AuroraMiddleware/qt5base-lts
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Minor improvements to bilinear filtering

Browse Source 
Duplicates the improved bounds check for rotated sampling in RGB64, and
improves two prologs by using that x1 == x2 during the prolog.

Change-Id: I562c5bee12e314c36d3b304f2f72d7635d22d7d4
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
This commit is contained in:
Allan Sandfeld Jensen 2016-09-01 16:54:43 +02:00 committed by Allan Sandfeld Jensen
parent f104e43a72
commit fe82f50a65
1 changed files with 28 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
src/gui/painting/qdrawhelper.cpp
View File

@ -2176,7 +2176,8 @@ static const uint * QT_FASTCALL fetchTransformedBilinearARGB32PM(uint *buffer, c
fetchTransformedBilinear_pixelBounds<blendType>(image_height, image_y1, image_y2, y1, y2);
const uint *s1 = (const uint *)data->texture.scanLine(y1);
const uint *s2 = (const uint *)data->texture.scanLine(y2);
int disty = ((fy & 0x0000ffff) + 0x0800) >> 12;
const int disty8 = (fy & 0x0000ffff) >> 8;
const int disty4 = (disty8 + 0x08) >> 4;
if (blendType != BlendTransformedBilinearTiled) {
#define BILINEAR_DOWNSCALE_BOUNDS_PROLOG \
@ -2188,12 +2189,9 @@ static const uint * QT_FASTCALL fetchTransformedBilinearARGB32PM(uint *buffer, c
fetchTransformedBilinear_pixelBounds<blendType>(image_width, image_x1, image_x2, x1, x2); \
if (x1 != x2) \
break; \
uint tl = s1[x1]; \
uint tr = s1[x2]; \
uint bl = s2[x1]; \
uint br = s2[x2]; \
int distx = ((fx & 0x0000ffff) + 0x0800) >> 12; \
*b = interpolate_4_pixels_16(tl, tr, bl, br, distx, disty); \
uint top = s1[x1]; \
uint bot = s2[x1]; \
*b = INTERPOLATE_PIXEL_256(top, 256 - disty8, bot, disty8); \
fx += fdx; \
++b; \
} \
@ -2209,7 +2207,7 @@ static const uint * QT_FASTCALL fetchTransformedBilinearARGB32PM(uint *buffer, c
const __m128i colorMask = _mm_set1_epi32(0x00ff00ff);
const __m128i v_256 = _mm_set1_epi16(256);
const __m128i v_disty = _mm_set1_epi16(disty);
const __m128i v_disty = _mm_set1_epi16(disty4);
const __m128i v_fdx = _mm_set1_epi32(fdx*4);
const __m128i v_fx_r = _mm_set1_epi32(0x8);
__m128i v_fx = _mm_setr_epi32(fx, fx + fdx, fx + fdx + fdx, fx + fdx + fdx + fdx);
@ -2241,7 +2239,7 @@ static const uint * QT_FASTCALL fetchTransformedBilinearARGB32PM(uint *buffer, c
const int16x8_t colorMask = vdupq_n_s16(0x00ff);
const int16x8_t invColorMask = vmvnq_s16(colorMask);
const int16x8_t v_256 = vdupq_n_s16(256);
const int16x8_t v_disty = vdupq_n_s16(disty);
const int16x8_t v_disty = vdupq_n_s16(disty4);
const int16x8_t v_disty_ = vshlq_n_s16(v_disty, 4);
int32x4_t v_fdx = vdupq_n_s32(fdx*4);
@ -2298,8 +2296,14 @@ static const uint * QT_FASTCALL fetchTransformedBilinearARGB32PM(uint *buffer, c
uint tr = s1[x2];
uint bl = s2[x1];
uint br = s2[x2];
int distx = ((fx & 0x0000ffff) + 0x0800) >> 12;
*b = interpolate_4_pixels_16(tl, tr, bl, br, distx, disty);
#if defined(__SSE2__) || defined(__ARM_NEON__)
// The optimized interpolate_4_pixels are faster than interpolate_4_pixels_16.
int distx8 = (fx & 0x0000ffff) >> 8;
*b = interpolate_4_pixels(tl, tr, bl, br, distx8, disty8);
#else
int distx4 = ((fx & 0x0000ffff) + 0x0800) >> 12;
*b = interpolate_4_pixels_16(tl, tr, bl, br, distx4, disty4);
#endif
fx += fdx;
++b;
}
@ -2980,10 +2984,8 @@ static const QRgba64 *QT_FASTCALL fetchTransformedBilinear64(QRgba64 *buffer, co
fetchTransformedBilinear_pixelBounds<blendType>(image_width, image_x1, image_x2, x1, x2);
if (x1 != x2)
break;
sbuf1[i * 2 + 0] = ((const uint*)s1)[x1];
sbuf1[i * 2 + 1] = ((const uint*)s1)[x2];
sbuf2[i * 2 + 0] = ((const uint*)s2)[x1];
sbuf2[i * 2 + 1] = ((const uint*)s2)[x2];
sbuf1[i * 2 + 0] = sbuf1[i * 2 + 1] = ((const uint*)s1)[x1];
sbuf2[i * 2 + 0] = sbuf2[i * 2 + 1] = ((const uint*)s2)[x1];
fx += fdx;
}
int fastLen;
@ -3102,6 +3104,16 @@ static const QRgba64 *QT_FASTCALL fetchTransformedBilinear64(QRgba64 *buffer, co
fx += fdx;
fy += fdy;
}
int fastLen = len;
if (fdx > 0)
fastLen = qMin(fastLen, int((qint64(image_x2) * fixed_scale - fx) / fdx));
else if (fdx < 0)
fastLen = qMin(fastLen, int((qint64(image_x1) * fixed_scale - fx) / fdx));
if (fdy > 0)
fastLen = qMin(fastLen, int((qint64(image_y2) * fixed_scale - fy) / fdy));
else if (fdy < 0)
fastLen = qMin(fastLen, int((qint64(image_y1) * fixed_scale - fy) / fdy));
fastLen -= 3;
const __m128i v_fdx = _mm_set1_epi32(fdx*4);
const __m128i v_fdy = _mm_set1_epi32(fdy*4);
@ -3111,15 +3123,7 @@ static const QRgba64 *QT_FASTCALL fetchTransformedBilinear64(QRgba64 *buffer, co
const uchar *s1 = data->texture.imageData;
const uchar *s2 = s1 + bytesPerLine;
const __m128i vbpl = _mm_shufflelo_epi16(_mm_cvtsi32_si128(bytesPerLine/4), _MM_SHUFFLE(0, 0, 0, 0));
for (; i < len-3; i+=4) {
if (fdx > 0 && (short)_mm_extract_epi16(v_fx, 7) >= image_x2)
break;
if (fdx < 0 && (short)_mm_extract_epi16(v_fx, 7) < image_x1)
break;
if (fdy > 0 && (short)_mm_extract_epi16(v_fy, 7) >= image_y2)
break;
if (fdy < 0 && (short)_mm_extract_epi16(v_fy, 7) < image_y1)
break;
for (; i < fastLen; i += 4) {
const __m128i vy = _mm_packs_epi32(_mm_srai_epi32(v_fy, 16), _mm_setzero_si128());
__m128i voffset = _mm_unpacklo_epi16(_mm_mullo_epi16(vy, vbpl), _mm_mulhi_epu16(vy, vbpl));
voffset = _mm_add_epi32(voffset, _mm_srli_epi32(v_fx, 16));