--- xvid.c	2003/11/30 16:13:15	1.45.2.22
+++ xvid.c	2004/04/14 19:51:06	1.48.2.3
@@ -19,7 +19,7 @@
  *  along with this program ; if not, write to the Free Software
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  *
- * $Id: xvid.c,v 1.45.2.22 2003/11/30 16:13:15 edgomez Exp $
+ * $Id: xvid.c,v 1.48.2.3 2004/04/14 19:51:06 edgomez Exp $
  *
  ****************************************************************************/
 
@@ -46,6 +46,7 @@
 #include "utils/timer.h"
 #include "bitstream/mbcoding.h"
 #include "image/qpel.h"
+#include "image/postprocessing.h"
 
 #if defined(_DEBUG)
 unsigned int xvid_debug = 0; /* xvid debug mask */
@@ -247,6 +248,7 @@
 	bgra_to_yv12    = bgra_to_yv12_c;
 	abgr_to_yv12    = abgr_to_yv12_c;
 	rgba_to_yv12    = rgba_to_yv12_c;
+	argb_to_yv12    = argb_to_yv12_c;
 	yuyv_to_yv12    = yuyv_to_yv12_c;
 	uyvy_to_yv12    = uyvy_to_yv12_c;
 
@@ -256,10 +258,10 @@
 	bgrai_to_yv12   = bgrai_to_yv12_c;
 	abgri_to_yv12   = abgri_to_yv12_c;
 	rgbai_to_yv12   = rgbai_to_yv12_c;
+	argbi_to_yv12   = argbi_to_yv12_c;
 	yuyvi_to_yv12   = yuyvi_to_yv12_c;
 	uyvyi_to_yv12   = uyvyi_to_yv12_c;
 
-
 	/* All colorspace transformation functions YV12->User format */
 	yv12_to_rgb555  = yv12_to_rgb555_c;
 	yv12_to_rgb565  = yv12_to_rgb565_c;
@@ -267,6 +269,7 @@
 	yv12_to_bgra    = yv12_to_bgra_c;
 	yv12_to_abgr    = yv12_to_abgr_c;
 	yv12_to_rgba    = yv12_to_rgba_c;
+	yv12_to_argb    = yv12_to_argb_c;
 	yv12_to_yuyv    = yv12_to_yuyv_c;
 	yv12_to_uyvy    = yv12_to_uyvy_c;
 
@@ -276,6 +279,7 @@
 	yv12_to_bgrai   = yv12_to_bgrai_c;
 	yv12_to_abgri   = yv12_to_abgri_c;
 	yv12_to_rgbai   = yv12_to_rgbai_c;
+	yv12_to_argbi   = yv12_to_argbi_c;
 	yv12_to_yuyvi   = yv12_to_yuyvi_c;
 	yv12_to_uyvyi   = yv12_to_uyvyi_c;
 
@@ -445,42 +449,45 @@
 
 	if ((cpu_flags & XVID_CPU_3DNOWEXT)) {
 
-		/* Inverse DCT */
-		idct =  idct_3dne;
-
 		/* Buffer transfer */
 		transfer_8to16copy =  transfer_8to16copy_3dne;
 		transfer_16to8copy = transfer_16to8copy_3dne;
 		transfer_8to16sub =  transfer_8to16sub_3dne;
 		transfer_8to16subro =  transfer_8to16subro_3dne;
-		transfer_8to16sub2 =  transfer_8to16sub2_3dne;
 		transfer_16to8add = transfer_16to8add_3dne;
 		transfer8x8_copy = transfer8x8_copy_3dne;
 
-		/* Quantization */
-		quant_h263_intra = quant_h263_intra_3dne;
-		quant_h263_inter = quant_h263_inter_3dne;
-		dequant_mpeg_intra = dequant_mpeg_intra_3dne;
-		dequant_mpeg_inter = dequant_mpeg_inter_3dne;
-		dequant_h263_intra = dequant_h263_intra_3dne;
-		dequant_h263_inter = dequant_h263_inter_3dne;
-
-		/* ME functions */
-		calc_cbp = calc_cbp_3dne;
-		sad16 = sad16_3dne;
-		sad8 = sad8_3dne;
-		sad16bi = sad16bi_3dne;
-		sad8bi = sad8bi_3dne;
-		dev16 = dev16_3dne;
-
-		/* Interpolation */
-		interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
-		interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
-		interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
+		if ((cpu_flags & XVID_CPU_MMXEXT)) {
+			/* Inverse DCT */
+			idct =  idct_3dne;
+
+			/* Buffer transfer */
+			transfer_8to16sub2 =  transfer_8to16sub2_3dne;
+
+			/* Interpolation */
+			interpolate8x8_halfpel_h = interpolate8x8_halfpel_h_3dne;
+			interpolate8x8_halfpel_v = interpolate8x8_halfpel_v_3dne;
+			interpolate8x8_halfpel_hv = interpolate8x8_halfpel_hv_3dne;
+
+			/* Quantization */
+			quant_h263_intra = quant_h263_intra_3dne;		/* cmov only */
+			quant_h263_inter = quant_h263_inter_3dne;
+			dequant_mpeg_intra = dequant_mpeg_intra_3dne;	/* cmov only */
+			dequant_mpeg_inter = dequant_mpeg_inter_3dne;
+			dequant_h263_intra = dequant_h263_intra_3dne;
+			dequant_h263_inter = dequant_h263_inter_3dne;
+
+			/* ME functions */
+			calc_cbp = calc_cbp_3dne;
+
+			sad16 = sad16_3dne;
+			sad8 = sad8_3dne;
+			sad16bi = sad16bi_3dne;
+			sad8bi = sad8bi_3dne;
+			dev16 = dev16_3dne;
+		}
 	}
 
-#if defined(EXPERIMENTAL_SSE2_CODE) /* mark the whole SSE2 stuff as experimental. At least on
-									   my P4, it crashes... */
 	if ((cpu_flags & XVID_CPU_SSE2)) {
 
 		calc_cbp = calc_cbp_sse2;
@@ -495,12 +502,11 @@
 		sad16    = sad16_sse2;
 		dev16    = dev16_sse2;
 
-		/* DCT operators */
+		/* DCT operators
+		 * no iDCT because it's not "Walken matching" */
 		fdct = fdct_sse2_skal;
-		idct = idct_sse2_dmitry;
 	}
-#endif
-#endif
+#endif /* ARCH_IS_IA32 */
 
 #if defined(ARCH_IS_IA64)
 	if ((cpu_flags & XVID_CPU_ASM)) { /* use assembler routines? */