x[i][1] * x[i + 1][0];
real_sum2 += x[i][0] * x[i + 2][0] + x[i][1] * x[i + 2][1];
imag_sum2 += x[i][0] * x[i + 2][1] - x[i][1] * x[i + 2][0];
}
phi[2-2][1][0] = real_sum2;
phi[2-2][1][1] = imag_sum2;
phi[2 ][1][0] = real_sum0 + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
phi[1 ][0][0] = real_sum0 + x[38][0] * x[38][0] + x[38][1] * x[38][1];
phi[2-1][1][0] = real_sum1 + x[ 0][0] * x[ 1][0] + x[ 0][1] * x[ 1][1];
phi[2-1][1][1] = imag_sum1 + x[ 0][0] * x[ 1][1] - x[ 0][1] * x[ 1][0];
phi[0 ][0][0] = real_sum1 + x[38][0] * x[39][0] + x[38][1] * x[39][1];
phi[0 ][0][1] = imag_sum1 + x[38][0] * x[39][1] - x[38][1] * x[39][0];
phi[2 - 2][1][0] = real_sum2;
phi[2 - 2][1][1] = imag_sum2;
phi[2 ][1][0] = real_sum0 + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
phi[1 ][0][0] = real_sum0 + x[38][0] * x[38][0] + x[38][1] * x[38][1];
phi[2 - 1][1][0] = real_sum1 + x[ 0][0] * x[ 1][0] + x[ 0][1] * x[ 1][1];
phi[2 - 1][1][1] = imag_sum1 + x[ 0][0] * x[ 1][1] - x[ 0][1] * x[ 1][0];
phi[0 ][0][0] = real_sum1 + x[38][0] * x[39][0] + x[38][1] * x[39][1];
phi[0 ][0][1] = imag_sum1 + x[38][0] * x[39][1] - x[38][1] * x[39][0];
#endif
}
......
libavcodec/x86/sbrdsp.asm
View file @ 711c8ee7
......@@ -252,36 +252,36 @@ cglobal sbr_neg_odd_64, 1,2,4,z
; sbr_qmf_deint_bfly(float *v, const float *src0, const float *src1)
%macro SBR_QMF_DEINT_BFLY 0
cglobal sbr_qmf_deint_bfly, 3,5,8, v,src0,src1,vrev,c
mov cq, 64*4-2*mmsize
lea vrevq, [vq + 64*4]
mov cq, 64*4-2*mmsize
lea vrevq, [vq + 64*4]
.loop:
mova m0, [src0q+cq]
mova m1, [src1q]
mova m4, [src0q+cq+mmsize]
mova m5, [src1q+mmsize]
mova m0, [src0q+cq]
mova m1, [src1q]
mova m4, [src0q+cq+mmsize]
mova m5, [src1q+mmsize]
%if cpuflag(sse2)
pshufd m2, m0, q0123
pshufd m3, m1, q0123
pshufd m6, m4, q0123
pshufd m7, m5, q0123
pshufd m2, m0, q0123
pshufd m3, m1, q0123
pshufd m6, m4, q0123
pshufd m7, m5, q0123
%else
shufps m2, m0, m0, q0123
shufps m3, m1, m1, q0123
shufps m6, m4, m4, q0123
shufps m7, m5, m5, q0123
shufps m2, m0, m0, q0123
shufps m3, m1, m1, q0123
shufps m6, m4, m4, q0123
shufps m7, m5, m5, q0123
%endif
addps m5, m2
subps m0, m7
addps m1, m6
subps m4, m3
mova [vrevq], m1
addps m5, m2
subps m0, m7
addps m1, m6
subps m4, m3
mova [vrevq], m1
mova [vrevq+mmsize], m5
mova [vq+cq], m0
mova [vq+cq], m0
mova [vq+cq+mmsize], m4
add src1q, 2*mmsize
add vrevq, 2*mmsize
sub cq, 2*mmsize
jge .loop
add src1q, 2*mmsize
add vrevq, 2*mmsize
sub cq, 2*mmsize
jge .loop
REP_RET
%endmacro
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment