diff --git a/Grid/simd/Fujitsu_A64FX_asm_double.h b/Grid/simd/Fujitsu_A64FX_asm_double.h
index 86ebba68..4ab27919 100644
--- a/Grid/simd/Fujitsu_A64FX_asm_double.h
+++ b/Grid/simd/Fujitsu_A64FX_asm_double.h
@@ -283,7 +283,7 @@ asm ( \
 // MULT_2SPIN
 #define MULT_2SPIN_A64FXd(A)  \
 { \
-    const auto & ref(U[sU][A]); \
+    const auto & ref(U[sU](A)); \
 asm ( \
     "ld1d { z24.d }, p5/z, [%[fetchptr], -6, mul vl] \n\t" \
     "ld1d { z25.d }, p5/z, [%[fetchptr], -3, mul vl] \n\t" \
diff --git a/Grid/simd/Fujitsu_A64FX_asm_single.h b/Grid/simd/Fujitsu_A64FX_asm_single.h
index ea869777..d2ba4176 100644
--- a/Grid/simd/Fujitsu_A64FX_asm_single.h
+++ b/Grid/simd/Fujitsu_A64FX_asm_single.h
@@ -295,7 +295,7 @@ asm ( \
 // MULT_2SPIN
 #define MULT_2SPIN_A64FXf(A)  \
 { \
-    const auto & ref(U[sU][A]); \
+    const auto & ref(U[sU](A)); \
 asm ( \
     "ld1d { z24.f }, p5/z, [%[fetchptr], -6, mul vl] \n\t" \
     "ld1d { z25.f }, p5/z, [%[fetchptr], -3, mul vl] \n\t" \
diff --git a/Grid/simd/Fujitsu_A64FX_intrin_double.h b/Grid/simd/Fujitsu_A64FX_intrin_double.h
index c3bb2fcc..edfc7ab1 100644
--- a/Grid/simd/Fujitsu_A64FX_intrin_double.h
+++ b/Grid/simd/Fujitsu_A64FX_intrin_double.h
@@ -261,7 +261,7 @@ Author: Nils Meyer <nils.meyer@ur.de>
 // MULT_2SPIN
 #define MULT_2SPIN_A64FXd(A)  \
 { \
-    const auto & ref(U[sU][A]); \
+    const auto & ref(U[sU](A)); \
     U_00 = svld1(pg1, (float64_t*)((uint64_t)&ref[2][0] + -6 * 64));  \
     U_10 = svld1(pg1, (float64_t*)((uint64_t)&ref[2][0] + -3 * 64));  \
     U_20 = svld1(pg1, (float64_t*)((uint64_t)&ref[2][0] + 0 * 64));  \
diff --git a/Grid/simd/Fujitsu_A64FX_intrin_single.h b/Grid/simd/Fujitsu_A64FX_intrin_single.h
index b80c71ec..26a04dfd 100644
--- a/Grid/simd/Fujitsu_A64FX_intrin_single.h
+++ b/Grid/simd/Fujitsu_A64FX_intrin_single.h
@@ -268,7 +268,7 @@ Author: Nils Meyer <nils.meyer@ur.de>
 // MULT_2SPIN
 #define MULT_2SPIN_A64FXf(A)  \
 { \
-    const auto & ref(U[sU][A]); \
+    const auto & ref(U[sU](A)); \
     U_00 = svld1(pg1, (float32_t*)((uint64_t)&ref[2][0] + -6 * 64));  \
     U_10 = svld1(pg1, (float32_t*)((uint64_t)&ref[2][0] + -3 * 64));  \
     U_20 = svld1(pg1, (float32_t*)((uint64_t)&ref[2][0] + 0 * 64));  \