diff --git a/aom_dsp/aom_dsp_rtcd_defs.pl b/aom_dsp/aom_dsp_rtcd_defs.pl
index b39bfaa..f571562 100755
--- a/aom_dsp/aom_dsp_rtcd_defs.pl
+++ b/aom_dsp/aom_dsp_rtcd_defs.pl
@@ -892,10 +892,10 @@
   #
   foreach (@block_sizes) {
     ($w, $h) = @$_;
-    add_proto qw/void/, "aom_sad${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-    add_proto qw/void/, "aom_sad${w}x${h}x4d_avg", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, const uint8_t *second_pred, uint32_t *sad_array";
-    add_proto qw/void/, "aom_sad_skip_${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-    add_proto qw/void/, "aom_masked_sad${w}x${h}x4d", "const uint8_t *src, int src_stride, const uint8_t *ref[], int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned sads[]";
+    add_proto qw/void/, "aom_sad${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[4], int ref_stride, uint32_t sad_array[4]";
+    add_proto qw/void/, "aom_sad${w}x${h}x4d_avg", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[4], int ref_stride, const uint8_t *second_pred, uint32_t sad_array[4]";
+    add_proto qw/void/, "aom_sad_skip_${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[4], int ref_stride, uint32_t sad_array[4]";
+    add_proto qw/void/, "aom_masked_sad${w}x${h}x4d", "const uint8_t *src, int src_stride, const uint8_t *ref[4], int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned sads[4]";
   }
 
   specialize qw/aom_sad128x128x4d avx2          sse2/;
@@ -1084,7 +1084,7 @@
     add_proto qw/void aom_highbd_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
   }
 
-  add_proto qw/void aom_int_pro_row/, "int16_t *hbuf, const uint8_t *ref, const int ref_stride, const int height";
+  add_proto qw/void aom_int_pro_row/, "int16_t hbuf[16], const uint8_t *ref, const int ref_stride, const int height";
   specialize qw/aom_int_pro_row sse2 neon/;
 
   add_proto qw/int16_t aom_int_pro_col/, "const uint8_t *ref, const int width";
diff --git a/aom_dsp/arm/sad4d_neon.c b/aom_dsp/arm/sad4d_neon.c
index 0e633b5..76eb153 100644
--- a/aom_dsp/arm/sad4d_neon.c
+++ b/aom_dsp/arm/sad4d_neon.c
@@ -82,7 +82,7 @@
 
 void aom_sad64x64x4d_neon(const uint8_t *src, int src_stride,
                           const uint8_t *const ref[4], int ref_stride,
-                          uint32_t *res) {
+                          uint32_t res[4]) {
   int i;
   uint16x8_t vec_sum_ref0_lo = vdupq_n_u16(0);
   uint16x8_t vec_sum_ref0_hi = vdupq_n_u16(0);
@@ -128,7 +128,7 @@
 
 void aom_sad32x32x4d_neon(const uint8_t *src, int src_stride,
                           const uint8_t *const ref[4], int ref_stride,
-                          uint32_t *res) {
+                          uint32_t res[4]) {
   int i;
   uint16x8_t vec_sum_ref0_lo = vdupq_n_u16(0);
   uint16x8_t vec_sum_ref0_hi = vdupq_n_u16(0);
@@ -172,7 +172,7 @@
 
 void aom_sad16x16x4d_neon(const uint8_t *src, int src_stride,
                           const uint8_t *const ref[4], int ref_stride,
-                          uint32_t *res) {
+                          uint32_t res[4]) {
   int i;
   uint16x8_t vec_sum_ref0_lo = vdupq_n_u16(0);
   uint16x8_t vec_sum_ref0_hi = vdupq_n_u16(0);
diff --git a/aom_dsp/mips/sad_msa.c b/aom_dsp/mips/sad_msa.c
index 58cdd80..01d4a52 100644
--- a/aom_dsp/mips/sad_msa.c
+++ b/aom_dsp/mips/sad_msa.c
@@ -162,9 +162,9 @@
 }
 
 static void sad_4width_x4d_msa(const uint8_t *src_ptr, int32_t src_stride,
-                               const uint8_t *const aref_ptr[],
+                               const uint8_t *const aref_ptr[4],
                                int32_t ref_stride, int32_t height,
-                               uint32_t *sad_array) {
+                               uint32_t sad_array[4]) {
   const uint8_t *ref0_ptr, *ref1_ptr, *ref2_ptr, *ref3_ptr;
   int32_t ht_cnt;
   uint32_t src0, src1, src2, src3;
@@ -223,9 +223,9 @@
 }
 
 static void sad_8width_x4d_msa(const uint8_t *src_ptr, int32_t src_stride,
-                               const uint8_t *const aref_ptr[],
+                               const uint8_t *const aref_ptr[4],
                                int32_t ref_stride, int32_t height,
-                               uint32_t *sad_array) {
+                               uint32_t sad_array[4]) {
   int32_t ht_cnt;
   const uint8_t *ref0_ptr, *ref1_ptr, *ref2_ptr, *ref3_ptr;
   v16u8 src0, src1, src2, src3;
@@ -274,9 +274,9 @@
 }
 
 static void sad_16width_x4d_msa(const uint8_t *src_ptr, int32_t src_stride,
-                                const uint8_t *const aref_ptr[],
+                                const uint8_t *const aref_ptr[4],
                                 int32_t ref_stride, int32_t height,
-                                uint32_t *sad_array) {
+                                uint32_t sad_array[4]) {
   int32_t ht_cnt;
   const uint8_t *ref0_ptr, *ref1_ptr, *ref2_ptr, *ref3_ptr;
   v16u8 src, ref0, ref1, ref2, ref3, diff;
@@ -339,9 +339,9 @@
 }
 
 static void sad_32width_x4d_msa(const uint8_t *src, int32_t src_stride,
-                                const uint8_t *const aref_ptr[],
+                                const uint8_t *const aref_ptr[4],
                                 int32_t ref_stride, int32_t height,
-                                uint32_t *sad_array) {
+                                uint32_t sad_array[4]) {
   const uint8_t *ref0_ptr, *ref1_ptr, *ref2_ptr, *ref3_ptr;
   int32_t ht_cnt;
   v16u8 src0, src1, ref0, ref1;
@@ -383,9 +383,9 @@
 }
 
 static void sad_64width_x4d_msa(const uint8_t *src, int32_t src_stride,
-                                const uint8_t *const aref_ptr[],
+                                const uint8_t *const aref_ptr[4],
                                 int32_t ref_stride, int32_t height,
-                                uint32_t *sad_array) {
+                                uint32_t sad_array[4]) {
   const uint8_t *ref0_ptr, *ref1_ptr, *ref2_ptr, *ref3_ptr;
   int32_t ht_cnt;
   v16u8 src0, src1, src2, src3;
@@ -659,36 +659,36 @@
 
 #define AOM_SAD_4xHEIGHTx4D_MSA(height)                                   \
   void aom_sad4x##height##x4d_msa(const uint8_t *src, int32_t src_stride, \
-                                  const uint8_t *const refs[],            \
-                                  int32_t ref_stride, uint32_t *sads) {   \
+                                  const uint8_t *const refs[4],           \
+                                  int32_t ref_stride, uint32_t sads[4]) { \
     sad_4width_x4d_msa(src, src_stride, refs, ref_stride, height, sads);  \
   }
 
 #define AOM_SAD_8xHEIGHTx4D_MSA(height)                                   \
   void aom_sad8x##height##x4d_msa(const uint8_t *src, int32_t src_stride, \
-                                  const uint8_t *const refs[],            \
-                                  int32_t ref_stride, uint32_t *sads) {   \
+                                  const uint8_t *const refs[4],           \
+                                  int32_t ref_stride, uint32_t sads[4]) { \
     sad_8width_x4d_msa(src, src_stride, refs, ref_stride, height, sads);  \
   }
 
 #define AOM_SAD_16xHEIGHTx4D_MSA(height)                                   \
   void aom_sad16x##height##x4d_msa(const uint8_t *src, int32_t src_stride, \
-                                   const uint8_t *const refs[],            \
-                                   int32_t ref_stride, uint32_t *sads) {   \
+                                   const uint8_t *const refs[4],           \
+                                   int32_t ref_stride, uint32_t sads[4]) { \
     sad_16width_x4d_msa(src, src_stride, refs, ref_stride, height, sads);  \
   }
 
 #define AOM_SAD_32xHEIGHTx4D_MSA(height)                                   \
   void aom_sad32x##height##x4d_msa(const uint8_t *src, int32_t src_stride, \
-                                   const uint8_t *const refs[],            \
-                                   int32_t ref_stride, uint32_t *sads) {   \
+                                   const uint8_t *const refs[4],           \
+                                   int32_t ref_stride, uint32_t sads[4]) { \
     sad_32width_x4d_msa(src, src_stride, refs, ref_stride, height, sads);  \
   }
 
 #define AOM_SAD_64xHEIGHTx4D_MSA(height)                                   \
   void aom_sad64x##height##x4d_msa(const uint8_t *src, int32_t src_stride, \
-                                   const uint8_t *const refs[],            \
-                                   int32_t ref_stride, uint32_t *sads) {   \
+                                   const uint8_t *const refs[4],           \
+                                   int32_t ref_stride, uint32_t sads[4]) { \
     sad_64width_x4d_msa(src, src_stride, refs, ref_stride, height, sads);  \
   }
 
diff --git a/aom_dsp/sad.c b/aom_dsp/sad.c
index d72b6c6..771d34d 100644
--- a/aom_dsp/sad.c
+++ b/aom_dsp/sad.c
@@ -69,33 +69,33 @@
   }
 
 // Calculate sad against 4 reference locations and store each in sad_array
-#define sadMxNx4D(m, n)                                                      \
-  void aom_sad##m##x##n##x4d_c(const uint8_t *src, int src_stride,           \
-                               const uint8_t *const ref_array[],             \
-                               int ref_stride, uint32_t *sad_array) {        \
-    int i;                                                                   \
-    for (i = 0; i < 4; ++i) {                                                \
-      sad_array[i] =                                                         \
-          aom_sad##m##x##n##_c(src, src_stride, ref_array[i], ref_stride);   \
-    }                                                                        \
-  }                                                                          \
-  void aom_sad##m##x##n##x4d_avg_c(                                          \
-      const uint8_t *src, int src_stride, const uint8_t *const ref_array[],  \
-      int ref_stride, const uint8_t *second_pred, uint32_t *sad_array) {     \
-    int i;                                                                   \
-    for (i = 0; i < 4; ++i) {                                                \
-      sad_array[i] = aom_sad##m##x##n##_avg_c(src, src_stride, ref_array[i], \
-                                              ref_stride, second_pred);      \
-    }                                                                        \
-  }                                                                          \
-  void aom_sad_skip_##m##x##n##x4d_c(const uint8_t *src, int src_stride,     \
-                                     const uint8_t *const ref_array[],       \
-                                     int ref_stride, uint32_t *sad_array) {  \
-    int i;                                                                   \
-    for (i = 0; i < 4; ++i) {                                                \
-      sad_array[i] = 2 * sad(src, 2 * src_stride, ref_array[i],              \
-                             2 * ref_stride, (m), (n / 2));                  \
-    }                                                                        \
+#define sadMxNx4D(m, n)                                                       \
+  void aom_sad##m##x##n##x4d_c(const uint8_t *src, int src_stride,            \
+                               const uint8_t *const ref_array[4],             \
+                               int ref_stride, uint32_t sad_array[4]) {       \
+    int i;                                                                    \
+    for (i = 0; i < 4; ++i) {                                                 \
+      sad_array[i] =                                                          \
+          aom_sad##m##x##n##_c(src, src_stride, ref_array[i], ref_stride);    \
+    }                                                                         \
+  }                                                                           \
+  void aom_sad##m##x##n##x4d_avg_c(                                           \
+      const uint8_t *src, int src_stride, const uint8_t *const ref_array[4],  \
+      int ref_stride, const uint8_t *second_pred, uint32_t sad_array[4]) {    \
+    int i;                                                                    \
+    for (i = 0; i < 4; ++i) {                                                 \
+      sad_array[i] = aom_sad##m##x##n##_avg_c(src, src_stride, ref_array[i],  \
+                                              ref_stride, second_pred);       \
+    }                                                                         \
+  }                                                                           \
+  void aom_sad_skip_##m##x##n##x4d_c(const uint8_t *src, int src_stride,      \
+                                     const uint8_t *const ref_array[4],       \
+                                     int ref_stride, uint32_t sad_array[4]) { \
+    int i;                                                                    \
+    for (i = 0; i < 4; ++i) {                                                 \
+      sad_array[i] = 2 * sad(src, 2 * src_stride, ref_array[i],               \
+                             2 * ref_stride, (m), (n / 2));                   \
+    }                                                                         \
   }
 
 // 128x128
diff --git a/aom_dsp/sad_av1.c b/aom_dsp/sad_av1.c
index 3c90bf8..f3d5847 100644
--- a/aom_dsp/sad_av1.c
+++ b/aom_dsp/sad_av1.c
@@ -51,9 +51,9 @@
                         msk_stride, m, n);                                     \
   }                                                                            \
   void aom_masked_sad##m##x##n##x4d_c(                                         \
-      const uint8_t *src, int src_stride, const uint8_t *ref[],                \
+      const uint8_t *src, int src_stride, const uint8_t *ref[4],               \
       int ref_stride, const uint8_t *second_pred, const uint8_t *msk,          \
-      int msk_stride, int invert_mask, unsigned sads[]) {                      \
+      int msk_stride, int invert_mask, unsigned sads[4]) {                     \
     if (!invert_mask)                                                          \
       for (int i = 0; i < 4; i++) {                                            \
         sads[i] = masked_sad(src, src_stride, ref[i], ref_stride, second_pred, \
diff --git a/aom_dsp/x86/avg_intrin_sse2.c b/aom_dsp/x86/avg_intrin_sse2.c
index a52abd0..8d27027 100644
--- a/aom_dsp/x86/avg_intrin_sse2.c
+++ b/aom_dsp/x86/avg_intrin_sse2.c
@@ -516,7 +516,7 @@
   return _mm_cvtsi128_si32(accum);
 }
 
-void aom_int_pro_row_sse2(int16_t *hbuf, const uint8_t *ref,
+void aom_int_pro_row_sse2(int16_t hbuf[16], const uint8_t *ref,
                           const int ref_stride, const int height) {
   int idx = 1;
   __m128i zero = _mm_setzero_si128();
diff --git a/aom_dsp/x86/masked_sad4d_ssse3.c b/aom_dsp/x86/masked_sad4d_ssse3.c
index 8ef7ee0..1235f27 100644
--- a/aom_dsp/x86/masked_sad4d_ssse3.c
+++ b/aom_dsp/x86/masked_sad4d_ssse3.c
@@ -37,11 +37,11 @@
   res##idx = _mm_add_epi32(res##idx, _mm_sad_epu8(pred, src));
 
 static INLINE void masked_sadx4d_ssse3(const uint8_t *src_ptr, int src_stride,
-                                       const uint8_t *a_ptr[], int a_stride,
+                                       const uint8_t *a_ptr[4], int a_stride,
                                        const uint8_t *b_ptr, int b_stride,
                                        const uint8_t *m_ptr, int m_stride,
                                        int width, int height, int inv_mask,
-                                       unsigned sad_array[]) {
+                                       unsigned sad_array[4]) {
   int x, y;
   __m128i a;
   __m128i data_l, data_r, mask_l, mask_r, pred_l, pred_r, pred;
@@ -104,10 +104,10 @@
   res##idx = _mm_add_epi32(res##idx, _mm_sad_epu8(pred, src));
 
 void aom_masked_sad8xhx4d_ssse3(const uint8_t *src_ptr, int src_stride,
-                                const uint8_t *ref_array[], int a_stride,
+                                const uint8_t *ref_array[4], int a_stride,
                                 const uint8_t *b_ptr, int b_stride,
                                 const uint8_t *m_ptr, int m_stride, int height,
-                                int inv_mask, unsigned sad_array[]) {
+                                int inv_mask, unsigned sad_array[4]) {
   const uint8_t *ref0 = ref_array[0];
   const uint8_t *ref1 = ref_array[1];
   const uint8_t *ref2 = ref_array[2];
@@ -165,10 +165,10 @@
   res##idx = _mm_add_epi32(res##idx, _mm_sad_epu8(pred, src));
 
 void aom_masked_sad4xhx4d_ssse3(const uint8_t *src_ptr, int src_stride,
-                                const uint8_t *ref_array[], int a_stride,
+                                const uint8_t *ref_array[4], int a_stride,
                                 const uint8_t *b_ptr, int b_stride,
                                 const uint8_t *m_ptr, int m_stride, int height,
-                                int inv_mask, unsigned sad_array[]) {
+                                int inv_mask, unsigned sad_array[4]) {
   const uint8_t *ref0 = ref_array[0];
   const uint8_t *ref1 = ref_array[1];
   const uint8_t *ref2 = ref_array[2];
@@ -217,27 +217,27 @@
 
 #define MASKSADMXN_SSSE3(m, n)                                                 \
   void aom_masked_sad##m##x##n##x4d_ssse3(                                     \
-      const uint8_t *src, int src_stride, const uint8_t *ref[],                \
+      const uint8_t *src, int src_stride, const uint8_t *ref[4],               \
       int ref_stride, const uint8_t *second_pred, const uint8_t *msk,          \
-      int msk_stride, int inv_mask, unsigned sad_array[]) {                    \
+      int msk_stride, int inv_mask, unsigned sad_array[4]) {                   \
     masked_sadx4d_ssse3(src, src_stride, ref, ref_stride, second_pred, m, msk, \
                         msk_stride, m, n, inv_mask, sad_array);                \
   }
 
 #define MASKSAD8XN_SSSE3(n)                                                   \
   void aom_masked_sad8x##n##x4d_ssse3(                                        \
-      const uint8_t *src, int src_stride, const uint8_t *ref[],               \
+      const uint8_t *src, int src_stride, const uint8_t *ref[4],              \
       int ref_stride, const uint8_t *second_pred, const uint8_t *msk,         \
-      int msk_stride, int inv_mask, unsigned sad_array[]) {                   \
+      int msk_stride, int inv_mask, unsigned sad_array[4]) {                  \
     aom_masked_sad8xhx4d_ssse3(src, src_stride, ref, ref_stride, second_pred, \
                                8, msk, msk_stride, n, inv_mask, sad_array);   \
   }
 
 #define MASKSAD4XN_SSSE3(n)                                                   \
   void aom_masked_sad4x##n##x4d_ssse3(                                        \
-      const uint8_t *src, int src_stride, const uint8_t *ref[],               \
+      const uint8_t *src, int src_stride, const uint8_t *ref[4],              \
       int ref_stride, const uint8_t *second_pred, const uint8_t *msk,         \
-      int msk_stride, int inv_mask, unsigned sad_array[]) {                   \
+      int msk_stride, int inv_mask, unsigned sad_array[4]) {                  \
     aom_masked_sad4xhx4d_ssse3(src, src_stride, ref, ref_stride, second_pred, \
                                4, msk, msk_stride, n, inv_mask, sad_array);   \
   }