[babl] extensions/simple: reuse some fast paths for more component counts

[Øyvind Kolås <ok src gnome org>]

commit 13b7091428060240e7a7701cb3bf783d6b705d2d
Author: Øyvind Kolås <pippin gimp org>
Date:   Fri Dec 9 22:17:15 2016 +0100

    extensions/simple: reuse some fast paths for more component counts

 extensions/simple.c |  200 ++++++++++++++++++++++++++++++++++++++++-----------
 1 files changed, 158 insertions(+), 42 deletions(-)
---
diff --git a/extensions/simple.c b/extensions/simple.c
index 4755921..d03981d 100644
--- a/extensions/simple.c
+++ b/extensions/simple.c
@@ -4,30 +4,43 @@
 
 int init (void);
 
+
 static inline long
-float_to_u8 (unsigned char *src_char, unsigned char *dst, long samples)
+float_to_u8_x1 (unsigned char *src_char, unsigned char *dst, long samples)
 {
   float *src = (float *)src_char;
   long n = samples;
   while (n--)
     {
       float r = src[0];
-      float g = src[1];
-      float b = src[2];
-      float a = src[3];
-            
       dst[0] = (r >= 1.0f) ? 0xFF : ((r <= 0.0f) ? 0x0 : 0xFF * r + 0.5f);
-      dst[1] = (g >= 1.0f) ? 0xFF : ((g <= 0.0f) ? 0x0 : 0xFF * g + 0.5f);
-      dst[2] = (b >= 1.0f) ? 0xFF : ((b <= 0.0f) ? 0x0 : 0xFF * b + 0.5f);
-      dst[3] = (a >= 1.0f) ? 0xFF : ((a <= 0.0f) ? 0x0 : 0xFF * a + 0.5f);
-      
-      dst += 4;
-      src += 4;
+      dst += 1;
+      src += 1;
     }
   return samples;
 }
 
 static inline long
+float_to_u8_x4 (unsigned char *src_char, unsigned char *dst, long samples)
+{
+  return float_to_u8_x1 (src_char, dst, samples * 4);
+}
+
+static inline long
+float_to_u8_x3 (unsigned char *src_char, unsigned char *dst, long samples)
+{
+  return float_to_u8_x1 (src_char, dst, samples * 3);
+}
+
+static inline long
+float_to_u8_x2 (unsigned char *src_char, unsigned char *dst, long samples)
+{
+  return float_to_u8_x1 (src_char, dst, samples * 2);
+}
+
+
+
+static inline long
 float_pre_to_u8_pre (unsigned char *src_char, unsigned char *dst, long samples)
 {
   float *src = (float *)src_char;
@@ -59,7 +72,7 @@ float_pre_to_u8_pre (unsigned char *src_char, unsigned char *dst, long samples)
 }
 
 static inline long
-float_to_u16 (unsigned char *src_char, unsigned char *dst_char, long samples)
+float_to_u16_x1 (unsigned char *src_char, unsigned char *dst_char, long samples)
 {
   float *src    = (float *)src_char;
   uint16_t *dst = (uint16_t *)dst_char;
@@ -67,20 +80,27 @@ float_to_u16 (unsigned char *src_char, unsigned char *dst_char, long samples)
   while (n--)
     {
       float r = src[0];
-      float g = src[1];
-      float b = src[2];
-      float a = src[3];
-            
       dst[0] = (r >= 1.0f) ? 0xFFFF : ((r <= 0.0f) ? 0x0 : 0xFFFF * r + 0.5f);
-      dst[1] = (g >= 1.0f) ? 0xFFFF : ((g <= 0.0f) ? 0x0 : 0xFFFF * g + 0.5f);
-      dst[2] = (b >= 1.0f) ? 0xFFFF : ((b <= 0.0f) ? 0x0 : 0xFFFF * b + 0.5f);
-      dst[3] = (a >= 1.0f) ? 0xFFFF : ((a <= 0.0f) ? 0x0 : 0xFFFF * a + 0.5f);
-      
-      dst += 4;
-      src += 4;
+      dst += 1;
+      src += 1;
     }
   return samples;
 }
+static inline long
+float_to_u16_x2 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u16_x2 (src_char, dst_char, samples * 2);
+}
+static inline long
+float_to_u16_x3 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u16_x3 (src_char, dst_char, samples * 3);
+}
+static inline long
+float_to_u16_x4 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u16_x4 (src_char, dst_char, samples * 4);
+}
 
 static inline long
 float_pre_to_u16_pre (unsigned char *src_char, unsigned char *dst_char, long samples)
@@ -146,7 +166,7 @@ float_pre_to_u32_pre (unsigned char *src_char, unsigned char *dst_char, long sam
 
 
 static inline long
-float_to_u32 (unsigned char *src_char, unsigned char *dst_char, long samples)
+float_to_u32_x1 (unsigned char *src_char, unsigned char *dst_char, long samples)
 {
   float *src    = (float *)src_char;
   uint32_t *dst = (uint32_t *)dst_char;
@@ -154,20 +174,30 @@ float_to_u32 (unsigned char *src_char, unsigned char *dst_char, long samples)
   while (n--)
     {
       float r = src[0];
-      float g = src[1];
-      float b = src[2];
-      float a = src[3];
             
       dst[0] = (r >= 1.0f) ? 0xFFFFFFFF : ((r <= 0.0f) ? 0x0 : 0xFFFFFFFF * r + 0.5f);
-      dst[1] = (g >= 1.0f) ? 0xFFFFFFFF : ((g <= 0.0f) ? 0x0 : 0xFFFFFFFF * g + 0.5f);
-      dst[2] = (b >= 1.0f) ? 0xFFFFFFFF : ((b <= 0.0f) ? 0x0 : 0xFFFFFFFF * b + 0.5f);
-      dst[3] = (a >= 1.0f) ? 0xFFFFFFFF : ((a <= 0.0f) ? 0x0 : 0xFFFFFFFF * a + 0.5f);
       
-      dst += 4;
-      src += 4;
+      dst += 1;
+      src += 1;
     }
   return samples;
 }
+static inline long
+float_to_u32_x2 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u32_x2 (src_char, dst_char, samples * 2);
+}
+static inline long
+float_to_u32_x3 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u32_x3 (src_char, dst_char, samples * 3);
+}
+static inline long
+float_to_u32_x4 (unsigned char *src_char, unsigned char *dst_char, long samples)
+{
+  return float_to_u32_x4 (src_char, dst_char, samples * 4);
+}
+
 
 static inline long
 u32_to_float (unsigned char *src_char, unsigned char *dst_char, long samples)
@@ -205,12 +235,42 @@ init (void)
   babl_conversion_new (babl_format ("R'G'B'A float"),
                        babl_format ("R'G'B'A u8"),
                       "linear", 
-                       float_to_u8,
+                       float_to_u8_x4,
                        NULL);
   babl_conversion_new (babl_format ("RGBA float"),
                        babl_format ("RGBA u8"),
                       "linear", 
-                       float_to_u8,
+                       float_to_u8_x4,
+                       NULL);
+  babl_conversion_new (babl_format ("R'G'B' float"),
+                       babl_format ("R'G'B' u8"),
+                      "linear", 
+                       float_to_u8_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("RGB float"),
+                       babl_format ("RGB u8"),
+                      "linear", 
+                       float_to_u8_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("Y'A float"),
+                       babl_format ("Y'A u8"),
+                      "linear", 
+                       float_to_u8_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("YA float"),
+                       babl_format ("YA u8"),
+                      "linear", 
+                       float_to_u8_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("YA float"),
+                       babl_format ("YA u8"),
+                      "linear", 
+                       float_to_u8_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("Y' float"),
+                       babl_format ("Y' u8"),
+                      "linear", 
+                       float_to_u8_x1,
                        NULL);
   babl_conversion_new (babl_format ("R'aG'aB'aA float"),
                        babl_format ("R'aG'aB'aA u8"),
@@ -227,12 +287,43 @@ init (void)
   babl_conversion_new (babl_format ("R'G'B'A float"),
                        babl_format ("R'G'B'A u16"),
                       "linear", 
-                       float_to_u16,
+                       float_to_u16_x4,
                        NULL);
   babl_conversion_new (babl_format ("RGBA float"),
                        babl_format ("RGBA u16"),
                       "linear", 
-                       float_to_u16,
+                       float_to_u16_x4,
+                       NULL);
+
+  babl_conversion_new (babl_format ("R'G'B' float"),
+                       babl_format ("R'G'B' u16"),
+                      "linear", 
+                       float_to_u16_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("RGB float"),
+                       babl_format ("RGB u16"),
+                      "linear", 
+                       float_to_u16_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("Y'A float"),
+                       babl_format ("Y'A u16"),
+                      "linear", 
+                       float_to_u16_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("YA float"),
+                       babl_format ("YA u16"),
+                      "linear", 
+                       float_to_u16_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("Y' float"),
+                       babl_format ("Y' u16"),
+                      "linear", 
+                       float_to_u16_x1,
+                       NULL);
+  babl_conversion_new (babl_format ("Y float"),
+                       babl_format ("Y u16"),
+                      "linear", 
+                       float_to_u16_x1,
                        NULL);
   babl_conversion_new (babl_format ("R'aG'aB'aA float"),
                        babl_format ("R'aG'aB'aA u16"),
@@ -250,12 +341,42 @@ init (void)
   babl_conversion_new (babl_format ("R'G'B'A float"),
                        babl_format ("R'G'B'A u32"),
                       "linear", 
-                       float_to_u32,
+                       float_to_u32_x4,
                        NULL);
   babl_conversion_new (babl_format ("RGBA float"),
                        babl_format ("RGBA u32"),
                       "linear", 
-                       float_to_u32,
+                       float_to_u32_x4,
+                       NULL);
+  babl_conversion_new (babl_format ("R'G'B' float"),
+                       babl_format ("R'G'B' u32"),
+                      "linear", 
+                       float_to_u32_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("RGB float"),
+                       babl_format ("RGB u32"),
+                      "linear", 
+                       float_to_u32_x3,
+                       NULL);
+  babl_conversion_new (babl_format ("Y'A float"),
+                       babl_format ("Y'A u32"),
+                      "linear", 
+                       float_to_u32_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("YA float"),
+                       babl_format ("YA u32"),
+                      "linear", 
+                       float_to_u32_x2,
+                       NULL);
+  babl_conversion_new (babl_format ("Y' float"),
+                       babl_format ("Y' u32"),
+                      "linear", 
+                       float_to_u32_x1,
+                       NULL);
+  babl_conversion_new (babl_format ("Y float"),
+                       babl_format ("Y u32"),
+                      "linear", 
+                       float_to_u32_x1,
                        NULL);
   babl_conversion_new (babl_format ("R'aG'aB'aA float"),
                        babl_format ("R'aG'aB'aA u32"),
@@ -274,31 +395,26 @@ init (void)
                       "linear", 
                        u32_to_float_x2,
                        NULL);
-
   babl_conversion_new (babl_format ("Y'A u32"),
                        babl_format ("Y'A float"),
                       "linear", 
                        u32_to_float_x2,
                        NULL);
-
   babl_conversion_new (babl_format ("Y u32"),
                        babl_format ("Y float"),
                       "linear", 
                        u32_to_float,
                        NULL);
-
   babl_conversion_new (babl_format ("Y' u32"),
                        babl_format ("Y' float"),
                       "linear", 
                        u32_to_float,
                        NULL);
-
   babl_conversion_new (babl_format ("RGBA u32"),
                        babl_format ("RGBA float"),
                       "linear", 
                        u32_to_float_x4,
                        NULL);
-
   babl_conversion_new (babl_format ("R'G'B'A u32"),
                        babl_format ("R'G'B'A float"),
                       "linear",