* $Id$
*
* Authors: Samuel Hocevar <sam@zoy.org>
+ * Damien Fouilleul <damienf@videolan.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
- *
+ *
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
/*****************************************************************************
* Preamble
*****************************************************************************/
-#include <string.h> /* strerror() */
-#include <stdlib.h> /* malloc(), free() */
#include <vlc/vlc.h>
-#include <vlc/vout.h>
+#include <vlc_vout.h>
#include "i422_yuy2.h"
set_description( _("MMX conversions from " SRC_FOURCC " to " DEST_FOURCC) );
set_capability( "chroma", 100 );
add_requirement( MMX );
+#elif defined (MODULE_NAME_IS_i422_yuy2_sse2)
+ set_description( _("SSE2 conversions from " SRC_FOURCC " to " DEST_FOURCC) );
+ set_capability( "chroma", 120 );
+ add_requirement( SSE2 );
#endif
set_callbacks( Activate, NULL );
vlc_module_end();
default:
return -1;
}
-
- return 0;
+ return 0;
}
/* Following functions are local */
static void I422_YUY2( vout_thread_t *p_vout, picture_t *p_source,
picture_t *p_dest )
{
- uint8_t *p_pixels = p_dest->p->p_pixels;
- int i_pitch = p_dest->p->i_pitch;
+ uint8_t *p_line = p_dest->p->p_pixels;
uint8_t *p_y = p_source->Y_PIXELS;
uint8_t *p_u = p_source->U_PIXELS;
uint8_t *p_v = p_source->V_PIXELS;
int i_x, i_y;
+ const int i_source_margin = p_source->p[0].i_pitch
+ - p_source->p[0].i_visible_pitch;
+ const int i_source_margin_c = p_source->p[1].i_pitch
+ - p_source->p[1].i_visible_pitch;
+ const int i_dest_margin = p_dest->p->i_pitch
+ - p_dest->p->i_visible_pitch;
+
+#if defined (MODULE_NAME_IS_i422_yuy2_sse2)
+
+ if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
+ ((intptr_t)p_line|(intptr_t)p_y))) )
+ {
+ /* use faster SSE2 aligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_YUYV_ALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YUYV( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ else {
+ /* use slower SSE2 unaligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_YUYV_UNALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YUYV( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ SSE2_END;
+
+#else
+
for( i_y = p_vout->render.i_height ; i_y-- ; )
{
- uint8_t *p_line = p_pixels;
for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
{
#if defined (MODULE_NAME_IS_i422_yuy2)
C_YUV422_YUYV( p_line, p_y, p_u, p_v );
C_YUV422_YUYV( p_line, p_y, p_u, p_v );
C_YUV422_YUYV( p_line, p_y, p_u, p_v );
-#else
- __asm__( ".p2align 3" MMX_YUV422_YUYV
- : : "r" (p_line), "r" (p_y), "r" (p_u), "r" (p_v) );
-
- p_line += 16; p_y += 8; p_u += 4; p_v += 4;
+#elif defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_CALL( MMX_YUV422_YUYV );
#endif
}
- p_pixels += i_pitch;
+ for( i_x = ( p_vout->render.i_width % 8 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YUYV( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
}
+#if defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_END;
+#endif
+
+#endif
}
/*****************************************************************************
static void I422_YVYU( vout_thread_t *p_vout, picture_t *p_source,
picture_t *p_dest )
{
- uint8_t *p_pixels = p_dest->p->p_pixels;
- int i_pitch = p_dest->p->i_pitch;
+ uint8_t *p_line = p_dest->p->p_pixels;
uint8_t *p_y = p_source->Y_PIXELS;
uint8_t *p_u = p_source->U_PIXELS;
uint8_t *p_v = p_source->V_PIXELS;
int i_x, i_y;
+ const int i_source_margin = p_source->p[0].i_pitch
+ - p_source->p[0].i_visible_pitch;
+ const int i_source_margin_c = p_source->p[1].i_pitch
+ - p_source->p[1].i_visible_pitch;
+ const int i_dest_margin = p_dest->p->i_pitch
+ - p_dest->p->i_visible_pitch;
+
+#if defined (MODULE_NAME_IS_i422_yuy2_sse2)
+
+ if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
+ ((intptr_t)p_line|(intptr_t)p_y))) )
+ {
+ /* use faster SSE2 aligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_YVYU_ALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YVYU( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ else {
+ /* use slower SSE2 unaligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_YVYU_UNALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YVYU( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ SSE2_END;
+
+#else
+
for( i_y = p_vout->render.i_height ; i_y-- ; )
{
- uint8_t *p_line = p_pixels;
for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
{
#if defined (MODULE_NAME_IS_i422_yuy2)
C_YUV422_YVYU( p_line, p_y, p_u, p_v );
C_YUV422_YVYU( p_line, p_y, p_u, p_v );
C_YUV422_YVYU( p_line, p_y, p_u, p_v );
-#else
- __asm__( ".p2align 3" MMX_YUV422_YVYU
- : : "r" (p_line), "r" (p_y), "r" (p_u), "r" (p_v) );
-
- p_line += 16; p_y += 8; p_u += 4; p_v += 4;
+#elif defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_CALL( MMX_YUV422_YVYU );
#endif
}
- p_pixels += i_pitch;
+ for( i_x = ( p_vout->render.i_width % 8 ) / 2; i_x-- ; )
+ {
+ C_YUV422_YVYU( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
}
+#if defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_END;
+#endif
+
+#endif
}
/*****************************************************************************
static void I422_UYVY( vout_thread_t *p_vout, picture_t *p_source,
picture_t *p_dest )
{
- uint8_t *p_pixels = p_dest->p->p_pixels;
- int i_pitch = p_dest->p->i_pitch;
+ uint8_t *p_line = p_dest->p->p_pixels;
uint8_t *p_y = p_source->Y_PIXELS;
uint8_t *p_u = p_source->U_PIXELS;
uint8_t *p_v = p_source->V_PIXELS;
int i_x, i_y;
+ const int i_source_margin = p_source->p[0].i_pitch
+ - p_source->p[0].i_visible_pitch;
+ const int i_source_margin_c = p_source->p[1].i_pitch
+ - p_source->p[1].i_visible_pitch;
+ const int i_dest_margin = p_dest->p->i_pitch
+ - p_dest->p->i_visible_pitch;
+
+#if defined (MODULE_NAME_IS_i422_yuy2_sse2)
+
+ if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
+ ((intptr_t)p_line|(intptr_t)p_y))) )
+ {
+ /* use faster SSE2 aligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_UYVY_ALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ else {
+ /* use slower SSE2 unaligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_UYVY_UNALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ SSE2_END;
+
+#else
+
for( i_y = p_vout->render.i_height ; i_y-- ; )
{
- uint8_t *p_line = p_pixels;
for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
{
#if defined (MODULE_NAME_IS_i422_yuy2)
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
-#else
- __asm__( ".p2align 3" MMX_YUV422_UYVY
- : : "r" (p_line), "r" (p_y), "r" (p_u), "r" (p_v) );
-
- p_line += 16; p_y += 8; p_u += 4; p_v += 4;
+#elif defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_CALL( MMX_YUV422_UYVY );
#endif
}
- p_pixels += i_pitch;
+ for( i_x = ( p_vout->render.i_width % 8 ) / 2; i_x-- ; )
+ {
+ C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
}
+#if defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_END;
+#endif
+
+#endif
}
/*****************************************************************************
int i_x, i_y;
+ const int i_source_margin = p_source->p[0].i_pitch
+ - p_source->p[0].i_visible_pitch;
+ const int i_source_margin_c = p_source->p[1].i_pitch
+ - p_source->p[1].i_visible_pitch;
+ const int i_dest_margin = p_dest->p->i_pitch
+ - p_dest->p->i_visible_pitch;
+
+#if defined (MODULE_NAME_IS_i422_yuy2_sse2)
+
+ if( 0 == (15 & (p_source->p[Y_PLANE].i_pitch|p_dest->p->i_pitch|
+ ((intptr_t)p_line|(intptr_t)p_y))) )
+ {
+ /* use faster SSE2 aligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ p_line -= 2 * p_dest->p->i_pitch;
+
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_UYVY_ALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ else {
+ /* use slower SSE2 unaligned fetch and store */
+ for( i_y = p_vout->render.i_height ; i_y-- ; )
+ {
+ p_line -= 2 * p_dest->p->i_pitch;
+
+ for( i_x = p_vout->render.i_width / 16 ; i_x-- ; )
+ {
+ SSE2_CALL( SSE2_YUV422_UYVY_UNALIGNED );
+ }
+ for( i_x = ( p_vout->render.i_width % 16 ) / 2; i_x-- ; )
+ {
+ C_YUV422_UYVY( p_line, p_y, p_u, p_v );
+ }
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
+ }
+ }
+ SSE2_END;
+
+#else
+
for( i_y = p_vout->render.i_height ; i_y-- ; )
{
for( i_x = p_vout->render.i_width / 8 ; i_x-- ; )
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
C_YUV422_UYVY( p_line, p_y, p_u, p_v );
-#else
- __asm__( ".p2align 3" MMX_YUV422_UYVY
- : : "r" (p_line), "r" (p_y), "r" (p_u), "r" (p_v) );
-
- p_line += 16; p_y += 8; p_u += 4; p_v += 4;
+#elif defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_CALL( MMX_YUV422_UYVY );
#endif
}
+ p_y += i_source_margin;
+ p_u += i_source_margin_c;
+ p_v += i_source_margin_c;
+ p_line += i_dest_margin;
}
+#if defined (MODULE_NAME_IS_i422_yuy2_mmx)
+ MMX_END;
+#elif defined (MODULE_NAME_IS_i422_yuy2_sse2)
+ SSE2_END;
+#endif
+
+#endif
}
/*****************************************************************************
projects / vlc / blobdiff