Skip to content

W
wine-winehq
Commit a875e3c6 authored by Huw Davies's avatar Huw Davies Committed by Alexandre Julliard
Browse files
Options

gdi32: Unroll the start- and end-of-line loops.

Signed-off-by: 's avatarHuw Davies <huw@codeweavers.com> Signed-off-by: 's avatarAlexandre Julliard <julliard@winehq.org>
parent 45b848cc
Show whitespace changes
Inline Side-by-side
Showing with 312 additions and 23 deletions
dlls/gdi32/dibdrv/primitives.c
View file @ a875e3c6
...@@ -5304,17 +5304,46 @@ static void mask_rect_32( const dib_info *dst, const RECT *rc, ...@@ -5304,17 +5304,46 @@ static void mask_rect_32( const dib_info *dst, const RECT *rc,
return; return;
} }
full = ((rc->right - rc->left) - ((8 - (origin->x & 7)) & 7)) / 8;
for (y = rc->top; y < rc->bottom; y++) for (y = rc->top; y < rc->bottom; y++)
{ {
pos = origin->x & 7; pos = origin->x & 7;
src_val = src_start[pos / 8];
x = 0;
for (x = 0; x < rc->right - rc->left && pos < 8; x++, pos++) switch (pos & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; case 1:
do_rop_codes_32( dst_start + x, dst_colors[bit_val], &codes ); bit_val = src_val >> 6;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = src_val >> 5;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = src_val >> 4;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = src_val >> 3;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = src_val >> 2;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = src_val >> 1;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 7:
bit_val = src_val;
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
pos = (pos + 7) & ~7;
} }
full = ((rc->right - rc->left) - x) / 8;
for (i = 0; i < full; i++, pos += 8) for (i = 0; i < full; i++, pos += 8)
{ {
src_val = src_start[pos / 8]; src_val = src_start[pos / 8];
...@@ -5337,11 +5366,42 @@ static void mask_rect_32( const dib_info *dst, const RECT *rc, ...@@ -5337,11 +5366,42 @@ static void mask_rect_32( const dib_info *dst, const RECT *rc,
do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes ); do_rop_codes_32( dst_start + x++, dst_colors[bit_val], &codes );
} }
for ( ; x < rc->right - rc->left; x++, pos++) if (origin_end & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; src_val = src_start[pos / 8];
x += (origin_end & 7) - 1;
switch (origin_end & 7)
{
case 7:
bit_val = src_val >> 1;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = src_val >> 2;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = src_val >> 3;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = src_val >> 4;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = src_val >> 5;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = src_val >> 6;
do_rop_codes_32( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 1:
bit_val = src_val >> 7;
do_rop_codes_32( dst_start + x, dst_colors[bit_val], &codes ); do_rop_codes_32( dst_start + x, dst_colors[bit_val], &codes );
} }
}
dst_start += dst->stride / 4; dst_start += dst->stride / 4;
src_start += src->stride; src_start += src->stride;
...@@ -5381,20 +5441,74 @@ static void mask_rect_24( const dib_info *dst, const RECT *rc, ...@@ -5381,20 +5441,74 @@ static void mask_rect_24( const dib_info *dst, const RECT *rc,
return; return;
} }
full = ((rc->right - rc->left) - ((8 - (origin->x & 7)) & 7)) / 8;
for (y = rc->top; y < rc->bottom; y++) for (y = rc->top; y < rc->bottom; y++)
{ {
pos = origin->x & 7; pos = origin->x & 7;
src_val = src_start[pos / 8];
x = 0;
for (x = 0; x < rc->right - rc->left && pos < 8; x++, pos++) switch (pos & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; case 1:
bit_val = (src_val >> 6) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 2:
bit_val = (src_val >> 5) & 1;
rgb = color_table[bit_val]; rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes ); do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes ); do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes ); do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 3:
bit_val = (src_val >> 4) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 4:
bit_val = (src_val >> 3) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 5:
bit_val = (src_val >> 2) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 6:
bit_val = (src_val >> 1) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
/* fall through */
case 7:
bit_val = src_val & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x++;
pos = (pos + 7) & ~7;
} }
full = ((rc->right - rc->left) - x) / 8;
for (i = 0; i < full; i++, pos += 8) for (i = 0; i < full; i++, pos += 8)
{ {
src_val = src_start[pos / 8]; src_val = src_start[pos / 8];
...@@ -5456,14 +5570,69 @@ static void mask_rect_24( const dib_info *dst, const RECT *rc, ...@@ -5456,14 +5570,69 @@ static void mask_rect_24( const dib_info *dst, const RECT *rc,
x++; x++;
} }
for ( ; x < rc->right - rc->left; x++, pos++) if (origin_end & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; src_val = src_start[pos / 8];
x += (origin_end & 7) - 1;
switch (origin_end & 7)
{
case 7:
bit_val = (src_val >> 1) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 6:
bit_val = (src_val >> 2) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 5:
bit_val = (src_val >> 3) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 4:
bit_val = (src_val >> 4) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 3:
bit_val = (src_val >> 5) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 2:
bit_val = (src_val >> 6) & 1;
rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
x--;
/* fall through */
case 1:
bit_val = (src_val >> 7) & 1;
rgb = color_table[bit_val]; rgb = color_table[bit_val];
do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes ); do_rop_codes_8( dst_start + x * 3, rgb.rgbBlue, &codes );
do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes ); do_rop_codes_8( dst_start + x * 3 + 1, rgb.rgbGreen, &codes );
do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes ); do_rop_codes_8( dst_start + x * 3 + 2, rgb.rgbRed, &codes );
} }
}
dst_start += dst->stride; dst_start += dst->stride;
src_start += src->stride; src_start += src->stride;
...@@ -5510,17 +5679,46 @@ static void mask_rect_16( const dib_info *dst, const RECT *rc, ...@@ -5510,17 +5679,46 @@ static void mask_rect_16( const dib_info *dst, const RECT *rc,
return; return;
} }
full = ((rc->right - rc->left) - ((8 - (origin->x & 7)) & 7)) / 8;
for (y = rc->top; y < rc->bottom; y++) for (y = rc->top; y < rc->bottom; y++)
{ {
pos = origin->x & 7; pos = origin->x & 7;
src_val = src_start[pos / 8];
x = 0;
for (x = 0; x < rc->right - rc->left && pos < 8; x++, pos++) switch (pos & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; case 1:
do_rop_codes_16( dst_start + x, dst_colors[bit_val], &codes ); bit_val = (src_val >> 6) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = (src_val >> 5) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = (src_val >> 4) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = (src_val >> 3) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = (src_val >> 2) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = (src_val >> 1) & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 7:
bit_val = src_val & 1;
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
pos = (pos + 7) & ~7;
} }
full = ((rc->right - rc->left) - x) / 8;
for (i = 0; i < full; i++, pos += 8) for (i = 0; i < full; i++, pos += 8)
{ {
src_val = src_start[pos / 8]; src_val = src_start[pos / 8];
...@@ -5543,11 +5741,42 @@ static void mask_rect_16( const dib_info *dst, const RECT *rc, ...@@ -5543,11 +5741,42 @@ static void mask_rect_16( const dib_info *dst, const RECT *rc,
do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes ); do_rop_codes_16( dst_start + x++, dst_colors[bit_val], &codes );
} }
for ( ; x < rc->right - rc->left; x++, pos++) if (origin_end & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; src_val = src_start[pos / 8];
x += (origin_end & 7) - 1;
switch (origin_end & 7)
{
case 7:
bit_val = (src_val >> 1) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = (src_val >> 2) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = (src_val >> 3) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = (src_val >> 4) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = (src_val >> 5) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = (src_val >> 6) & 1;
do_rop_codes_16( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 1:
bit_val = (src_val >> 7) & 1;
do_rop_codes_16( dst_start + x, dst_colors[bit_val], &codes ); do_rop_codes_16( dst_start + x, dst_colors[bit_val], &codes );
} }
}
dst_start += dst->stride / 2; dst_start += dst->stride / 2;
src_start += src->stride; src_start += src->stride;
...@@ -5587,17 +5816,46 @@ static void mask_rect_8( const dib_info *dst, const RECT *rc, ...@@ -5587,17 +5816,46 @@ static void mask_rect_8( const dib_info *dst, const RECT *rc,
return; return;
} }
full = ((rc->right - rc->left) - ((8 - (origin->x & 7)) & 7)) / 8;
for (y = rc->top; y < rc->bottom; y++) for (y = rc->top; y < rc->bottom; y++)
{ {
pos = origin->x & 7; pos = origin->x & 7;
src_val = src_start[pos / 8];
x = 0;
for (x = 0; x < rc->right - rc->left && pos < 8; x++, pos++) switch (pos & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; case 1:
do_rop_codes_8( dst_start + x, dst_colors[bit_val], &codes ); bit_val = (src_val >> 6) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = (src_val >> 5) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = (src_val >> 4) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = (src_val >> 3) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = (src_val >> 2) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = (src_val >> 1) & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
/* fall through */
case 7:
bit_val = src_val & 1;
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
pos = (pos + 7) & ~7;
} }
full = ((rc->right - rc->left) - x) / 8;
for (i = 0; i < full; i++, pos += 8) for (i = 0; i < full; i++, pos += 8)
{ {
src_val = src_start[pos / 8]; src_val = src_start[pos / 8];
...@@ -5620,11 +5878,42 @@ static void mask_rect_8( const dib_info *dst, const RECT *rc, ...@@ -5620,11 +5878,42 @@ static void mask_rect_8( const dib_info *dst, const RECT *rc,
do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes ); do_rop_codes_8( dst_start + x++, dst_colors[bit_val], &codes );
} }
for ( ; x < rc->right - rc->left; x++, pos++) if (origin_end & 7)
{ {
bit_val = (src_start[pos / 8] & pixel_masks_1[pos % 8]) ? 1 : 0; src_val = src_start[pos / 8];
x += (origin_end & 7) - 1;
switch (origin_end & 7)
{
case 7:
bit_val = (src_val >> 1) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 6:
bit_val = (src_val >> 2) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 5:
bit_val = (src_val >> 3) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 4:
bit_val = (src_val >> 4) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 3:
bit_val = (src_val >> 5) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 2:
bit_val = (src_val >> 6) & 1;
do_rop_codes_8( dst_start + x--, dst_colors[bit_val], &codes );
/* fall through */
case 1:
bit_val = (src_val >> 7) & 1;
do_rop_codes_8( dst_start + x, dst_colors[bit_val], &codes ); do_rop_codes_8( dst_start + x, dst_colors[bit_val], &codes );
} }
}
dst_start += dst->stride; dst_start += dst->stride;
src_start += src->stride; src_start += src->stride;
......
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