Emulation/DirkSimple
1
0
Fork 0
mirror of https://github.com/icculus/DirkSimple.git synced 2024-06-02 03:07:42 -04:00
Code Issues Packages Projects Releases Wiki Activity

theoraplay: Calculate two rows per iteration in yuv->rgb conversion.

Browse source 
Since we use the same Cr and Cb values for each of these rows, this
lets us calculate them once and reuse them with both rows' Y values.

The test case went from 2.0 seconds to 1.9.

Reference Issue #23.
This commit is contained in:
Ryan C. Gordon 2023-04-10 19:40:40 -04:00
parent 6116eb368d
commit ed6705db21
No known key found for this signature in database
GPG key ID: FA148B892AB48044
2 changed files with 28 additions and 27 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
thirdparty/theoraplay/theoraplay.c vendored
View file

@ -97,7 +97,7 @@ static unsigned char *ConvertVideoFrame420ToIYUV(const THEORAPLAY_Allocator *all
// RGB
#define THEORAPLAY_CVT_FNNAME_420 ConvertVideoFrame420ToRGB
#define THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, h) ((w) * (h) * 3)
#define THEORAPLAY_CVT_RGB_OUTPUT(r, g, b) { \
#define THEORAPLAY_CVT_RGB_OUTPUT(dst, r, g, b) { \
*(dst++) = (unsigned char) ((r < 0) ? 0 : (r > 255) ? 255 : r); \
*(dst++) = (unsigned char) ((g < 0) ? 0 : (g > 255) ? 255 : g); \
*(dst++) = (unsigned char) ((b < 0) ? 0 : (b > 255) ? 255 : b); \
@ -107,7 +107,7 @@ static unsigned char *ConvertVideoFrame420ToIYUV(const THEORAPLAY_Allocator *all
// RGBA
#define THEORAPLAY_CVT_FNNAME_420 ConvertVideoFrame420ToRGBA
#define THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, h) ((w) * (h) * 4)
#define THEORAPLAY_CVT_RGB_OUTPUT(r, g, b) { \
#define THEORAPLAY_CVT_RGB_OUTPUT(dst, r, g, b) { \
*(dst++) = (unsigned char) ((r < 0) ? 0 : (r > 255) ? 255 : r); \
*(dst++) = (unsigned char) ((g < 0) ? 0 : (g > 255) ? 255 : g); \
*(dst++) = (unsigned char) ((b < 0) ? 0 : (b > 255) ? 255 : b); \
@ -118,7 +118,7 @@ static unsigned char *ConvertVideoFrame420ToIYUV(const THEORAPLAY_Allocator *all
// BGRA
#define THEORAPLAY_CVT_FNNAME_420 ConvertVideoFrame420ToBGRA
#define THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, h) ((w) * (h) * 4)
#define THEORAPLAY_CVT_RGB_OUTPUT(r, g, b) { \
#define THEORAPLAY_CVT_RGB_OUTPUT(dst, r, g, b) { \
*(dst++) = (unsigned char) ((b < 0) ? 0 : (b > 255) ? 255 : b); \
*(dst++) = (unsigned char) ((g < 0) ? 0 : (g > 255) ? 255 : g); \
*(dst++) = (unsigned char) ((r < 0) ? 0 : (r > 255) ? 255 : r); \
@ -129,7 +129,7 @@ static unsigned char *ConvertVideoFrame420ToIYUV(const THEORAPLAY_Allocator *all
// RGB565
#define THEORAPLAY_CVT_FNNAME_420 ConvertVideoFrame420ToRGB565
#define THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, h) ((w) * (h) * 2)
#define THEORAPLAY_CVT_RGB_OUTPUT(r, g, b) { \
#define THEORAPLAY_CVT_RGB_OUTPUT(dst, r, g, b) { \
unsigned short *dst16 = (unsigned short *) dst; \
const int r5 = ((r < 0) ? 0 : (r > 255) ? 255 : r) >> 3; \
const int g6 = ((g < 0) ? 0 : (g > 255) ? 255 : g) >> 2; \

47
thirdparty/theoraplay/theoraplay_cvtrgb.h vendored
View file

@ -33,6 +33,7 @@ static unsigned char *THEORAPLAY_CVT_FNNAME_420(const THEORAPLAY_Allocator *allo
if (pixels)
{
unsigned char *dst = pixels;
unsigned char *dst2 = dst + THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, 1);
const int ystride = ycbcr[0].stride;
const int cbstride = ycbcr[1].stride;
const unsigned char *py;
@ -70,16 +71,15 @@ static unsigned char *THEORAPLAY_CVT_FNNAME_420(const THEORAPLAY_Allocator *allo
pcr = ycbcr[2].data + cboff;
}
for (posy = 0; posy < h; posy++)
for (posy = 0; posy < h; posy += 2)
{
int posx, poshalfx;
int posx = 0;
int poshalfx;
posx = 0;
for (poshalfx = 0; poshalfx < halfw; poshalfx++, posx += 2)
{
const int pb = pcb[poshalfx] - cbcroffset;
const int pr = pcr[poshalfx] - cbcroffset;
#if 0
const int pb_factored = ((pb * kbfactor) >> FIXED_POINT_BITS);
const int pr_factored = ((pr * krfactor) >> FIXED_POINT_BITS);
const int pg_factored = (((green_krfactor * pr) + (green_kbfactor * pb)) >> FIXED_POINT_BITS);
@ -88,37 +88,38 @@ static unsigned char *THEORAPLAY_CVT_FNNAME_420(const THEORAPLAY_Allocator *allo
const int r1 = y1 + pr_factored;
const int g1 = y1 - pg_factored;
const int b1 = y1 + pb_factored;
THEORAPLAY_CVT_RGB_OUTPUT(r1, g1, b1);
THEORAPLAY_CVT_RGB_OUTPUT(dst, r1, g1, b1);
}
{
const int y2 = ((py[posx+1] - yoffset) * yfactor) >> FIXED_POINT_BITS;
const int r2 = y2 + pr_factored;
const int g2 = y2 - pg_factored;
const int b2 = y2 + pb_factored;
THEORAPLAY_CVT_RGB_OUTPUT(r2, g2, b2);
}
#else
{
const int y1 = ((py[posx] - yoffset) * yfactor) >> FIXED_POINT_BITS;
const int r1 = y1 + ((pr * krfactor) >> FIXED_POINT_BITS);
const int g1 = y1 - (((green_krfactor * pr) + (green_kbfactor * pb)) >> FIXED_POINT_BITS);
const int b1 = y1 + ((pb * kbfactor) >> FIXED_POINT_BITS);
THEORAPLAY_CVT_RGB_OUTPUT(r1, g1, b1);
THEORAPLAY_CVT_RGB_OUTPUT(dst, r2, g2, b2);
}
{
const int y2 = ((py[posx+1] - yoffset) * yfactor) >> FIXED_POINT_BITS;
const int r2 = y2 + ((pr * krfactor) >> FIXED_POINT_BITS);
const int g2 = y2 - (((green_krfactor * pr) + (green_kbfactor * pb)) >> FIXED_POINT_BITS);
const int b2 = y2 + ((pb * kbfactor) >> FIXED_POINT_BITS);
THEORAPLAY_CVT_RGB_OUTPUT(r2, g2, b2);
const int y3 = ((py[ystride+posx] - yoffset) * yfactor) >> FIXED_POINT_BITS;
const int r3 = y3 + pr_factored;
const int g3 = y3 - pg_factored;
const int b3 = y3 + pb_factored;
THEORAPLAY_CVT_RGB_OUTPUT(dst2, r3, g3, b3);
}
{
const int y4 = ((py[ystride+posx+1] - yoffset) * yfactor) >> FIXED_POINT_BITS;
const int r4 = y4 + pr_factored;
const int g4 = y4 - pg_factored;
const int b4 = y4 + pb_factored;
THEORAPLAY_CVT_RGB_OUTPUT(dst2, r4, g4, b4);
}
#endif
} // for
dst += THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, 1);
dst2 += THEORAPLAY_CVT_RGB_DST_BUFFER_SIZE(w, 1);
// adjust to the start of the next line.
py += ystride;
pcb += cbstride * (posy % 2);
pcr += crstride * (posy % 2);
py += ystride * 2;
pcb += cbstride;
pcr += crstride;
} // for
} // if