diff '--color=auto' -ruN a/source/row_common.cc b/source/row_common.cc
--- a/source/row_common.cc	2025-07-15 17:55:24.611751521 +0200
+++ b/source/row_common.cc	2025-07-15 18:01:57.808312551 +0200
@@ -104,8 +104,13 @@
 #if defined(__x86_64__) || defined(_M_X64) || defined(__i386__) || \
     defined(_M_IX86) || defined(__arm__) || defined(_M_ARM) ||     \
     (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
+#define WRITE16(p, v) *(uint16_t*)(p) = v
 #define WRITEWORD(p, v) *(uint32_t*)(p) = v
 #else
+static inline void WRITE16(uint8_t* p, uint16_t v) {
+  p[0] = (uint8_t)(v & 255);
+  p[1] = (uint8_t)((v >> 8) & 255);
+}
 static inline void WRITEWORD(uint8_t* p, uint32_t v) {
   p[0] = (uint8_t)(v & 255);
   p[1] = (uint8_t)((v >> 8) & 255);
@@ -408,10 +413,10 @@
     uint8_t b1 = STATIC_CAST(uint8_t, clamp255(src_argb[4] + dither1) >> 3);
     uint8_t g1 = STATIC_CAST(uint8_t, clamp255(src_argb[5] + dither1) >> 2);
     uint8_t r1 = STATIC_CAST(uint8_t, clamp255(src_argb[6] + dither1) >> 3);
-    *(uint16_t*)(dst_rgb + 0) =
-        STATIC_CAST(uint16_t, b0 | (g0 << 5) | (r0 << 11));
-    *(uint16_t*)(dst_rgb + 2) =
-        STATIC_CAST(uint16_t, b1 | (g1 << 5) | (r1 << 11));
+    WRITE16((dst_rgb + 0),
+        STATIC_CAST(uint16_t, b0 | (g0 << 5) | (r0 << 11)));
+    WRITE16((dst_rgb + 2),
+        STATIC_CAST(uint16_t, b1 | (g1 << 5) | (r1 << 11)));
     dst_rgb += 4;
     src_argb += 8;
   }
