use a 8b table for C-version for clz()

30% faster on x86, 5% faster on N5.

New generic function: WebPLog2FloorC()
This function is called as fallback for BitsLog2Floor() when there's
no clz() available.

Change-Id: Ica15c6092112e514c0e200fab89c434de48d4b19

src/utils/utils.c

@@ -305,3 +305,26 @@ int WebPGetColorPalette(const WebPPicture* const pic, uint32_t* const palette) {
 #undef COLOR_HASH_RIGHT_SHIFT
 
 //------------------------------------------------------------------------------
+
+#if defined(WEBP_NEED_LOG_TABLE_8BIT)
+const uint8_t WebPLogTable8bit[256] = {   // 31 ^ clz(i)
+  0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3,
+  4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
+  5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
+  5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
+  6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
+  6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
+  6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
+  6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
+  7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7
+};
+#endif
+
+//------------------------------------------------------------------------------

src/utils/utils.h

@@ -107,6 +107,19 @@ static WEBP_INLINE void PutLE32(uint8_t* const data, uint32_t val) {
   PutLE16(data + 2, (int)(val >> 16));
 }
 
+// Returns 31 ^ clz(n) = log2(n). This is the default C-implementation, either
+// based on table or not. Can be used as fallback if clz() is not available.
+#define WEBP_NEED_LOG_TABLE_8BIT
+extern const uint8_t WebPLogTable8bit[256];
+static WEBP_INLINE int WebPLog2FloorC(uint32_t n) {
+  int log = 0;
+  while (n >= 256) {
+    log += 8;
+    n >>= 8;
+  }
+  return log + WebPLogTable8bit[n];
+}
+
 // Returns (int)floor(log2(n)). n must be > 0.
 // use GNU builtins where available.
 #if defined(__GNUC__) && \
@@ -124,22 +137,8 @@ static WEBP_INLINE int BitsLog2Floor(uint32_t n) {
   _BitScanReverse(&first_set_bit, n);
   return first_set_bit;
 }
-#else
+#else   // default: use the C-version.
-static WEBP_INLINE int BitsLog2Floor(uint32_t n) {
+static WEBP_INLINE int BitsLog2Floor(uint32_t n) { return WebPLog2FloorC(n); }
-  int log = 0;
-  uint32_t value = n;
-  int i;
-
-  for (i = 4; i >= 0; --i) {
-    const int shift = (1 << i);
-    const uint32_t x = value >> shift;
-    if (x != 0) {
-      value = x;
-      log += shift;
-    }
-  }
-  return log;
-}
 #endif
 
 //------------------------------------------------------------------------------