Speed up rounding to byte test

unit_test/color_test.cc

 /*
  *  Copyright 2015 The LibYuv Project Authors. All rights reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS. All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
@@ skipping 211 matching lines @@
   *r = orig_pixels[2];
 }
 
 // Pick a method for clamping.
 #define CLAMPMETHOD_IF 1
 //  #define CLAMPMETHOD_TABLE 1
 //  #define CLAMPMETHOD_TERNARY 1
 //  #define CLAMPMETHOD_MASK 1
 
 // Pick a method for rounding.
-#define ROUND(f) static_cast<int>(f + 0.5)
-//  #define ROUND(f) lrintf(f)
-//  #define ROUND(f) static_cast<int>(round(f))
-//  #define ROUND(f) _mm_cvt_ss2si(_mm_load_ss(&f))
+#define ROUND(f) lrintf(f)

[brucedawson, 2015/10/02 18:48:07]

But, lrintf is ~376 times slower on VC++
(https://connect.microsoft.com/VisualStudio/feedback/details/1179496/cq-intrin...)

[fbarchard, 2015/10/02 21:44:53]

Done.

+//#define ROUND(f) static_cast<int>(f + 0.5)
+//#define ROUND(f) static_cast<int>(round(f))
+//#define ROUND(f) _mm_cvt_ss2si(_mm_load_ss(&f))
 
 #if defined(CLAMPMETHOD_IF)
 static int RoundToByte(float f) {
   int i =  ROUND(f);
   if (i < 0) {
     i = 0;
   }
   if (i > 255) {
     i = 255;
   }
@@ skipping 61 matching lines @@
   int i = ROUND(f);
   i =  ((-(i) >> 31) & (i));  // clamp to 0.
   return (((255 - (i)) >> 31) | (i)) & 255;  // clamp to 255.
 }
 #endif
 
 #define RANDOM256(s) ((s & 1) ? ((s >> 1) ^ 0xb8) : (s >> 1))
 
 TEST_F(libyuvTest, TestRoundToByte) {
   int allb = 0;
-  for (int i = 0; i < benchmark_iterations_; ++i) {
-    for (int u2 = 0; u2 < 256; ++u2) {
-      for (int v2 = 0; v2 < 256; ++v2) {
-        for (int y2 = 0; y2 < 256; ++y2) {
-          int y = RANDOM256(y2);
-          int b = RoundToByte(y * 810.33 - 257);
-          allb |= b;
-        }
-      }
-    }
+  int count = benchmark_width_ * benchmark_height_ * benchmark_iterations_;
+  for (int i = 0; i < count; ++i) {
+    int y = RANDOM256(i);
+    int b = RoundToByte(y * 810.33 - 257);
+    allb |= b;
   }
   EXPECT_GE(allb, 0);
   EXPECT_LE(allb, 255);
 }
 
 static void YUVToRGBReference(int y, int u, int v, int* r, int* g, int* b) {
   *r = RoundToByte((y - 16) * 1.164 - (v - 128) * -1.596);
   *g = RoundToByte((y - 16) * 1.164 - (u - 128) * 0.391 - (v - 128) * 0.813);
   *b = RoundToByte((y - 16) * 1.164 - (u - 128) * -2.018);
 }
@@ skipping 214 matching lines @@
     EXPECT_EQ(r0, r1);
     EXPECT_EQ(g0, g1);
     EXPECT_EQ(b0, b1);
     EXPECT_EQ(r0, r2);
     EXPECT_EQ(g0, g2);
     EXPECT_EQ(b0, b2);
   }
 }
 
 }  // namespace libyuv