Commit 93681fbd authored by Ronald S. Bultje's avatar Ronald S. Bultje
Browse files

swscale: fix compile on ppc.

parent e758573a
......@@ -86,9 +86,11 @@ altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW)
}
static inline void
yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc,
int lumFilterSize, const int16_t *chrFilter,
const int16_t **chrUSrc, const int16_t **chrVSrc,
int chrFilterSize, uint8_t *dest, uint8_t *uDest,
uint8_t *vDest, int dstW, int chrDstW)
{
const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
register int i, j;
......@@ -159,22 +161,22 @@ yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc, int lumF
vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter
perm = vec_lvsl(0, chrSrc[j]);
l1 = vec_ld(0, chrSrc[j]);
l1_V = vec_ld(2048 << 1, chrSrc[j]);
perm = vec_lvsl(0, chrUSrc[j]);
l1 = vec_ld(0, chrUSrc[j]);
l1_V = vec_ld(0, chrVSrc[j]);
for (i = 0; i < (chrDstW - 7); i+=8) {
int offset = i << 2;
vector signed short l2 = vec_ld((i << 1) + 16, chrSrc[j]);
vector signed short l2_V = vec_ld(((i + 2048) << 1) + 16, chrSrc[j]);
vector signed short l2 = vec_ld((i << 1) + 16, chrUSrc[j]);
vector signed short l2_V = vec_ld((i << 1) + 16, chrVSrc[j]);
vector signed int v1 = vec_ld(offset, u);
vector signed int v2 = vec_ld(offset + 16, u);
vector signed int v1_V = vec_ld(offset, v);
vector signed int v2_V = vec_ld(offset + 16, v);
vector signed short ls = vec_perm(l1, l2, perm); // chrSrc[j][i] ... chrSrc[j][i+7]
vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrSrc[j][i+2048] ... chrSrc[j][i+2055]
vector signed short ls = vec_perm(l1, l2, perm); // chrUSrc[j][i] ... chrUSrc[j][i+7]
vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrVSrc[j][i] ... chrVSrc[j][i]
vector signed int i1 = vec_mule(vChrFilter, ls);
vector signed int i2 = vec_mulo(vChrFilter, ls);
......@@ -182,9 +184,9 @@ yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc, int lumF
vector signed int i2_V = vec_mulo(vChrFilter, ls_V);
vector signed int vf1 = vec_mergeh(i1, i2);
vector signed int vf2 = vec_mergel(i1, i2); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
vector signed int vf2 = vec_mergel(i1, i2); // chrUSrc[j][i] * chrFilter[j] ... chrUSrc[j][i+7] * chrFilter[j]
vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrVSrc[j][i] * chrFilter[j] ... chrVSrc[j][i+7] * chrFilter[j]
vector signed int vo1 = vec_add(v1, vf1);
vector signed int vo2 = vec_add(v2, vf2);
......@@ -200,8 +202,8 @@ yuv2yuvX_altivec_real(const int16_t *lumFilter, const int16_t **lumSrc, int lumF
l1_V = l2_V;
}
for ( ; i < chrDstW; i++) {
u[i] += chrSrc[j][i] * chrFilter[j];
v[i] += chrSrc[j][i + 2048] * chrFilter[j];
u[i] += chrUSrc[j][i] * chrFilter[j];
v[i] += chrVSrc[j][i] * chrFilter[j];
}
}
altivec_packIntArrayToCharArray(u, uDest, chrDstW);
......
......@@ -24,21 +24,28 @@
#endif
#if COMPILE_TEMPLATE_ALTIVEC
static inline void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize, const int16_t **alpSrc,
uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, long dstW, long chrDstW)
static inline void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter,
const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrUSrc,
const int16_t **chrVSrc, int chrFilterSize,
const int16_t **alpSrc,
uint8_t *dest, uint8_t *uDest, uint8_t *vDest,
uint8_t *aDest, long dstW, long chrDstW)
{
yuv2yuvX_altivec_real(lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
chrFilter, chrUSrc, chrVSrc, chrFilterSize,
dest, uDest, vDest, dstW, chrDstW);
}
/**
* vertical scale YV12 to RGB
*/
static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
const int16_t **alpSrc, uint8_t *dest, long dstW, long dstY)
static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter,
const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrUSrc,
const int16_t **chrVSrc, int chrFilterSize,
const int16_t **alpSrc, uint8_t *dest,
long dstW, long dstY)
{
/* The following list of supported dstFormat values should
match what's found in the body of ff_yuv2packedX_altivec() */
......@@ -47,11 +54,11 @@ static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter,
c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 ||
c->dstFormat==PIX_FMT_RGBA || c->dstFormat==PIX_FMT_ARGB))
ff_yuv2packedX_altivec(c, lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
chrFilter, chrUSrc, chrVSrc, chrFilterSize,
dest, dstW, dstY);
else
yuv2packedXinC(c, lumFilter, lumSrc, lumFilterSize,
chrFilter, chrSrc, chrFilterSize,
chrFilter, chrUSrc, chrVSrc, chrFilterSize,
alpSrc, dest, dstW, dstY);
}
#endif
......
......@@ -778,10 +778,11 @@ void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4], int b
void
ff_yuv2packedX_altivec(SwsContext *c,
const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
uint8_t *dest, int dstW, int dstY)
ff_yuv2packedX_altivec(SwsContext *c, const int16_t *lumFilter,
const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrUSrc,
const int16_t **chrVSrc, int chrFilterSize,
uint8_t *dest, int dstW, int dstY)
{
int i,j;
vector signed short X,X0,X1,Y0,U0,V0,Y1,U1,V1,U,V;
......@@ -816,9 +817,9 @@ ff_yuv2packedX_altivec(SwsContext *c,
V = RND;
/* extract 8 coeffs from U,V */
for (j=0; j<chrFilterSize; j++) {
X = vec_ld (0, &chrSrc[j][i/2]);
X = vec_ld (0, &chrUSrc[j][i/2]);
U = vec_mradds (X, CCoeffs[j], U);
X = vec_ld (0, &chrSrc[j][i/2+2048]);
X = vec_ld (0, &chrVSrc[j][i/2]);
V = vec_mradds (X, CCoeffs[j], V);
}
......@@ -894,9 +895,9 @@ ff_yuv2packedX_altivec(SwsContext *c,
V = RND;
/* extract 8 coeffs from U,V */
for (j=0; j<chrFilterSize; j++) {
X = vec_ld (0, &chrSrc[j][i/2]);
X = vec_ld (0, &chrUSrc[j][i/2]);
U = vec_mradds (X, CCoeffs[j], U);
X = vec_ld (0, &chrSrc[j][i/2+2048]);
X = vec_ld (0, &chrVSrc[j][i/2]);
V = vec_mradds (X, CCoeffs[j], V);
}
......
......@@ -333,9 +333,10 @@ SwsFunc ff_yuv2rgb_init_mlib(SwsContext *c);
SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c);
SwsFunc ff_yuv2rgb_get_func_ptr_bfin(SwsContext *c);
void ff_bfin_get_unscaled_swscale(SwsContext *c);
void ff_yuv2packedX_altivec(SwsContext *c,
const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
void ff_yuv2packedX_altivec(SwsContext *c, const int16_t *lumFilter,
const int16_t **lumSrc, int lumFilterSize,
const int16_t *chrFilter, const int16_t **chrUSrc,
const int16_t **chrVSrc, int chrFilterSize,
uint8_t *dest, int dstW, int dstY);
const char *sws_format_name(enum PixelFormat format);
......
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