Skip to content

L
libswscale
Commit d03c61d5 authored by ramiro's avatar ramiro
Browse files
Options

Move fast bilinear scaler code to the existing h[yc]scale_fast() functions. 

git-svn-id: file:///var/local/repositories/mplayer/trunk/libswscale@30098 b3059339-0415-0410-9bf9-f77b7e298cf2
parent 1f13448e
Hide whitespace changes
Inline Side-by-side
Showing with 196 additions and 196 deletions
swscale_internal.h
View file @ d03c61d5
...@@ -261,10 +261,10 @@ typedef struct SwsContext { ...@@ -261,10 +261,10 @@ typedef struct SwsContext {
const uint8_t *src1, const uint8_t *src2, const uint8_t *src1, const uint8_t *src2,
long width, uint32_t *pal); ///< Unscaled conversion of chroma planes to YV12 for horizontal scaler. long width, uint32_t *pal); ///< Unscaled conversion of chroma planes to YV12 for horizontal scaler.
void (*hyscale_fast)(struct SwsContext *c, void (*hyscale_fast)(struct SwsContext *c,
int16_t *dst, int dstWidth, int16_t *dst, long dstWidth,
const uint8_t *src, int srcW, int xInc); const uint8_t *src, int srcW, int xInc);
void (*hcscale_fast)(struct SwsContext *c, void (*hcscale_fast)(struct SwsContext *c,
int16_t *dst, int dstWidth, int16_t *dst, long dstWidth,
const uint8_t *src1, const uint8_t *src2, const uint8_t *src1, const uint8_t *src2,
int srcW, int xInc); int srcW, int xInc);
......
swscale_template.c
View file @ d03c61d5
...@@ -2259,9 +2259,116 @@ static void RENAME(lumRangeFromJpeg)(uint16_t *dst, int width) ...@@ -2259,9 +2259,116 @@ static void RENAME(lumRangeFromJpeg)(uint16_t *dst, int width)
"shrl $9, %%esi \n\t" \ "shrl $9, %%esi \n\t" \
static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst, static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
int dstWidth, const uint8_t *src, int srcW, long dstWidth, const uint8_t *src, int srcW,
int xInc) int xInc)
{ {
#if ARCH_X86 && CONFIG_GPL
#if COMPILE_TEMPLATE_MMX2
int32_t *mmx2FilterPos = c->lumMmx2FilterPos;
int16_t *mmx2Filter = c->lumMmx2Filter;
int canMMX2BeUsed = c->canMMX2BeUsed;
void *mmx2FilterCode= c->lumMmx2FilterCode;
int i;
#if defined(PIC)
DECLARE_ALIGNED(8, uint64_t, ebxsave);
#endif
if (canMMX2BeUsed) {
__asm__ volatile(
#if defined(PIC)
"mov %%"REG_b", %5 \n\t"
#endif
"pxor %%mm7, %%mm7 \n\t"
"mov %0, %%"REG_c" \n\t"
"mov %1, %%"REG_D" \n\t"
"mov %2, %%"REG_d" \n\t"
"mov %3, %%"REG_b" \n\t"
"xor %%"REG_a", %%"REG_a" \n\t" // i
PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t"
PREFETCH" 64(%%"REG_c") \n\t"
#if ARCH_X86_64
#define CALL_MMX2_FILTER_CODE \
"movl (%%"REG_b"), %%esi \n\t"\
"call *%4 \n\t"\
"movl (%%"REG_b", %%"REG_a"), %%esi \n\t"\
"add %%"REG_S", %%"REG_c" \n\t"\
"add %%"REG_a", %%"REG_D" \n\t"\
"xor %%"REG_a", %%"REG_a" \n\t"\
#else
#define CALL_MMX2_FILTER_CODE \
"movl (%%"REG_b"), %%esi \n\t"\
"call *%4 \n\t"\
"addl (%%"REG_b", %%"REG_a"), %%"REG_c" \n\t"\
"add %%"REG_a", %%"REG_D" \n\t"\
"xor %%"REG_a", %%"REG_a" \n\t"\
#endif /* ARCH_X86_64 */
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
#if defined(PIC)
"mov %5, %%"REG_b" \n\t"
#endif
:: "m" (src), "m" (dst), "m" (mmx2Filter), "m" (mmx2FilterPos),
"m" (mmx2FilterCode)
#if defined(PIC)
,"m" (ebxsave)
#endif
: "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
#if !defined(PIC)
,"%"REG_b
#endif
);
for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) dst[i] = src[srcW-1]*128;
} else {
#endif /* COMPILE_TEMPLATE_MMX2 */
x86_reg xInc_shr16 = xInc >> 16;
uint16_t xInc_mask = xInc & 0xffff;
//NO MMX just normal asm ...
__asm__ volatile(
"xor %%"REG_a", %%"REG_a" \n\t" // i
"xor %%"REG_d", %%"REG_d" \n\t" // xx
"xorl %%ecx, %%ecx \n\t" // xalpha
ASMALIGN(4)
"1: \n\t"
"movzbl (%0, %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%0, %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86
"movw %%si, (%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"movzbl (%0, %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%0, %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86
"movw %%si, 2(%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"add $2, %%"REG_a" \n\t"
"cmp %2, %%"REG_a" \n\t"
" jb 1b \n\t"
:: "r" (src), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask)
: "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi"
);
#if COMPILE_TEMPLATE_MMX2
} //if MMX2 can't be used
#endif
#else
int i; int i;
unsigned int xpos=0; unsigned int xpos=0;
for (i=0;i<dstWidth;i++) { for (i=0;i<dstWidth;i++) {
...@@ -2270,6 +2377,7 @@ static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst, ...@@ -2270,6 +2377,7 @@ static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
dst[i]= (src[xx]<<7) + (src[xx+1] - src[xx])*xalpha; dst[i]= (src[xx]<<7) + (src[xx+1] - src[xx])*xalpha;
xpos+=xInc; xpos+=xInc;
} }
#endif /* ARCH_X86 */
} }
// *** horizontal scale Y line to temp buffer // *** horizontal scale Y line to temp buffer
...@@ -2279,10 +2387,6 @@ static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth, ...@@ -2279,10 +2387,6 @@ static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth,
enum PixelFormat srcFormat, uint8_t *formatConvBuffer, enum PixelFormat srcFormat, uint8_t *formatConvBuffer,
uint32_t *pal, int isAlpha) uint32_t *pal, int isAlpha)
{ {
int32_t av_unused *mmx2FilterPos = c->lumMmx2FilterPos;
int16_t av_unused *mmx2Filter = c->lumMmx2Filter;
int av_unused canMMX2BeUsed = c->canMMX2BeUsed;
void av_unused *mmx2FilterCode= c->lumMmx2FilterCode;
void (*toYV12)(uint8_t *, const uint8_t *, long, uint32_t *) = isAlpha ? c->alpToYV12 : c->lumToYV12; void (*toYV12)(uint8_t *, const uint8_t *, long, uint32_t *) = isAlpha ? c->alpToYV12 : c->lumToYV12;
void (*convertRange)(uint16_t *, int) = isAlpha ? NULL : c->lumConvertRange; void (*convertRange)(uint16_t *, int) = isAlpha ? NULL : c->lumConvertRange;
...@@ -2297,121 +2401,118 @@ static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth, ...@@ -2297,121 +2401,118 @@ static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth,
{ {
c->hScale(dst, dstWidth, src, srcW, xInc, hLumFilter, hLumFilterPos, hLumFilterSize); c->hScale(dst, dstWidth, src, srcW, xInc, hLumFilter, hLumFilterPos, hLumFilterSize);
} else { // fast bilinear upscale / crap downscale } else { // fast bilinear upscale / crap downscale
c->hyscale_fast(c, dst, dstWidth, src, srcW, xInc);
}
if (convertRange)
convertRange(dst, dstWidth);
}
static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst,
long dstWidth, const uint8_t *src1,
const uint8_t *src2, int srcW, int xInc)
{
#if ARCH_X86 && CONFIG_GPL #if ARCH_X86 && CONFIG_GPL
#if COMPILE_TEMPLATE_MMX2 #if COMPILE_TEMPLATE_MMX2
int i; int32_t *mmx2FilterPos = c->chrMmx2FilterPos;
int16_t *mmx2Filter = c->chrMmx2Filter;
int canMMX2BeUsed = c->canMMX2BeUsed;
void *mmx2FilterCode= c->chrMmx2FilterCode;
int i;
#if defined(PIC) #if defined(PIC)
DECLARE_ALIGNED(8, uint64_t, ebxsave); DECLARE_ALIGNED(8, uint64_t, ebxsave);
#endif #endif
if (canMMX2BeUsed) { if (canMMX2BeUsed) {
__asm__ volatile( __asm__ volatile(
#if defined(PIC) #if defined(PIC)
"mov %%"REG_b", %5 \n\t" "mov %%"REG_b", %6 \n\t"
#endif #endif
"pxor %%mm7, %%mm7 \n\t" "pxor %%mm7, %%mm7 \n\t"
"mov %0, %%"REG_c" \n\t" "mov %0, %%"REG_c" \n\t"
"mov %1, %%"REG_D" \n\t" "mov %1, %%"REG_D" \n\t"
"mov %2, %%"REG_d" \n\t" "mov %2, %%"REG_d" \n\t"
"mov %3, %%"REG_b" \n\t" "mov %3, %%"REG_b" \n\t"
"xor %%"REG_a", %%"REG_a" \n\t" // i "xor %%"REG_a", %%"REG_a" \n\t" // i
PREFETCH" (%%"REG_c") \n\t" PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t" PREFETCH" 32(%%"REG_c") \n\t"
PREFETCH" 64(%%"REG_c") \n\t" PREFETCH" 64(%%"REG_c") \n\t"
#if ARCH_X86_64 CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
#define CALL_MMX2_FILTER_CODE \ CALL_MMX2_FILTER_CODE
"movl (%%"REG_b"), %%esi \n\t"\ CALL_MMX2_FILTER_CODE
"call *%4 \n\t"\ "xor %%"REG_a", %%"REG_a" \n\t" // i
"movl (%%"REG_b", %%"REG_a"), %%esi \n\t"\ "mov %5, %%"REG_c" \n\t" // src
"add %%"REG_S", %%"REG_c" \n\t"\ "mov %1, %%"REG_D" \n\t" // buf1
"add %%"REG_a", %%"REG_D" \n\t"\ "add $"AV_STRINGIFY(VOF)", %%"REG_D" \n\t"
"xor %%"REG_a", %%"REG_a" \n\t"\ PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t"
#else PREFETCH" 64(%%"REG_c") \n\t"
#define CALL_MMX2_FILTER_CODE \ CALL_MMX2_FILTER_CODE
"movl (%%"REG_b"), %%esi \n\t"\ CALL_MMX2_FILTER_CODE
"call *%4 \n\t"\ CALL_MMX2_FILTER_CODE
"addl (%%"REG_b", %%"REG_a"), %%"REG_c" \n\t"\ CALL_MMX2_FILTER_CODE
"add %%"REG_a", %%"REG_D" \n\t"\
"xor %%"REG_a", %%"REG_a" \n\t"\
#endif /* ARCH_X86_64 */
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
#if defined(PIC) #if defined(PIC)
"mov %5, %%"REG_b" \n\t" "mov %6, %%"REG_b" \n\t"
#endif #endif
:: "m" (src), "m" (dst), "m" (mmx2Filter), "m" (mmx2FilterPos), :: "m" (src1), "m" (dst), "m" (mmx2Filter), "m" (mmx2FilterPos),
"m" (mmx2FilterCode) "m" (mmx2FilterCode), "m" (src2)
#if defined(PIC) #if defined(PIC)
,"m" (ebxsave) ,"m" (ebxsave)
#endif #endif
: "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
#if !defined(PIC) #if !defined(PIC)
,"%"REG_b ,"%"REG_b
#endif #endif
); );
for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) dst[i] = src[srcW-1]*128; for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) {
} else { //printf("%d %d %d\n", dstWidth, i, srcW);
dst[i] = src1[srcW-1]*128;
dst[i+VOFW] = src2[srcW-1]*128;
}
} else {
#endif /* COMPILE_TEMPLATE_MMX2 */ #endif /* COMPILE_TEMPLATE_MMX2 */
x86_reg xInc_shr16 = xInc >> 16; x86_reg xInc_shr16 = (x86_reg) (xInc >> 16);
uint16_t xInc_mask = xInc & 0xffff; uint16_t xInc_mask = xInc & 0xffff;
//NO MMX just normal asm ...
__asm__ volatile( __asm__ volatile(
"xor %%"REG_a", %%"REG_a" \n\t" // i "xor %%"REG_a", %%"REG_a" \n\t" // i
"xor %%"REG_d", %%"REG_d" \n\t" // xx "xor %%"REG_d", %%"REG_d" \n\t" // xx
"xorl %%ecx, %%ecx \n\t" // xalpha "xorl %%ecx, %%ecx \n\t" // xalpha
ASMALIGN(4) ASMALIGN(4)
"1: \n\t" "1: \n\t"
"movzbl (%0, %%"REG_d"), %%edi \n\t" //src[xx] "mov %0, %%"REG_S" \n\t"
"movzbl 1(%0, %%"REG_d"), %%esi \n\t" //src[xx+1] "movzbl (%%"REG_S", %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%%"REG_S", %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86 FAST_BILINEAR_X86
"movw %%si, (%%"REG_D", %%"REG_a", 2) \n\t" "movw %%si, (%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"movzbl (%0, %%"REG_d"), %%edi \n\t" //src[xx] "movzbl (%5, %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%0, %%"REG_d"), %%esi \n\t" //src[xx+1] "movzbl 1(%5, %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86 FAST_BILINEAR_X86
"movw %%si, 2(%%"REG_D", %%"REG_a", 2) \n\t" "movw %%si, "AV_STRINGIFY(VOF)"(%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"add $2, %%"REG_a" \n\t"
"cmp %2, %%"REG_a" \n\t"
" jb 1b \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"add $1, %%"REG_a" \n\t"
"cmp %2, %%"REG_a" \n\t"
" jb 1b \n\t"
:: "r" (src), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask) /* GCC 3.3 makes MPlayer crash on IA-32 machines when using "g" operand here,
which is needed to support GCC 4.0. */
#if ARCH_X86_64 && AV_GCC_VERSION_AT_LEAST(3,4)
:: "m" (src1), "m" (dst), "g" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
#else
:: "m" (src1), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
#endif
"r" (src2)
: "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi" : "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi"
); );
#if COMPILE_TEMPLATE_MMX2 #if COMPILE_TEMPLATE_MMX2
} //if MMX2 can't be used } //if MMX2 can't be used
#endif #endif
#else #else
c->hyscale_fast(c, dst, dstWidth, src, srcW, xInc);
#endif /* ARCH_X86 */
}
if (convertRange)
convertRange(dst, dstWidth);
}
static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst,
int dstWidth, const uint8_t *src1,
const uint8_t *src2, int srcW, int xInc)
{
int i; int i;
unsigned int xpos=0; unsigned int xpos=0;
for (i=0;i<dstWidth;i++) { for (i=0;i<dstWidth;i++) {
...@@ -2425,6 +2526,7 @@ static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst, ...@@ -2425,6 +2526,7 @@ static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst,
*/ */
xpos+=xInc; xpos+=xInc;
} }
#endif /* ARCH_X86 */
} }
inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src1, const uint8_t *src2, inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src1, const uint8_t *src2,
...@@ -2433,10 +2535,6 @@ inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, ...@@ -2433,10 +2535,6 @@ inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth,
enum PixelFormat srcFormat, uint8_t *formatConvBuffer, enum PixelFormat srcFormat, uint8_t *formatConvBuffer,
uint32_t *pal) uint32_t *pal)
{ {
int32_t av_unused *mmx2FilterPos = c->chrMmx2FilterPos;
int16_t av_unused *mmx2Filter = c->chrMmx2Filter;
int av_unused canMMX2BeUsed = c->canMMX2BeUsed;
void av_unused *mmx2FilterCode= c->chrMmx2FilterCode;
src1 += c->chrSrcOffset; src1 += c->chrSrcOffset;
src2 += c->chrSrcOffset; src2 += c->chrSrcOffset;
...@@ -2452,105 +2550,7 @@ inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, ...@@ -2452,105 +2550,7 @@ inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth,
c->hScale(dst , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize); c->hScale(dst , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
c->hScale(dst+VOFW, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize); c->hScale(dst+VOFW, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
} else { // fast bilinear upscale / crap downscale } else { // fast bilinear upscale / crap downscale
#if ARCH_X86 && CONFIG_GPL
#if COMPILE_TEMPLATE_MMX2
int i;
#if defined(PIC)
DECLARE_ALIGNED(8, uint64_t, ebxsave);
#endif
if (canMMX2BeUsed) {
__asm__ volatile(
#if defined(PIC)
"mov %%"REG_b", %6 \n\t"
#endif
"pxor %%mm7, %%mm7 \n\t"
"mov %0, %%"REG_c" \n\t"
"mov %1, %%"REG_D" \n\t"
"mov %2, %%"REG_d" \n\t"
"mov %3, %%"REG_b" \n\t"
"xor %%"REG_a", %%"REG_a" \n\t" // i
PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t"
PREFETCH" 64(%%"REG_c") \n\t"
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
"xor %%"REG_a", %%"REG_a" \n\t" // i
"mov %5, %%"REG_c" \n\t" // src
"mov %1, %%"REG_D" \n\t" // buf1
"add $"AV_STRINGIFY(VOF)", %%"REG_D" \n\t"
PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t"
PREFETCH" 64(%%"REG_c") \n\t"
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
CALL_MMX2_FILTER_CODE
#if defined(PIC)
"mov %6, %%"REG_b" \n\t"
#endif
:: "m" (src1), "m" (dst), "m" (mmx2Filter), "m" (mmx2FilterPos),
"m" (mmx2FilterCode), "m" (src2)
#if defined(PIC)
,"m" (ebxsave)
#endif
: "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
#if !defined(PIC)
,"%"REG_b
#endif
);
for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) {
//printf("%d %d %d\n", dstWidth, i, srcW);
dst[i] = src1[srcW-1]*128;
dst[i+VOFW] = src2[srcW-1]*128;
}
} else {
#endif /* COMPILE_TEMPLATE_MMX2 */
x86_reg xInc_shr16 = (x86_reg) (xInc >> 16);
uint16_t xInc_mask = xInc & 0xffff;
__asm__ volatile(
"xor %%"REG_a", %%"REG_a" \n\t" // i
"xor %%"REG_d", %%"REG_d" \n\t" // xx
"xorl %%ecx, %%ecx \n\t" // xalpha
ASMALIGN(4)
"1: \n\t"
"mov %0, %%"REG_S" \n\t"
"movzbl (%%"REG_S", %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%%"REG_S", %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86
"movw %%si, (%%"REG_D", %%"REG_a", 2) \n\t"
"movzbl (%5, %%"REG_d"), %%edi \n\t" //src[xx]
"movzbl 1(%5, %%"REG_d"), %%esi \n\t" //src[xx+1]
FAST_BILINEAR_X86
"movw %%si, "AV_STRINGIFY(VOF)"(%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //xalpha += xInc&0xFFFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>16 + carry
"add $1, %%"REG_a" \n\t"
"cmp %2, %%"REG_a" \n\t"
" jb 1b \n\t"
/* GCC 3.3 makes MPlayer crash on IA-32 machines when using "g" operand here,
which is needed to support GCC 4.0. */
#if ARCH_X86_64 && AV_GCC_VERSION_AT_LEAST(3,4)
:: "m" (src1), "m" (dst), "g" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
#else
:: "m" (src1), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
#endif
"r" (src2)
: "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi"
);
#if COMPILE_TEMPLATE_MMX2
} //if MMX2 can't be used
#endif
#else
c->hcscale_fast(c, dst, dstWidth, src1, src2, srcW, xInc); c->hcscale_fast(c, dst, dstWidth, src1, src2, srcW, xInc);
#endif /* ARCH_X86 */
} }
if (c->chrConvertRange) if (c->chrConvertRange)
......
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