diff --git a/src/strategies/avx2/avx2_common_functions.h b/src/strategies/avx2/avx2_common_functions.h
index 3b6063ad..701f3ee9 100644
--- a/src/strategies/avx2/avx2_common_functions.h
+++ b/src/strategies/avx2/avx2_common_functions.h
@@ -15,6 +15,16 @@
   #define FIX_W32
 #endif
 
+// Non-inline functions defined in this header are likely to trigger a
+// warning for each module including this header that does NOT use them,
+// at least on unix-ish platforms (GCC/Clang both on native Unix and MinGW).
+// Tell 'em we actually want to do that, it's not an accident.
+#if defined __GNUC__ || defined __clang__ || defined __MINGW32__ || defined __MINGW64__
+  #define FIX_UNUSED __attribute__((unused))
+#endif
+
+#define FIX_NOINLINE FIX_W32 FIX_UNUSED
+
 /*
  * Reorder coefficients from raster to scan order
  * Fun fact: Once upon a time, doing this in a loop looked like this:
@@ -123,8 +133,7 @@ static INLINE void get_first_last_nz_int16(__m256i ints, int32_t *first, int32_t
   *last = (31 - (int32_t)_lzcnt_u32(nonzero_bytes)) >> 1;
 }
 
-/* MOVED TO SAO-AVX2.C WHERE THIS IS USED
-int32_t FIX_W32 kvz_hsum_8x32b(const __m256i v)
+static int32_t FIX_NOINLINE hsum_8x32b(const __m256i v)
 {
   __m256i sum1 = v;
   __m256i sum2 = _mm256_permute4x64_epi64(sum1, _MM_SHUFFLE(1, 0, 3, 2));
@@ -138,5 +147,5 @@ int32_t FIX_W32 kvz_hsum_8x32b(const __m256i v)
   int32_t sum9 = _mm_cvtsi128_si32       (sum8);
   return  sum9;
 }
-*/
+
 #endif
diff --git a/src/strategies/avx2/sao-avx2.c b/src/strategies/avx2/sao-avx2.c
index 5bd2dd91..e42f911a 100644
--- a/src/strategies/avx2/sao-avx2.c
+++ b/src/strategies/avx2/sao-avx2.c
@@ -271,21 +271,6 @@ static INLINE __m256i FIX_W32 do_one_edge_ymm(const __m256i a,
   return calc_diff_off_delta(diff_lo, diff_hi, offset, orig);
 }
 
-int32_t FIX_W32 kvz_hsum_8x32b(const __m256i v)
-{
-  __m256i sum1 = v;
-  __m256i sum2 = _mm256_permute4x64_epi64(sum1, _MM_SHUFFLE(1, 0, 3, 2));
-  __m256i sum3 = _mm256_add_epi32(sum1, sum2);
-  __m256i sum4 = _mm256_shuffle_epi32(sum3, _MM_SHUFFLE(1, 0, 3, 2));
-  __m256i sum5 = _mm256_add_epi32(sum3, sum4);
-  __m256i sum6 = _mm256_shuffle_epi32(sum5, _MM_SHUFFLE(2, 3, 0, 1));
-  __m256i sum7 = _mm256_add_epi32(sum5, sum6);
-
-  __m128i sum8 = _mm256_castsi256_si128(sum7);
-  int32_t sum9 = _mm_cvtsi128_si32(sum8);
-  return  sum9;
-}
-
 static int32_t sao_edge_ddistortion_avx2(const kvz_pixel *orig_data,
                                          const kvz_pixel *rec_data,
                                                int32_t    block_width,
@@ -384,7 +369,7 @@ static int32_t sao_edge_ddistortion_avx2(const kvz_pixel *orig_data,
               sum   = _mm256_add_epi32(sum, curr);
     }
   }
-  return kvz_hsum_8x32b(sum);
+  return hsum_8x32b(sum);
 }
 
 static void FIX_W32 calc_edge_dir_one_ymm(const __m256i  a,
@@ -499,7 +484,7 @@ static void calc_sao_edge_dir_avx2(const kvz_pixel *orig_data,
     }
   }
   for (uint32_t i = 0; i < 5; i++) {
-    int32_t sum = kvz_hsum_8x32b(diff_accum[i]);
+    int32_t sum = hsum_8x32b(diff_accum[i]);
     diff_sum[i] += sum;
   }
 }
@@ -892,7 +877,7 @@ static int32_t sao_band_ddistortion_avx2(const encoder_state_t *state,
               sum          = _mm256_add_epi32    (sum,          curr_sum);
     }
   }
-  return kvz_hsum_8x32b(sum);
+  return hsum_8x32b(sum);
 
 use_generic:
   return sao_band_ddistortion_generic(state, orig_data, rec_data, block_width,