diff --git a/adapters/opencv/kleidicv_hal.cpp b/adapters/opencv/kleidicv_hal.cpp
index b62f736bcb7ae0ab92679ea82cd329aa8741de8b..2fa3c244db08ee6c5bf99dc53ca5e431c6e95130 100644
--- a/adapters/opencv/kleidicv_hal.cpp
+++ b/adapters/opencv/kleidicv_hal.cpp
@@ -1023,6 +1023,12 @@ int sum(const uchar *src_data, size_t src_step, int src_type, size_t width,
 int rotate(int src_type, const uchar *src_data, size_t src_step, int src_width,
            int src_height, uchar *dst_data, size_t dst_step, int angle) {
   int element_size = CV_ELEM_SIZE(src_type);
+#if !KLEIDICV_ENABLE_ALL_OPENCV_HAL
+  // KleidiCV has regression on some devices for 4-byte and 8-byte element size
+  if ((element_size != 1) && (element_size != 2)) {
+    return CV_HAL_ERROR_NOT_IMPLEMENTED;
+  }
+#endif  // KLEIDICV_ENABLE_ALL_OPENCV_HAL
 
   int multithread_min_elements = 0;
   switch (element_size) {
diff --git a/test/api/test_rotate_thread.cpp b/test/api/test_rotate_thread.cpp
index 4268bb8576c2baeca6e8f205a9a1352ea317072f..9f3d38edef52a7f303ab43479af0a8b616350958 100644
--- a/test/api/test_rotate_thread.cpp
+++ b/test/api/test_rotate_thread.cpp
@@ -75,9 +75,9 @@ class RotateThread : public testing::TestWithParam<size_t> {
                           size_t height, size_t stride,
                           size_t element_size) const {
     for (size_t i = 0; i < height; i++) {
-      for (size_t j = 0; j < width * element_size; j++) {
-        ASSERT_EQ(lhs[i * stride + j], rhs[i * stride + j]);
-      }
+      ASSERT_EQ(
+          std::memcmp(lhs + i * stride, rhs + i * stride, width * element_size),
+          0);
     }
   }
 };