avcodec/put_bits: Implement put_bits32() in a single pass instead of 2 passes writing 16bits each

820 cpu cycles -> 660 cpu cycles for 100 put_bits32()
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/put_bits.h

@@ -209,15 +209,37 @@ static inline void put_sbits(PutBitContext *pb, int n, int32_t value)
  */
 static void av_unused put_bits32(PutBitContext *s, uint32_t value)
 {
-    int lo = value & 0xffff;
-    int hi = value >> 16;
+    unsigned int bit_buf;
+    int bit_left;
+
+    bit_buf  = s->bit_buf;
+    bit_left = s->bit_left;
+
 #ifdef BITSTREAM_WRITER_LE
-    put_bits(s, 16, lo);
-    put_bits(s, 16, hi);
+    bit_buf |= value << (32 - bit_left);
+    if (3 < s->buf_end - s->buf_ptr) {
+        AV_WL32(s->buf_ptr, bit_buf);
+        s->buf_ptr += 4;
+    } else {
+        av_log(NULL, AV_LOG_ERROR, "Internal error, put_bits buffer too small\n");
+        av_assert2(0);
+    }
+    bit_buf     = (uint64_t)value >> bit_left;
 #else
-    put_bits(s, 16, hi);
-    put_bits(s, 16, lo);
+    bit_buf     = (uint64_t)bit_buf << bit_left;
+    bit_buf    |= value >> (32 - bit_left);
+    if (3 < s->buf_end - s->buf_ptr) {
+        AV_WB32(s->buf_ptr, bit_buf);
+        s->buf_ptr += 4;
+    } else {
+        av_log(NULL, AV_LOG_ERROR, "Internal error, put_bits buffer too small\n");
+        av_assert2(0);
+    }
+    bit_buf     = value;
 #endif
+
+    s->bit_buf  = bit_buf;
+    s->bit_left = bit_left;
 }
 
 /**