Speed up HQ2x/HQ3x C++ version a bit (about 20% faster with -O3 on my system)

svn-id: r36087

4 files changed, 207 insertions, 101 deletions

diff --git a/graphics/colormasks.h b/graphics/colormasks.h
index 9330585b6b..0cd8e86050 100644
--- a/graphics/colormasks.h
+++ b/graphics/colormasks.h
@@ -94,7 +94,10 @@ struct ColorMasks<565> {
 		kBlueMask   = ((1 << kBlueBits) - 1) << kBlueShift,
 
 		kRedBlueMask = kRedMask | kBlueMask,
-		kLowBits    = (1 << kRedShift) | (1 << kGreenShift) | (1 << kBlueShift)
+
+		kLowBits    = (1 << kRedShift) | (1 << kGreenShift) | (1 << kBlueShift),
+		kLow2Bits   = (3 << kRedShift) | (3 << kGreenShift) | (3 << kBlueShift),
+		kLow3Bits   = (7 << kRedShift) | (7 << kGreenShift) | (7 << kBlueShift)
 	};
 };
 
@@ -125,7 +128,10 @@ struct ColorMasks<555> {
 		kBlueMask  = ((1 << kBlueBits) - 1) << kBlueShift,
 
 		kRedBlueMask = kRedMask | kBlueMask,
-		kLowBits    = (1 << kRedShift) | (1 << kGreenShift) | (1 << kBlueShift)
+
+		kLowBits    = (1 << kRedShift) | (1 << kGreenShift) | (1 << kBlueShift),
+		kLow2Bits   = (3 << kRedShift) | (3 << kGreenShift) | (3 << kBlueShift),
+		kLow3Bits   = (7 << kRedShift) | (7 << kGreenShift) | (7 << kBlueShift)
 	};
 };
 
diff --git a/graphics/scaler/hq2x.cpp b/graphics/scaler/hq2x.cpp
index bef97e7251..df1c34626f 100644
--- a/graphics/scaler/hq2x.cpp
+++ b/graphics/scaler/hq2x.cpp
@@ -38,78 +38,78 @@ void hq2x_16(const byte *, byte *, uint32, uint32, uint32, uint32);
 
 }
 
-void HQ2x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
+void HQ2x_ASM(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	hq2x_16(srcPtr, dstPtr, width, height, srcPitch, dstPitch);
 }
 
-#else
+#endif
 
 #define PIXEL00_0	*(q) = w5;
-#define PIXEL00_10	*(q) = interpolate16_2<bitFormat,3,1>(w5, w1);
-#define PIXEL00_11	*(q) = interpolate16_2<bitFormat,3,1>(w5, w4);
-#define PIXEL00_12	*(q) = interpolate16_2<bitFormat,3,1>(w5, w2);
-#define PIXEL00_20	*(q) = interpolate16_3<bitFormat,2,1,1>(w5, w4, w2);
-#define PIXEL00_21	*(q) = interpolate16_3<bitFormat,2,1,1>(w5, w1, w2);
-#define PIXEL00_22	*(q) = interpolate16_3<bitFormat,2,1,1>(w5, w1, w4);
-#define PIXEL00_60	*(q) = interpolate16_3<bitFormat,5,2,1>(w5, w2, w4);
-#define PIXEL00_61	*(q) = interpolate16_3<bitFormat,5,2,1>(w5, w4, w2);
-#define PIXEL00_70	*(q) = interpolate16_3<bitFormat,6,1,1>(w5, w4, w2);
-#define PIXEL00_90	*(q) = interpolate16_3<bitFormat,2,3,3>(w5, w4, w2);
-#define PIXEL00_100	*(q) = interpolate16_3<bitFormat,14,1,1>(w5, w4, w2);
+#define PIXEL00_10	*(q) = interpolate16_3_1<ColorMask>(w5, w1);
+#define PIXEL00_11	*(q) = interpolate16_3_1<ColorMask>(w5, w4);
+#define PIXEL00_12	*(q) = interpolate16_3_1<ColorMask>(w5, w2);
+#define PIXEL00_20	*(q) = interpolate16_2_1_1<ColorMask>(w5, w4, w2);
+#define PIXEL00_21	*(q) = interpolate16_2_1_1<ColorMask>(w5, w1, w2);
+#define PIXEL00_22	*(q) = interpolate16_2_1_1<ColorMask>(w5, w1, w4);
+#define PIXEL00_60	*(q) = interpolate16_5_2_1<ColorMask>(w5, w2, w4);
+#define PIXEL00_61	*(q) = interpolate16_5_2_1<ColorMask>(w5, w4, w2);
+#define PIXEL00_70	*(q) = interpolate16_6_1_1<ColorMask>(w5, w4, w2);
+#define PIXEL00_90	*(q) = interpolate16_2_3_3<ColorMask>(w5, w4, w2);
+#define PIXEL00_100	*(q) = interpolate16_14_1_1<ColorMask>(w5, w4, w2);
 
 #define PIXEL01_0	*(q+1) = w5;
-#define PIXEL01_10	*(q+1) = interpolate16_2<bitFormat,3,1>(w5, w3);
-#define PIXEL01_11	*(q+1) = interpolate16_2<bitFormat,3,1>(w5, w2);
-#define PIXEL01_12	*(q+1) = interpolate16_2<bitFormat,3,1>(w5, w6);
-#define PIXEL01_20	*(q+1) = interpolate16_3<bitFormat,2,1,1>(w5, w2, w6);
-#define PIXEL01_21	*(q+1) = interpolate16_3<bitFormat,2,1,1>(w5, w3, w6);
-#define PIXEL01_22	*(q+1) = interpolate16_3<bitFormat,2,1,1>(w5, w3, w2);
-#define PIXEL01_60	*(q+1) = interpolate16_3<bitFormat,5,2,1>(w5, w6, w2);
-#define PIXEL01_61	*(q+1) = interpolate16_3<bitFormat,5,2,1>(w5, w2, w6);
-#define PIXEL01_70	*(q+1) = interpolate16_3<bitFormat,6,1,1>(w5, w2, w6);
-#define PIXEL01_90	*(q+1) = interpolate16_3<bitFormat,2,3,3>(w5, w2, w6);
-#define PIXEL01_100	*(q+1) = interpolate16_3<bitFormat,14,1,1>(w5, w2, w6);
+#define PIXEL01_10	*(q+1) = interpolate16_3_1<ColorMask>(w5, w3);
+#define PIXEL01_11	*(q+1) = interpolate16_3_1<ColorMask>(w5, w2);
+#define PIXEL01_12	*(q+1) = interpolate16_3_1<ColorMask>(w5, w6);
+#define PIXEL01_20	*(q+1) = interpolate16_2_1_1<ColorMask>(w5, w2, w6);
+#define PIXEL01_21	*(q+1) = interpolate16_2_1_1<ColorMask>(w5, w3, w6);
+#define PIXEL01_22	*(q+1) = interpolate16_2_1_1<ColorMask>(w5, w3, w2);
+#define PIXEL01_60	*(q+1) = interpolate16_5_2_1<ColorMask>(w5, w6, w2);
+#define PIXEL01_61	*(q+1) = interpolate16_5_2_1<ColorMask>(w5, w2, w6);
+#define PIXEL01_70	*(q+1) = interpolate16_6_1_1<ColorMask>(w5, w2, w6);
+#define PIXEL01_90	*(q+1) = interpolate16_2_3_3<ColorMask>(w5, w2, w6);
+#define PIXEL01_100	*(q+1) = interpolate16_14_1_1<ColorMask>(w5, w2, w6);
 
 #define PIXEL10_0	*(q+nextlineDst) = w5;
-#define PIXEL10_10	*(q+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w7);
-#define PIXEL10_11	*(q+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w8);
-#define PIXEL10_12	*(q+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w4);
-#define PIXEL10_20	*(q+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w8, w4);
-#define PIXEL10_21	*(q+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w7, w4);
-#define PIXEL10_22	*(q+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w7, w8);
-#define PIXEL10_60	*(q+nextlineDst) = interpolate16_3<bitFormat,5,2,1>(w5, w4, w8);
-#define PIXEL10_61	*(q+nextlineDst) = interpolate16_3<bitFormat,5,2,1>(w5, w8, w4);
-#define PIXEL10_70	*(q+nextlineDst) = interpolate16_3<bitFormat,6,1,1>(w5, w8, w4);
-#define PIXEL10_90	*(q+nextlineDst) = interpolate16_3<bitFormat,2,3,3>(w5, w8, w4);
-#define PIXEL10_100	*(q+nextlineDst) = interpolate16_3<bitFormat,14,1,1>(w5, w8, w4);
+#define PIXEL10_10	*(q+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w7);
+#define PIXEL10_11	*(q+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w8);
+#define PIXEL10_12	*(q+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w4);
+#define PIXEL10_20	*(q+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w8, w4);
+#define PIXEL10_21	*(q+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w7, w4);
+#define PIXEL10_22	*(q+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w7, w8);
+#define PIXEL10_60	*(q+nextlineDst) = interpolate16_5_2_1<ColorMask>(w5, w4, w8);
+#define PIXEL10_61	*(q+nextlineDst) = interpolate16_5_2_1<ColorMask>(w5, w8, w4);
+#define PIXEL10_70	*(q+nextlineDst) = interpolate16_6_1_1<ColorMask>(w5, w8, w4);
+#define PIXEL10_90	*(q+nextlineDst) = interpolate16_2_3_3<ColorMask>(w5, w8, w4);
+#define PIXEL10_100	*(q+nextlineDst) = interpolate16_14_1_1<ColorMask>(w5, w8, w4);
 
 #define PIXEL11_0	*(q+1+nextlineDst) = w5;
-#define PIXEL11_10	*(q+1+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w9);
-#define PIXEL11_11	*(q+1+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w6);
-#define PIXEL11_12	*(q+1+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w8);
-#define PIXEL11_20	*(q+1+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w6, w8);
-#define PIXEL11_21	*(q+1+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w9, w8);
-#define PIXEL11_22	*(q+1+nextlineDst) = interpolate16_3<bitFormat,2,1,1>(w5, w9, w6);
-#define PIXEL11_60	*(q+1+nextlineDst) = interpolate16_3<bitFormat,5,2,1>(w5, w8, w6);
-#define PIXEL11_61	*(q+1+nextlineDst) = interpolate16_3<bitFormat,5,2,1>(w5, w6, w8);
-#define PIXEL11_70	*(q+1+nextlineDst) = interpolate16_3<bitFormat,6,1,1>(w5, w6, w8);
-#define PIXEL11_90	*(q+1+nextlineDst) = interpolate16_3<bitFormat,2,3,3>(w5, w6, w8);
-#define PIXEL11_100	*(q+1+nextlineDst) = interpolate16_3<bitFormat,14,1,1>(w5, w6, w8);
+#define PIXEL11_10	*(q+1+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w9);
+#define PIXEL11_11	*(q+1+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w6);
+#define PIXEL11_12	*(q+1+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w8);
+#define PIXEL11_20	*(q+1+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w6, w8);
+#define PIXEL11_21	*(q+1+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w9, w8);
+#define PIXEL11_22	*(q+1+nextlineDst) = interpolate16_2_1_1<ColorMask>(w5, w9, w6);
+#define PIXEL11_60	*(q+1+nextlineDst) = interpolate16_5_2_1<ColorMask>(w5, w8, w6);
+#define PIXEL11_61	*(q+1+nextlineDst) = interpolate16_5_2_1<ColorMask>(w5, w6, w8);
+#define PIXEL11_70	*(q+1+nextlineDst) = interpolate16_6_1_1<ColorMask>(w5, w6, w8);
+#define PIXEL11_90	*(q+1+nextlineDst) = interpolate16_2_3_3<ColorMask>(w5, w6, w8);
+#define PIXEL11_100	*(q+1+nextlineDst) = interpolate16_14_1_1<ColorMask>(w5, w6, w8);
 
 #define YUV(x)	RGBtoYUV[w ## x]
 
 
-#define bitFormat 565
+#define ColorMask Graphics::ColorMasks<565>
 void HQ2x_565(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	#include "graphics/scaler/hq2x.h"
 }
-#undef bitFormat
+#undef ColorMask
 
-#define bitFormat 555
+#define ColorMask Graphics::ColorMasks<555>
 void HQ2x_555(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	#include "graphics/scaler/hq2x.h"
 }
-#undef bitFormat
+#undef ColorMask
 
 
 void HQ2x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
@@ -118,5 +118,3 @@ void HQ2x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
 	else
 		HQ2x_555(srcPtr, srcPitch, dstPtr, dstPitch, width, height);
 }
-
-#endif //Assembly version
diff --git a/graphics/scaler/hq3x.cpp b/graphics/scaler/hq3x.cpp
index b225fcae87..b40c85d56a 100644
--- a/graphics/scaler/hq3x.cpp
+++ b/graphics/scaler/hq3x.cpp
@@ -39,80 +39,80 @@ void hq3x_16(const byte *, byte *, uint32, uint32, uint32, uint32);
 
 }
 
-void HQ3x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
+void HQ3x_ASM(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	hq3x_16(srcPtr, dstPtr, width, height, srcPitch, dstPitch);
 }
 
-#else
+#endif
 
-#define PIXEL00_1M  *(q) = interpolate16_2<bitFormat,3,1>(w5, w1);
-#define PIXEL00_1U  *(q) = interpolate16_2<bitFormat,3,1>(w5, w2);
-#define PIXEL00_1L  *(q) = interpolate16_2<bitFormat,3,1>(w5, w4);
-#define PIXEL00_2   *(q) = interpolate16_3<bitFormat,2,1,1>(w5, w4, w2);
-#define PIXEL00_4   *(q) = interpolate16_3<bitFormat,2,7,7>(w5, w4, w2);
-#define PIXEL00_5   *(q) = interpolate16_2<bitFormat,1,1>(w4, w2);
+#define PIXEL00_1M  *(q) = interpolate16_3_1<ColorMask>(w5, w1);
+#define PIXEL00_1U  *(q) = interpolate16_3_1<ColorMask>(w5, w2);
+#define PIXEL00_1L  *(q) = interpolate16_3_1<ColorMask>(w5, w4);
+#define PIXEL00_2   *(q) = interpolate16_2_1_1<ColorMask>(w5, w4, w2);
+#define PIXEL00_4   *(q) = interpolate16_2_7_7<ColorMask>(w5, w4, w2);
+#define PIXEL00_5   *(q) = interpolate16_1_1<ColorMask>(w4, w2);
 #define PIXEL00_C   *(q) = w5;
 
-#define PIXEL01_1   *(q+1) = interpolate16_2<bitFormat,3,1>(w5, w2);
-#define PIXEL01_3   *(q+1) = interpolate16_2<bitFormat,7,1>(w5, w2);
-#define PIXEL01_6   *(q+1) = interpolate16_2<bitFormat,3,1>(w2, w5);
+#define PIXEL01_1   *(q+1) = interpolate16_3_1<ColorMask>(w5, w2);
+#define PIXEL01_3   *(q+1) = interpolate16_7_1<ColorMask>(w5, w2);
+#define PIXEL01_6   *(q+1) = interpolate16_3_1<ColorMask>(w2, w5);
 #define PIXEL01_C   *(q+1) = w5;
 
-#define PIXEL02_1M  *(q+2) = interpolate16_2<bitFormat,3,1>(w5, w3);
-#define PIXEL02_1U  *(q+2) = interpolate16_2<bitFormat,3,1>(w5, w2);
-#define PIXEL02_1R  *(q+2) = interpolate16_2<bitFormat,3,1>(w5, w6);
-#define PIXEL02_2   *(q+2) = interpolate16_3<bitFormat,2,1,1>(w5, w2, w6);
-#define PIXEL02_4   *(q+2) = interpolate16_3<bitFormat,2,7,7>(w5, w2, w6);
-#define PIXEL02_5   *(q+2) = interpolate16_2<bitFormat,1,1>(w2, w6);
+#define PIXEL02_1M  *(q+2) = interpolate16_3_1<ColorMask>(w5, w3);
+#define PIXEL02_1U  *(q+2) = interpolate16_3_1<ColorMask>(w5, w2);
+#define PIXEL02_1R  *(q+2) = interpolate16_3_1<ColorMask>(w5, w6);
+#define PIXEL02_2   *(q+2) = interpolate16_2_1_1<ColorMask>(w5, w2, w6);
+#define PIXEL02_4   *(q+2) = interpolate16_2_7_7<ColorMask>(w5, w2, w6);
+#define PIXEL02_5   *(q+2) = interpolate16_1_1<ColorMask>(w2, w6);
 #define PIXEL02_C   *(q+2) = w5;
 
-#define PIXEL10_1   *(q+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w4);
-#define PIXEL10_3   *(q+nextlineDst) = interpolate16_2<bitFormat,7,1>(w5, w4);
-#define PIXEL10_6   *(q+nextlineDst) = interpolate16_2<bitFormat,3,1>(w4, w5);
+#define PIXEL10_1   *(q+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w4);
+#define PIXEL10_3   *(q+nextlineDst) = interpolate16_7_1<ColorMask>(w5, w4);
+#define PIXEL10_6   *(q+nextlineDst) = interpolate16_3_1<ColorMask>(w4, w5);
 #define PIXEL10_C   *(q+nextlineDst) = w5;
 
 #define PIXEL11     *(q+1+nextlineDst) = w5;
 
-#define PIXEL12_1   *(q+2+nextlineDst) = interpolate16_2<bitFormat,3,1>(w5, w6);
-#define PIXEL12_3   *(q+2+nextlineDst) = interpolate16_2<bitFormat,7,1>(w5, w6);
-#define PIXEL12_6   *(q+2+nextlineDst) = interpolate16_2<bitFormat,3,1>(w6, w5);
+#define PIXEL12_1   *(q+2+nextlineDst) = interpolate16_3_1<ColorMask>(w5, w6);
+#define PIXEL12_3   *(q+2+nextlineDst) = interpolate16_7_1<ColorMask>(w5, w6);
+#define PIXEL12_6   *(q+2+nextlineDst) = interpolate16_3_1<ColorMask>(w6, w5);
 #define PIXEL12_C   *(q+2+nextlineDst) = w5;
 
-#define PIXEL20_1M  *(q+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w7);
-#define PIXEL20_1D  *(q+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w8);
-#define PIXEL20_1L  *(q+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w4);
-#define PIXEL20_2   *(q+nextlineDst2) = interpolate16_3<bitFormat,2,1,1>(w5, w8, w4);
-#define PIXEL20_4   *(q+nextlineDst2) = interpolate16_3<bitFormat,2,7,7>(w5, w8, w4);
-#define PIXEL20_5   *(q+nextlineDst2) = interpolate16_2<bitFormat,1,1>(w8, w4);
+#define PIXEL20_1M  *(q+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w7);
+#define PIXEL20_1D  *(q+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w8);
+#define PIXEL20_1L  *(q+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w4);
+#define PIXEL20_2   *(q+nextlineDst2) = interpolate16_2_1_1<ColorMask>(w5, w8, w4);
+#define PIXEL20_4   *(q+nextlineDst2) = interpolate16_2_7_7<ColorMask>(w5, w8, w4);
+#define PIXEL20_5   *(q+nextlineDst2) = interpolate16_1_1<ColorMask>(w8, w4);
 #define PIXEL20_C   *(q+nextlineDst2) = w5;
 
-#define PIXEL21_1   *(q+1+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w8);
-#define PIXEL21_3   *(q+1+nextlineDst2) = interpolate16_2<bitFormat,7,1>(w5, w8);
-#define PIXEL21_6   *(q+1+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w8, w5);
+#define PIXEL21_1   *(q+1+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w8);
+#define PIXEL21_3   *(q+1+nextlineDst2) = interpolate16_7_1<ColorMask>(w5, w8);
+#define PIXEL21_6   *(q+1+nextlineDst2) = interpolate16_3_1<ColorMask>(w8, w5);
 #define PIXEL21_C   *(q+1+nextlineDst2) = w5;
 
-#define PIXEL22_1M  *(q+2+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w9);
-#define PIXEL22_1D  *(q+2+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w8);
-#define PIXEL22_1R  *(q+2+nextlineDst2) = interpolate16_2<bitFormat,3,1>(w5, w6);
-#define PIXEL22_2   *(q+2+nextlineDst2) = interpolate16_3<bitFormat,2,1,1>(w5, w6, w8);
-#define PIXEL22_4   *(q+2+nextlineDst2) = interpolate16_3<bitFormat,2,7,7>(w5, w6, w8);
-#define PIXEL22_5   *(q+2+nextlineDst2) = interpolate16_2<bitFormat,1,1>(w6, w8);
+#define PIXEL22_1M  *(q+2+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w9);
+#define PIXEL22_1D  *(q+2+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w8);
+#define PIXEL22_1R  *(q+2+nextlineDst2) = interpolate16_3_1<ColorMask>(w5, w6);
+#define PIXEL22_2   *(q+2+nextlineDst2) = interpolate16_2_1_1<ColorMask>(w5, w6, w8);
+#define PIXEL22_4   *(q+2+nextlineDst2) = interpolate16_2_7_7<ColorMask>(w5, w6, w8);
+#define PIXEL22_5   *(q+2+nextlineDst2) = interpolate16_1_1<ColorMask>(w6, w8);
 #define PIXEL22_C   *(q+2+nextlineDst2) = w5;
 
 #define YUV(x)	RGBtoYUV[w ## x]
 
 
-#define bitFormat 565
+#define ColorMask Graphics::ColorMasks<565>
 void HQ3x_565(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	#include "graphics/scaler/hq3x.h"
 }
-#undef bitFormat
+#undef ColorMask
 
-#define bitFormat 555
+#define ColorMask Graphics::ColorMasks<555>
 void HQ3x_555(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
 	#include "graphics/scaler/hq3x.h"
 }
-#undef bitFormat
+#undef ColorMask
 
 
 void HQ3x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch, int width, int height) {
@@ -121,5 +121,3 @@ void HQ3x(const uint8 *srcPtr, uint32 srcPitch, uint8 *dstPtr, uint32 dstPitch,
 	else
 		HQ3x_555(srcPtr, srcPitch, dstPtr, dstPitch, width, height);
 }
-
-#endif
diff --git a/graphics/scaler/intern.h b/graphics/scaler/intern.h
index c9fd550d35..7546d39021 100644
--- a/graphics/scaler/intern.h
+++ b/graphics/scaler/intern.h
@@ -31,7 +31,7 @@
 
 
 #define kHighBitsMask	Graphics::ColorMasks<bitFormat>::kHighBitsMask
-#define kLowBitsMask		Graphics::ColorMasks<bitFormat>::kLowBitsMask
+#define kLowBitsMask	Graphics::ColorMasks<bitFormat>::kLowBitsMask
 #define qhighBits	Graphics::ColorMasks<bitFormat>::qhighBits
 #define qlowBits	Graphics::ColorMasks<bitFormat>::qlowBits
 #define redblueMask	Graphics::ColorMasks<bitFormat>::kRedBlueMask
@@ -78,7 +78,7 @@ static inline uint32 interpolate32_1_1_1_1(uint32 p1, uint32 p2, uint32 p3, uint
 
 /**
  * Interpolate two 16 bit pixels with the weights specified in the template
- * parameters. Used by the hq scaler family.
+ * parameters.
  * @note w1 and w2 must sum up to 2, 4, 8 or 16.
  */
 template<int bitFormat, int w1, int w2>
@@ -89,7 +89,7 @@ static inline uint16 interpolate16_2(uint16 p1, uint16 p2) {
 
 /**
  * Interpolate three 16 bit pixels with the weights specified in the template
- * parameters. Used by the hq scaler family.
+ * parameters.
  * @note w1, w2 and w3 must sum up to 2, 4, 8 or 16.
  */
 template<int bitFormat, int w1, int w2, int w3>
@@ -100,6 +100,110 @@ static inline uint16 interpolate16_3(uint16 p1, uint16 p2, uint16 p3) {
 
 
 /**
+ * Interpolate two 16 bit pixels with weights 1 and 1, i.e., (p1+p2)/2.
+ * See <http://www.slack.net/~ant/info/rgb_mixing.html> for details on how this works.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_1_1(unsigned p1, unsigned p2) {
+	const unsigned lowbits = (p1 ^ p2) & ColorMask::kLowBits;
+	return ((p1 + p2) - lowbits) >> 1;
+}
+
+/**
+ * Interpolate two 16 bit pixels with weights 3 and 1, i.e., (3*p1+p2)/4.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_3_1(unsigned p1, unsigned p2) {
+	const unsigned lowbits = (((p1 & ColorMask::kLowBits) << 1) + (p1 & ColorMask::kLow2Bits)
+		                   + (p2 & ColorMask::kLow2Bits)) & ColorMask::kLow2Bits;
+	return ((p1*3 + p2) - lowbits) >> 2;
+}
+
+/**
+ * Interpolate two 16 bit pixels with weights 7 and 1, i.e., (7*p1+p2)/8.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_7_1(unsigned p1, unsigned p2) {
+	const unsigned lowbits = (((p1 & ColorMask::kLowBits) << 2) + ((p1 & ColorMask::kLow2Bits) << 1) + (p1 & ColorMask::kLow3Bits)
+		                   +  (p2 & ColorMask::kLow3Bits)) & ColorMask::kLow3Bits;
+	return ((p1*7+p2) - lowbits) >> 3;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 2, 1, and 1, i.e., (2*p1+p2+p3)/4.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_2_1_1(unsigned p1, unsigned p2, unsigned p3) {
+	p1<<=1;
+	const unsigned lowbits = ((p1 & (ColorMask::kLowBits << 1))
+		                   +  (p2 & ColorMask::kLow2Bits)
+		                   +  (p3 & ColorMask::kLow2Bits)) & ColorMask::kLow2Bits;
+	return ((p1+p2+p3) - lowbits) >> 2;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 5, 2, and 1, i.e., (5*p1+2*p2+p3)/8.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_5_2_1(unsigned p1, unsigned p2, unsigned p3) {
+	p2<<=1;
+	const unsigned lowbits = (((p1 & ColorMask::kLowBits) << 2) + (p1 & ColorMask::kLow3Bits)
+		                   +  (p2 & (ColorMask::kLow2Bits << 1))
+		                   +  (p3 & ColorMask::kLow3Bits)) & ColorMask::kLow3Bits;
+	return ((p1*5 + p2 + p3) - lowbits) >> 3;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 6, 1, and 1, i.e., (6*p1+p2+p3)/8.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_6_1_1(unsigned p1, unsigned p2, unsigned p3) {
+	const unsigned lowbits = (((((p1 & ColorMask::kLowBits) << 1) + (p1 & ColorMask::kLow2Bits)) << 1)
+		                   + (p2 & ColorMask::kLow3Bits)
+		                   + (p3 & ColorMask::kLow3Bits)) & ColorMask::kLow3Bits;
+	return ((p1*6 + p2 + p3) - lowbits) >> 3;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 2, 3, and 3, i.e., (2*p1+3*(p2+p3))/8.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_2_3_3(unsigned p1, unsigned p2, unsigned p3) {
+	p1 <<= 1;
+	const unsigned rb = (p1 & (ColorMask::kRedBlueMask<<1))
+		              + ((p2 & ColorMask::kRedBlueMask) + (p3 & ColorMask::kRedBlueMask))*3;
+	const unsigned  g = (p1 & (ColorMask::kGreenMask<<1))
+		              + ((p2 & ColorMask::kGreenMask) + (p3 & ColorMask::kGreenMask))*3;
+	return ((rb & (ColorMask::kRedBlueMask<<3)) | (g & (ColorMask::kGreenMask<<3))) >> 3;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 2, 7, and 7, i.e., (2*p1+7*(p2+p3))/16.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_2_7_7(unsigned p1, unsigned p2, unsigned p3) {
+	p1 <<= 1;
+	const unsigned rb = (p1 & (ColorMask::kRedBlueMask<<1))
+		              + ((p2 & ColorMask::kRedBlueMask) + (p3 & ColorMask::kRedBlueMask))*7;
+	const unsigned  g = (p1 & (ColorMask::kGreenMask<<1))
+		              + ((p2 & ColorMask::kGreenMask) + (p3 & ColorMask::kGreenMask))*7;
+	return ((rb & (ColorMask::kRedBlueMask<<4)) | (g & (ColorMask::kGreenMask<<4))) >> 4;
+}
+
+/**
+ * Interpolate three 16 bit pixels with weights 14, 1, and 1, i.e., (14*p1+p2+p3)/16.
+ */
+template<typename ColorMask>
+static inline unsigned interpolate16_14_1_1(unsigned p1, unsigned p2, unsigned p3) {
+	const unsigned rb = (p1&ColorMask::kRedBlueMask)*14
+	                  + (p2&ColorMask::kRedBlueMask)
+	                  + (p3&ColorMask::kRedBlueMask);
+	const unsigned  g = (p1&ColorMask::kGreenMask)*14
+	                  + (p2&ColorMask::kGreenMask) + (p3&ColorMask::kGreenMask);
+	return ((rb&(ColorMask::kRedBlueMask<<4)) | (g&(ColorMask::kGreenMask<<4))) >> 4;
+}
+
+/**
  * Compare two YUV values (encoded 8-8-8) and check if they differ by more than
  * a certain hard coded threshold. Used by the hq scaler family.
  */