arm neon: define native alias only under the inverse of the condition…

…s of a pass-through
simd-everywhere · Jan 11, 2025 · 2b450c0 · 2b450c0
1 parent f4f5904
commit 2b450c0
Show file tree

Hide file tree

Showing 157 changed files with 2,280 additions and 1,134 deletions.
diff --git a/simde/arm/neon/abd.h b/simde/arm/neon/abd.h
@@ -50,7 +50,7 @@ simde_vabdh_f16(simde_float16_t a, simde_float16_t b) {
     return r_ < 0 ? simde_float16_from_float32(-r_) : simde_float16_from_float32(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARM_NEON_FP16))
   #undef vabdh_f16
   #define vabdh_f16(a, b) simde_vabdh_f16((a), (b))
 #endif
@@ -94,7 +94,7 @@ simde_vabd_f16(simde_float16x4_t a, simde_float16x4_t b) {
     return simde_vabs_f16(simde_vsub_f16(a, b));
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARM_NEON_FP16))
   #undef vabd_f16
   #define vabd_f16(a, b) simde_vabd_f16((a), (b))
 #endif
@@ -315,7 +315,7 @@ simde_vabdq_f16(simde_float16x8_t a, simde_float16x8_t b) {
     return simde_vabsq_f16(simde_vsubq_f16(a, b));
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && !defined(SIMDE_ARM_NEON_FP16))
   #undef vabdq_f16
   #define vabdq_f16(a, b) simde_vabdq_f16((a), (b))
 #endif

diff --git a/simde/arm/neon/abs.h b/simde/arm/neon/abs.h
@@ -44,7 +44,8 @@ simde_vabsd_s64(int64_t a) {
     return a < 0 ? -a : a;
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9,1,0)))
   #undef vabsd_s64
   #define vabsd_s64(a) simde_vabsd_s64(a)
 #endif
@@ -60,7 +61,8 @@ simde_vabsh_f16(simde_float16_t a) {
     return (a_ >= 0.0f) ? simde_float16_from_float32(a_) : simde_float16_from_float32(-a_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vabsh_f16
   #define vabsh_f16(a) simde_vabsh_f16(a)
 #endif
@@ -87,7 +89,8 @@ simde_vabs_f16(simde_float16x4_t a) {
     return simde_float16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vabs_f16
   #define vabs_f16(a) simde_vabs_f16(a)
 #endif
@@ -294,7 +297,8 @@ simde_vabsq_f16(simde_float16x8_t a) {
     return simde_float16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vabsq_f16
   #define vabsq_f16(a) simde_vabsq_f16(a)
 #endif

diff --git a/simde/arm/neon/add.h b/simde/arm/neon/add.h
@@ -46,7 +46,8 @@ simde_vaddh_f16(simde_float16_t a, simde_float16_t b) {
     return simde_float16_from_float32(af + bf);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vaddh_f16
   #define vaddh_f16(a, b) simde_vaddh_f16((a), (b))
 #endif
@@ -102,7 +103,8 @@ simde_vadd_f16(simde_float16x4_t a, simde_float16x4_t b) {
     return simde_float16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vadd_f16
   #define vadd_f16(a, b) simde_vadd_f16((a), (b))
 #endif
@@ -437,7 +439,8 @@ simde_vaddq_f16(simde_float16x8_t a, simde_float16x8_t b) {
     return simde_float16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vaddq_f16
   #define vaddq_f16(a, b) simde_vaddq_f16((a), (b))
 #endif
@@ -809,7 +812,8 @@ simde_vadd_p8(simde_poly8x8_t a, simde_poly8x8_t b) {
     return simde_poly8x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    defined(_GCC_ARM_NEON_H))
   #undef vadd_p8
   #define vadd_p8(a, b) simde_vadd_p8((a), (b))
 #endif
@@ -833,7 +837,8 @@ simde_vadd_p16(simde_poly16x4_t a, simde_poly16x4_t b) {
     return simde_poly16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    defined(_GCC_ARM_NEON_H))
   #undef vadd_p16
   #define vadd_p16(a, b) simde_vadd_p16((a), (b))
 #endif
@@ -858,7 +863,9 @@ simde_vadd_p64(simde_poly64x1_t a, simde_poly64x1_t b) {
     return simde_poly64x1_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_CRYPTO) && \
+      !defined(_GCC_ARM_NEON_H)))
   #undef vadd_p64
   #define vadd_p64(a, b) simde_vadd_p64((a), (b))
 #endif
@@ -882,7 +889,8 @@ simde_vaddq_p8(simde_poly8x16_t a, simde_poly8x16_t b) {
     return simde_poly8x16_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    defined(_GCC_ARM_NEON_H))
   #undef vaddq_p8
   #define vaddq_p8(a, b) simde_vaddq_p8((a), (b))
 #endif
@@ -906,7 +914,8 @@ simde_vaddq_p16(simde_poly16x8_t a, simde_poly16x8_t b) {
     return simde_poly16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    defined(_GCC_ARM_NEON_H))
   #undef vaddq_p16
   #define vaddq_p16(a, b) simde_vaddq_p16((a), (b))
 #endif
@@ -931,7 +940,9 @@ simde_vaddq_p64(simde_poly64x2_t a, simde_poly64x2_t b) {
     return simde_poly64x2_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_CRYPTO) && \
+      !defined(_GCC_ARM_NEON_H)))
   #undef vaddq_p64
   #define vaddq_p64(a, b) simde_vaddq_p64((a), (b))
 #endif
@@ -950,7 +961,9 @@ simde_vaddq_p128(simde_poly128_t a, simde_poly128_t b) {
     return b ^ ((0 ^ a) & mask);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_CRYPTO) && \
+      !defined(_GCC_ARM_NEON_H)))
   #undef vaddq_p128
   #define vaddq_p128(a, b) simde_vaddq_p128((a), (b))
 #endif

diff --git a/simde/arm/neon/aes.h b/simde/arm/neon/aes.h
@@ -84,7 +84,8 @@ simde_vaeseq_u8(simde_uint8x16_t data, simde_uint8x16_t key) {
     return simde_uint8x16_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARCH_ARM_AES)))
   #undef vaeseq_u8
   #define vaeseq_u8(data, key) simde_vaeseq_u8((data), (key))
 #endif
@@ -132,7 +133,8 @@ simde_vaesdq_u8(simde_uint8x16_t data, simde_uint8x16_t key) {
     return simde_uint8x16_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARCH_ARM_AES)))
   #undef vaesdq_u8
   #define vaesdq_u8(data, key) simde_vaesdq_u8((data), (key))
 #endif
@@ -160,7 +162,8 @@ simde_vaesmcq_u8(simde_uint8x16_t data) {
     return simde_uint8x16_from_private(a_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARCH_ARM_AES)))
   #undef vaesmcq_u8
   #define vaesmcq_u8(data) simde_vaesmcq_u8((data))
 #endif
@@ -207,7 +210,8 @@ simde_vaesimcq_u8(simde_uint8x16_t data) {
     return simde_uint8x16_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARCH_ARM_AES)))
   #undef vaesimcq_u8
   #define vaesimcq_u8(data) simde_vaesimcq_u8((data))
 #endif

diff --git a/simde/arm/neon/bsl.h b/simde/arm/neon/bsl.h
@@ -62,7 +62,8 @@ simde_vbsl_f16(simde_uint16x4_t a, simde_float16x4_t b, simde_float16x4_t c) {
     return simde_vreinterpret_f16_u16(simde_uint16x4_from_private(r_));
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vbsl_f16
   #define vbsl_f16(a, b, c) simde_vbsl_f16((a), (b), (c))
 #endif
@@ -381,7 +382,8 @@ simde_vbslq_f16(simde_uint16x8_t a, simde_float16x8_t b, simde_float16x8_t c) {
     return simde_vreinterpretq_f16_u16(simde_uint16x8_from_private(r_));
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vbslq_f16
   #define vbslq_f16(a, b, c) simde_vbslq_f16((a), (b), (c))
 #endif

diff --git a/simde/arm/neon/cadd_rot270.h b/simde/arm/neon/cadd_rot270.h
@@ -41,8 +41,8 @@ _Pragma("clang diagnostic ignored \"-Wimplicit-float-conversion\"")
 SIMDE_FUNCTION_ATTRIBUTES
 simde_float16x4_t simde_vcadd_rot270_f16(simde_float16x4_t a, simde_float16x4_t b)
 {
-  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && SIMDE_ARCH_ARM_CHECK(8, 3) &&                                                   \
-      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) &&                                                  \
+  #if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
       (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))
     return vcadd_rot270_f16(a, b);
   #else
@@ -68,7 +68,10 @@ simde_float16x4_t simde_vcadd_rot270_f16(simde_float16x4_t a, simde_float16x4_t
     return simde_float16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcadd_rot270_f16
   #define vcadd_rot270_f16(a, b) simde_vcadd_rot270_f16(a, b)
 #endif
@@ -104,7 +107,10 @@ simde_float16x8_t simde_vcaddq_rot270_f16(simde_float16x8_t a, simde_float16x8_t
     return simde_float16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcaddq_rot270_f16
   #define vcaddq_rot270_f16(a, b) simde_vcaddq_rot270_f16(a, b)
 #endif
@@ -137,7 +143,10 @@ simde_float32x2_t simde_vcadd_rot270_f32(simde_float32x2_t a, simde_float32x2_t
     return simde_float32x2_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcadd_rot270_f32
   #define vcadd_rot270_f32(a, b) simde_vcadd_rot270_f32(a, b)
 #endif
@@ -171,7 +180,10 @@ simde_float32x4_t simde_vcaddq_rot270_f32(simde_float32x4_t a, simde_float32x4_t
     return simde_float32x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcaddq_rot270_f32
   #define vcaddq_rot270_f32(a, b) simde_vcaddq_rot270_f32(a, b)
 #endif
@@ -205,7 +217,10 @@ simde_float64x2_t simde_vcaddq_rot270_f64(simde_float64x2_t a, simde_float64x2_t
     return simde_float64x2_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcaddq_rot270_f64
   #define vcaddq_rot270_f64(a, b) simde_vcaddq_rot270_f64(a, b)
 #endif

diff --git a/simde/arm/neon/cadd_rot90.h b/simde/arm/neon/cadd_rot90.h
@@ -68,7 +68,10 @@ simde_float16x4_t simde_vcadd_rot90_f16(simde_float16x4_t a, simde_float16x4_t b
     return simde_float16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcadd_rot90_f16
   #define vcadd_rot90_f16(a, b) simde_vcadd_rot90_f16(a, b)
 #endif
@@ -104,7 +107,10 @@ simde_float16x8_t simde_vcaddq_rot90_f16(simde_float16x8_t a, simde_float16x8_t
     return simde_float16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcaddq_rot90_f16
   #define vcaddq_rot90_f16(a, b) simde_vcaddq_rot90_f16(a, b)
 #endif
@@ -137,7 +143,10 @@ simde_float32x2_t simde_vcadd_rot90_f32(simde_float32x2_t a, simde_float32x2_t b
     return simde_float32x2_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcadd_rot90_f32
   #define vcadd_rot90_f32(a, b) simde_vcadd_rot90_f32(a, b)
 #endif
@@ -171,7 +180,10 @@ simde_float32x4_t simde_vcaddq_rot90_f32(simde_float32x4_t a, simde_float32x4_t
     return simde_float32x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
   #undef vcaddq_rot90_f32
   #define vcaddq_rot90_f32(a, b) simde_vcaddq_rot90_f32(a, b)
 #endif
@@ -205,9 +217,12 @@ simde_float64x2_t simde_vcaddq_rot90_f64(simde_float64x2_t a, simde_float64x2_t
     return simde_float64x2_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
-#undef vcaddq_rot90_f64
-#define vcaddq_rot90_f64(a, b) simde_vcaddq_rot90_f64(a, b)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(SIMDE_ARCH_ARM_CHECK(8, 3) && \
+      (!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(9, 0, 0)) && \
+      (!defined(__clang__) || SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0))))
+  #undef vcaddq_rot90_f64
+  #define vcaddq_rot90_f64(a, b) simde_vcaddq_rot90_f64(a, b)
 #endif
 
 SIMDE_END_DECLS_

diff --git a/simde/arm/neon/cage.h b/simde/arm/neon/cage.h
@@ -47,7 +47,8 @@ simde_vcageh_f16(simde_float16_t a, simde_float16_t b) {
     return (simde_math_fabsf(a_) >= simde_math_fabsf(b_)) ? UINT16_MAX : UINT16_C(0);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vcageh_f16
   #define vcageh_f16(a, b) simde_vcageh_f16((a), (b))
 #endif
@@ -99,7 +100,8 @@ simde_vcage_f16(simde_float16x4_t a, simde_float16x4_t b) {
     return simde_uint16x4_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vcage_f16
   #define vcage_f16(a, b) simde_vcage_f16((a), (b))
 #endif
@@ -150,7 +152,8 @@ simde_vcageq_f16(simde_float16x8_t a, simde_float16x8_t b) {
     return simde_uint16x8_from_private(r_);
   #endif
 }
-#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES)
+#if defined(SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES) || (defined(SIMDE_ENABLE_NATIVE_ALIASES) && \
+    !(defined(SIMDE_ARM_NEON_FP16)))
   #undef vcageq_f16
   #define vcageq_f16(a, b) simde_vcageq_f16((a), (b))
 #endif