diff --git a/src/strategies/avx2/picture-avx2.c b/src/strategies/avx2/picture-avx2.c
index d3b6ca29..2c47f02d 100644
--- a/src/strategies/avx2/picture-avx2.c
+++ b/src/strategies/avx2/picture-avx2.c
@@ -916,28 +916,28 @@ static INLINE void bipred_average_px_px_avx2(kvz_pixel *dst,
 {
   // Use scalar code for yet unoptimized block sizes (4x4, 2x8)
   if (!(pu_w == 4 && pu_h == 4) && pu_w > 2) {
-    switch (pu_w) {
-    case  4: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  4, pu_h, dst_stride); break;
-    case  8: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  8, pu_h, dst_stride); break;
-    case 16: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 16, pu_h, dst_stride); break;
-    case 32: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 32, pu_h, dst_stride); break;
-    case 64: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 64, pu_h, dst_stride); break;
-    
-    case  6: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  6, pu_h, dst_stride); break;
-    case 12: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 12, pu_h, dst_stride); break;
-    case 24: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 24, pu_h, dst_stride); break;
-    case 48: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 48, pu_h, dst_stride); break;
+    switch (pu_w) {
+    case  4: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  4, pu_h, dst_stride); break;
+    case  8: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  8, pu_h, dst_stride); break;
+    case 16: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 16, pu_h, dst_stride); break;
+    case 32: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 32, pu_h, dst_stride); break;
+    case 64: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 64, pu_h, dst_stride); break;
+    
+    case  6: bipred_average_px_px_template_avx2(dst, px_L0, px_L1,  6, pu_h, dst_stride); break;
+    case 12: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 12, pu_h, dst_stride); break;
+    case 24: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 24, pu_h, dst_stride); break;
+    case 48: bipred_average_px_px_template_avx2(dst, px_L0, px_L1, 48, pu_h, dst_stride); break;
     default:
       assert(0 && "Unexpected block width.");
-      break;
+      break;
     }
   } else {
     int32_t shift = 15 - KVZ_BIT_DEPTH; // TODO: defines
     int32_t offset = 1 << (shift - 1);
 
-    for (int i = 0; i < pu_w * pu_h; ++i)
-    {
-      int y = i / pu_w;
+    for (int i = 0; i < pu_w * pu_h; ++i)
+    {
+      int y = i / pu_w;
       int x = i % pu_w;
       int16_t sample_L0 = px_L0[i] << (14 - KVZ_BIT_DEPTH);
       int16_t sample_L1 = px_L1[i] << (14 - KVZ_BIT_DEPTH);
@@ -1144,29 +1144,29 @@ static void bipred_average_ip_ip_avx2(kvz_pixel *dst,
 {
   // Use scalar code for yet unoptimized block sizes (4x4, 2x8)
   if (!(pu_w == 4 && pu_h == 4) && pu_w > 2) {
-    switch (pu_w) {
-    case  4: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 4, pu_h, dst_stride); break;
-    case  8: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 8, pu_h, dst_stride); break;
-    case 16: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 16, pu_h, dst_stride); break;
-    case 32: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 32, pu_h, dst_stride); break;
-    case 64: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 64, pu_h, dst_stride); break;
-
-    case  6: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1,  6, pu_h, dst_stride); break;
-    case 12: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 12, pu_h, dst_stride); break;
-    case 24: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 24, pu_h, dst_stride); break;
-    case 48: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 48, pu_h, dst_stride); break;
+    switch (pu_w) {
+    case  4: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 4, pu_h, dst_stride); break;
+    case  8: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 8, pu_h, dst_stride); break;
+    case 16: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 16, pu_h, dst_stride); break;
+    case 32: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 32, pu_h, dst_stride); break;
+    case 64: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 64, pu_h, dst_stride); break;
+
+    case  6: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1,  6, pu_h, dst_stride); break;
+    case 12: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 12, pu_h, dst_stride); break;
+    case 24: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 24, pu_h, dst_stride); break;
+    case 48: bipred_average_ip_ip_template_avx2(dst, ip_L0, ip_L1, 48, pu_h, dst_stride); break;
     default:
       printf("%dx%d\n", pu_w, pu_h);
       assert(0 && "Unexpected block width.");
-      break;
+      break;
     }
   } else {
     int32_t shift = 15 - KVZ_BIT_DEPTH; // TODO: defines
     int32_t offset = 1 << (shift - 1);
 
-    for (int i = 0; i < pu_w * pu_h; ++i)
-    {
-      int y = i / pu_w;
+    for (int i = 0; i < pu_w * pu_h; ++i)
+    {
+      int y = i / pu_w;
       int x = i % pu_w;
       int16_t sample_L0 = ip_L0[i];
       int16_t sample_L1 = ip_L1[i];
@@ -1385,28 +1385,28 @@ static void bipred_average_px_ip_avx2(kvz_pixel *dst,
 {
   // Use scalar code for yet unoptimized block sizes (4x4, 2x8)
   if (!(pu_w == 4 && pu_h == 4) && pu_w > 2) {
-    switch (pu_w) {
-    case  4: bipred_average_px_ip_template_avx2(dst, px, ip, 4, pu_h, dst_stride); break;
-    case  8: bipred_average_px_ip_template_avx2(dst, px, ip, 8, pu_h, dst_stride); break;
-    case 16: bipred_average_px_ip_template_avx2(dst, px, ip, 16, pu_h, dst_stride); break;
-    case 32: bipred_average_px_ip_template_avx2(dst, px, ip, 32, pu_h, dst_stride); break;
-    case 64: bipred_average_px_ip_template_avx2(dst, px, ip, 64, pu_h, dst_stride); break;
-
-    case  6: bipred_average_px_ip_template_avx2(dst, px, ip,  6, pu_h, dst_stride); break;
-    case 12: bipred_average_px_ip_template_avx2(dst, px, ip, 12, pu_h, dst_stride); break;
-    case 24: bipred_average_px_ip_template_avx2(dst, px, ip, 24, pu_h, dst_stride); break;
-    case 48: bipred_average_px_ip_template_avx2(dst, px, ip, 48, pu_h, dst_stride); break;
+    switch (pu_w) {
+    case  4: bipred_average_px_ip_template_avx2(dst, px, ip, 4, pu_h, dst_stride); break;
+    case  8: bipred_average_px_ip_template_avx2(dst, px, ip, 8, pu_h, dst_stride); break;
+    case 16: bipred_average_px_ip_template_avx2(dst, px, ip, 16, pu_h, dst_stride); break;
+    case 32: bipred_average_px_ip_template_avx2(dst, px, ip, 32, pu_h, dst_stride); break;
+    case 64: bipred_average_px_ip_template_avx2(dst, px, ip, 64, pu_h, dst_stride); break;
+
+    case  6: bipred_average_px_ip_template_avx2(dst, px, ip,  6, pu_h, dst_stride); break;
+    case 12: bipred_average_px_ip_template_avx2(dst, px, ip, 12, pu_h, dst_stride); break;
+    case 24: bipred_average_px_ip_template_avx2(dst, px, ip, 24, pu_h, dst_stride); break;
+    case 48: bipred_average_px_ip_template_avx2(dst, px, ip, 48, pu_h, dst_stride); break;
     default:
       assert(0 && "Unexpected block width.");
-      break;
+      break;
     }
   } else {
     int32_t shift = 15 - KVZ_BIT_DEPTH; // TODO: defines
     int32_t offset = 1 << (shift - 1);
 
-    for (int i = 0; i < pu_w * pu_h; ++i)
-    {
-      int y = i / pu_w;
+    for (int i = 0; i < pu_w * pu_h; ++i)
+    {
+      int y = i / pu_w;
       int x = i % pu_w;
       int16_t sample_px = px[i] << (14 - KVZ_BIT_DEPTH);
       int16_t sample_ip = ip[i];