WIP

Author: serge-sans-paille

include/xsimd/arch/xsimd_neon.hpp

@@ -2957,24 +2957,24 @@ namespace xsimd
             {
                 uint8x8_t low = vreinterpret_u8_u64(vget_low_u64(vreinterpretq_u64_u32(self)));
                 uint8x8_t mask_lo = {
-                    V0 * 4 + 0,
-                    V0 * 4 + 1,
-                    V0 * 4 + 2,
-                    V0 * 4 + 3,
-                    V1 * 4 + 0,
-                    V1 * 4 + 1,
-                    V1 * 4 + 2,
-                    V1 * 4 + 3,
+                    static_cast<uint8_t>(V0 * 4 + 0),
+                    static_cast<uint8_t>(V0 * 4 + 1),
+                    static_cast<uint8_t>(V0 * 4 + 2),
+                    static_cast<uint8_t>(V0 * 4 + 3),
+                    static_cast<uint8_t>(V1 * 4 + 0),
+                    static_cast<uint8_t>(V1 * 4 + 1),
+                    static_cast<uint8_t>(V1 * 4 + 2),
+                    static_cast<uint8_t>(V1 * 4 + 3),
                 };
                 uint8x8_t mask_hi = {
-                    V2 * 4 + 0,
-                    V2 * 4 + 1,
-                    V2 * 4 + 2,
-                    V2 * 4 + 3,
-                    V3 * 4 + 0,
-                    V3 * 4 + 1,
-                    V3 * 4 + 2,
-                    V3 * 4 + 3,
+                    static_cast<uint8_t>(V2 * 4 + 0),
+                    static_cast<uint8_t>(V2 * 4 + 1),
+                    static_cast<uint8_t>(V2 * 4 + 2),
+                    static_cast<uint8_t>(V2 * 4 + 3),
+                    static_cast<uint8_t>(V3 * 4 + 0),
+                    static_cast<uint8_t>(V3 * 4 + 1),
+                    static_cast<uint8_t>(V3 * 4 + 2),
+                    static_cast<uint8_t>(V3 * 4 + 3),
                 };
                 uint8x8_t lo = vtbl1_u8(low, mask_lo);
                 uint8x8_t hi = vtbl1_u8(low, mask_hi);
@@ -2984,24 +2984,24 @@ namespace xsimd
             {
                 uint8x8_t high = vreinterpret_u8_u64(vget_high_u64(vreinterpretq_u64_u32(self)));
                 uint8x8_t mask_lo = {
-                    V0 * 4 + 0,
-                    V0 * 4 + 1,
-                    V0 * 4 + 2,
-                    V0 * 4 + 3,
-                    V1 * 4 + 0,
-                    V1 * 4 + 1,
-                    V1 * 4 + 2,
-                    V1 * 4 + 3,
+                    static_cast<uint8_t>((V0 - 2) * 4 + 0),
+                    static_cast<uint8_t>((V0 - 2) * 4 + 1),
+                    static_cast<uint8_t>((V0 - 2) * 4 + 2),
+                    static_cast<uint8_t>((V0 - 2) * 4 + 3),
+                    static_cast<uint8_t>((V1 - 2) * 4 + 0),
+                    static_cast<uint8_t>((V1 - 2) * 4 + 1),
+                    static_cast<uint8_t>((V1 - 2) * 4 + 2),
+                    static_cast<uint8_t>((V1 - 2) * 4 + 3),
                 };
                 uint8x8_t mask_hi = {
-                    V2 * 4 + 0,
-                    V2 * 4 + 1,
-                    V2 * 4 + 2,
-                    V2 * 4 + 3,
-                    V3 * 4 + 0,
-                    V3 * 4 + 1,
-                    V3 * 4 + 2,
-                    V3 * 4 + 3,
+                    static_cast<uint8_t>((V2 - 2) * 4 + 0),
+                    static_cast<uint8_t>((V2 - 2) * 4 + 1),
+                    static_cast<uint8_t>((V2 - 2) * 4 + 2),
+                    static_cast<uint8_t>((V2 - 2) * 4 + 3),
+                    static_cast<uint8_t>((V3 - 2) * 4 + 0),
+                    static_cast<uint8_t>((V3 - 2) * 4 + 1),
+                    static_cast<uint8_t>((V3 - 2) * 4 + 2),
+                    static_cast<uint8_t>((V3 - 2) * 4 + 3),
                 };
                 uint8x8_t lo = vtbl1_u8(high, mask_lo);
                 uint8x8_t hi = vtbl1_u8(high, mask_hi);