diff --git a/include/bmat8_impl.hpp b/include/bmat8_impl.hpp
index 913e94f2..3b2b0da0 100644
--- a/include/bmat8_impl.hpp
+++ b/include/bmat8_impl.hpp
@@ -253,7 +253,7 @@ constexpr std::array<epu8, 4> masks {{
 static const epu8 shiftres {1, 2, 4, 8, 0x10, 0x20, 0x40, 0x80};
 
 inline void update_bitset(epu8 block, epu8 &set0, epu8 &set1) {
-    static const epu8 bound08 = simde_mm_slli_epi32(epu8id, 3); // shift for *8
+    static const epu8 bound08 = simde_mm_slli_epi32(static_cast<simde__m128i>(epu8id), 3); // shift for *8
 static const epu8 bound18 = bound08 + Epu8(0x80);
     for (size_t slice8 = 0; slice8 < 16; slice8++) {
         epu8 bm5 = Epu8(0xf8) & block; /* 11111000 */
diff --git a/include/perm16_impl.hpp b/include/perm16_impl.hpp
index 0408fc4d..b2d74a45 100644
--- a/include/perm16_impl.hpp
+++ b/include/perm16_impl.hpp
@@ -129,7 +129,8 @@ inline Transf16::Transf16(uint64_t compressed) {
 }
 
 inline Transf16::operator uint64_t() const {
-    epu8 res = static_cast<epu8>(simde_mm_slli_epi32(v, 4));
+    epu8 res =
+        static_cast<epu8>(simde_mm_slli_epi32(static_cast<simde__m128i>(v), 4));
     res = HPCombi::permuted(res, hilo_exchng) + v;
     return simde_mm_extract_epi64(res, 0);
 }
@@ -218,7 +219,9 @@ inline Perm16 Perm16::inverse_sort() const {
     // G++-7 compile this shift by 3 additions.
     // epu8 res = (v << 4) + one().v;
     // I call directly the shift intrinsic
-    epu8 res = static_cast<epu8>(simde_mm_slli_epi32(v, 4)) + one().v;
+    epu8 res = static_cast<epu8>(
+                   simde_mm_slli_epi32(static_cast<simde__m128i>(v), 4)) +
+               one().v;
     res = sorted(res) & Epu8(0x0F);
     return res;
 }