--- interpolate8x8.h	2002/11/20 19:53:12	1.5.2.6
+++ interpolate8x8.h	2002/12/19 00:39:47	1.5.2.9
@@ -83,6 +83,10 @@
 INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn;
 INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn;
 
+INTERPOLATE8X8 interpolate8x8_halfpel_h_3dne;
+INTERPOLATE8X8 interpolate8x8_halfpel_v_3dne;
+INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dne;
+
 INTERPOLATE8X8 interpolate8x8_halfpel_h_ia64;
 INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64;
 INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64;
@@ -155,6 +159,41 @@
 	}
 }
 
+
+static __inline void
+interpolate16x16_switch(uint8_t * const cur,
+					  const uint8_t * const refn,
+					  const uint32_t x,
+					  const uint32_t y,
+					  const int32_t dx,
+					  const int dy,
+					  const uint32_t stride,
+					  const uint32_t rounding)
+{
+	interpolate8x8_switch(cur, refn, x,   y,   dx, dy, stride, rounding);
+	interpolate8x8_switch(cur, refn, x+8, y,   dx, dy, stride, rounding);
+	interpolate8x8_switch(cur, refn, x,   y+8, dx, dy, stride, rounding);
+	interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding);
+}
+
+
+static __inline void
+interpolate32x32_switch(uint8_t * const cur,
+					  const uint8_t * const refn,
+					  const uint32_t x,
+					  const uint32_t y,
+					  const int32_t dx,
+					  const int dy,
+					  const uint32_t stride,
+					  const uint32_t rounding)
+{
+	interpolate16x16_switch(cur, refn, x,    y,    dx, dy, stride, rounding);
+	interpolate16x16_switch(cur, refn, x+16, y,    dx, dy, stride, rounding);
+	interpolate16x16_switch(cur, refn, x,    y+16, dx, dy, stride, rounding);
+	interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding);
+}
+
+
 static __inline uint8_t *
 interpolate8x8_switch2(uint8_t * const buffer,
 					  const uint8_t * const refn,
@@ -243,7 +282,7 @@
 
 	case 1:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8);
   		break;
   
 	case 2:
@@ -252,31 +291,31 @@
   
 	case 3:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8);
   		break;
   
 	case 4:
 		interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding);
-		interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8);
  		break;
    
 	case 5:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
 		interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
 		interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8);
  		break;
    
 	case 6:
 		interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8);
  		break;
    
 	case 7:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
 		interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9);
 		interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8);
  		break;
    
 	case 8:
@@ -301,26 +340,26 @@
  
 	case 12:
 		interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding);
-		interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8);
    		break;
  
 	case 13:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
 		interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9);
 		interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8);
  		break;
    
 	case 14:
 		interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8);
  		break;
    
 	case 15:
 		interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding);
 		interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9);
 		interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8);
 		break;
 	}
 }
@@ -372,10 +411,10 @@
 
 	case 1:
 		interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8);
 		break;
 
 	case 2:
@@ -384,18 +423,18 @@
   
 	case 3:
 		interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8);
 		break;
 
 	case 4:
 		interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding);
-		interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
 		break;
 	
 	case 5:
@@ -406,18 +445,18 @@
 		interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
 
 		interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
 		break;
    
 	case 6:
 		interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8);
 		break;
    
 	case 7:
@@ -428,10 +467,10 @@
 		interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
 
 		interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
 		break;
    
 	case 8:
@@ -462,10 +501,10 @@
  
 	case 12:
 		interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding);
-		interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8);
 		break;
  
 	case 13:
@@ -476,18 +515,18 @@
 		interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
 
 		interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8);
 		break;
    
 	case 14:
 		interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8);
 		break;
    
 	case 15:
@@ -498,10 +537,10 @@
 		interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9);
 
 		interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding);
-		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 9);
-		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 9);
+		interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8);
+		interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8);
 		break;
 	}
 }