diff --git a/ewah/ewok.h b/ewah/ewok.h
index f6ad190..e54d822 100644
--- a/ewah/ewok.h
+++ b/ewah/ewok.h
@@ -34,17 +34,14 @@ typedef uint64_t eword_t;
 #define BITS_IN_WORD (sizeof(eword_t) * 8)
 
 /**
- * Do not use __builtin_popcountll. The GCC implementation
- * is notoriously slow on all platforms.
+ * Use -march=native and latest GCC to get fast popcount.
  *
  * See: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36041
+ *      https://gcc.gnu.org/viewcvs/gcc/trunk/libgcc/libgcc2.c?r1=200506&r2=200505&pathrev=200506
  */
 static inline uint32_t ewah_bit_popcount64(uint64_t x)
 {
-	x = (x & 0x5555555555555555ULL) + ((x >>  1) & 0x5555555555555555ULL);
-	x = (x & 0x3333333333333333ULL) + ((x >>  2) & 0x3333333333333333ULL);
-	x = (x & 0x0F0F0F0F0F0F0F0FULL) + ((x >>  4) & 0x0F0F0F0F0F0F0F0FULL);
-	return (x * 0x0101010101010101ULL) >> 56;
+	return __builtin_popcountll(x);
 }
 
 #ifdef __GNUC__