Enable SIMD register calling convention with gmx_simdcall

[alexxy/gromacs.git] / src / gromacs / simd / simd_math.h

diff --git a/src/gromacs/simd/simd_math.h b/src/gromacs/simd/simd_math.h
index 4c03a250bf84fff2cb5f666ad7d9304aa75a9643..1f6dc08361ade3a6ee3c77a1cba1c839b10c9a2a 100644 (file)
--- a/src/gromacs/simd/simd_math.h
+++ b/src/gromacs/simd/simd_math.h
@@ -101,7 +101,7 @@
  * \param d term 4 (multiple values)
  * \return sum of terms 1-4 (multiple values)
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_sum4_f(gmx_simd_float_t a, gmx_simd_float_t b,
                 gmx_simd_float_t c, gmx_simd_float_t d)
 {
@@ -120,7 +120,7 @@ gmx_simd_sum4_f(gmx_simd_float_t a, gmx_simd_float_t b,
  * with the exception that negative zero is not considered to be negative
  * on architectures where \ref GMX_SIMD_HAVE_LOGICAL is not set.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_xor_sign_f(gmx_simd_float_t a, gmx_simd_float_t b)
 {
 #ifdef GMX_SIMD_HAVE_LOGICAL
@@ -139,7 +139,7 @@ gmx_simd_xor_sign_f(gmx_simd_float_t a, gmx_simd_float_t b)
  *  \param x  The reference (starting) value x for which we want 1/sqrt(x).
  *  \return   An improved approximation with roughly twice as many bits of accuracy.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_rsqrt_iter_f(gmx_simd_float_t lu, gmx_simd_float_t x)
 {
 #    ifdef GMX_SIMD_HAVE_FMA
@@ -156,7 +156,7 @@ gmx_simd_rsqrt_iter_f(gmx_simd_float_t lu, gmx_simd_float_t x)
  *  \param x Argument that must be >0. This routine does not check arguments.
  *  \return 1/sqrt(x). Result is undefined if your argument was invalid.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_invsqrt_f(gmx_simd_float_t x)
 {
     gmx_simd_float_t lu = gmx_simd_rsqrt_f(x);
@@ -184,7 +184,7 @@ gmx_simd_invsqrt_f(gmx_simd_float_t x)
  *  In particular for double precision we can sometimes calculate square root
  *  pairs slightly faster by using single precision until the very last step.
  */
-static gmx_inline void
+static gmx_inline void gmx_simdcall
 gmx_simd_invsqrt_pair_f(gmx_simd_float_t x0,    gmx_simd_float_t x1,
                         gmx_simd_float_t *out0, gmx_simd_float_t *out1)
 {
@@ -201,7 +201,7 @@ gmx_simd_invsqrt_pair_f(gmx_simd_float_t x0,    gmx_simd_float_t x1,
  *  \param x  The reference (starting) value x for which we want 1/x.
  *  \return   An improved approximation with roughly twice as many bits of accuracy.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_rcp_iter_f(gmx_simd_float_t lu, gmx_simd_float_t x)
 {
     return gmx_simd_mul_f(lu, gmx_simd_fnmadd_f(lu, x, gmx_simd_set1_f(2.0f)));
@@ -214,7 +214,7 @@ gmx_simd_rcp_iter_f(gmx_simd_float_t lu, gmx_simd_float_t x)
  *  \param x Argument that must be nonzero. This routine does not check arguments.
  *  \return 1/x. Result is undefined if your argument was invalid.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_inv_f(gmx_simd_float_t x)
 {
     gmx_simd_float_t lu = gmx_simd_rcp_f(x);
@@ -238,7 +238,7 @@ gmx_simd_inv_f(gmx_simd_float_t x)
  *  \return sqrt(x). If x=0, the result will correctly be set to 0.
  *          The result is undefined if the input value is negative.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_sqrt_f(gmx_simd_float_t x)
 {
     gmx_simd_fbool_t  mask;
@@ -257,7 +257,7 @@ gmx_simd_sqrt_f(gmx_simd_float_t x)
  * \result The natural logarithm of x. Undefined if argument is invalid.
  */
 #ifndef gmx_simd_log_f
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_log_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t  half       = gmx_simd_set1_f(0.5f);
@@ -301,7 +301,7 @@ gmx_simd_log_f(gmx_simd_float_t x)
  * \param x Argument.
  * \result 2^x. Undefined if input argument caused overflow.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_exp2_f(gmx_simd_float_t x)
 {
     /* Lower bound: Disallow numbers that would lead to an IEEE fp exponent reaching +-127. */
@@ -348,7 +348,7 @@ gmx_simd_exp2_f(gmx_simd_float_t x)
  * \result exp(x). Undefined if input argument caused overflow,
  * which can happen if abs(x) \> 7e13.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_exp_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t  argscale     = gmx_simd_set1_f(1.44269504088896341f);
@@ -398,7 +398,7 @@ gmx_simd_exp_f(gmx_simd_float_t x)
  * This routine achieves very close to full precision, but we do not care about
  * the last bit or the subnormal result range.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_erf_f(gmx_simd_float_t x)
 {
     /* Coefficients for minimax approximation of erf(x)=x*P(x^2) in range [-1,1] */
@@ -521,7 +521,7 @@ gmx_simd_erf_f(gmx_simd_float_t x)
  * (think results that are in the ballpark of 10^-30 for single precision,
  * or 10^-200 for double) since that is not relevant for MD.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_erfc_f(gmx_simd_float_t x)
 {
     /* Coefficients for minimax approximation of erf(x)=x*P(x^2) in range [-1,1] */
@@ -699,7 +699,7 @@ gmx_simd_erfc_f(gmx_simd_float_t x)
  * magnitudes of the argument we inherently begin to lose accuracy due to the
  * argument reduction, despite using extended precision arithmetics internally.
  */
-static gmx_inline void
+static gmx_inline void gmx_simdcall
 gmx_simd_sincos_f(gmx_simd_float_t x, gmx_simd_float_t *sinval, gmx_simd_float_t *cosval)
 {
     /* Constants to subtract Pi/4*x from y while minimizing precision loss */
@@ -820,7 +820,7 @@ gmx_simd_sincos_f(gmx_simd_float_t x, gmx_simd_float_t *sinval, gmx_simd_float_t
  * \attention Do NOT call both sin & cos if you need both results, since each of them
  * will then call \ref gmx_simd_sincos_r and waste a factor 2 in performance.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_sin_f(gmx_simd_float_t x)
 {
     gmx_simd_float_t s, c;
@@ -838,7 +838,7 @@ gmx_simd_sin_f(gmx_simd_float_t x)
  * \attention Do NOT call both sin & cos if you need both results, since each of them
  * will then call \ref gmx_simd_sincos_r and waste a factor 2 in performance.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_cos_f(gmx_simd_float_t x)
 {
     gmx_simd_float_t s, c;
@@ -853,7 +853,7 @@ gmx_simd_cos_f(gmx_simd_float_t x)
  * \param x The argument to evaluate tan for
  * \result Tan(x)
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_tan_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t  argred0         = gmx_simd_set1_f(1.5703125);
@@ -929,7 +929,7 @@ gmx_simd_tan_f(gmx_simd_float_t x)
  * \param x The argument to evaluate asin for
  * \result Asin(x)
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_asin_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t limitlow   = gmx_simd_set1_f(1e-4f);
@@ -981,7 +981,7 @@ gmx_simd_asin_f(gmx_simd_float_t x)
  * \param x The argument to evaluate acos for
  * \result Acos(x)
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_acos_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t one       = gmx_simd_set1_f(1.0f);
@@ -1018,7 +1018,7 @@ gmx_simd_acos_f(gmx_simd_float_t x)
  * \param x The argument to evaluate atan for
  * \result Atan(x), same argument/value range as standard math library.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_atan_f(gmx_simd_float_t x)
 {
     const gmx_simd_float_t halfpi    = gmx_simd_set1_f(M_PI/2);
@@ -1071,7 +1071,7 @@ gmx_simd_atan_f(gmx_simd_float_t x)
  * of any concern in Gromacs, and in particular it will not affect calculations
  * of angles from vectors.
  */
-static gmx_inline gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_atan2_f(gmx_simd_float_t y, gmx_simd_float_t x)
 {
     const gmx_simd_float_t pi          = gmx_simd_set1_f(M_PI);
@@ -1179,7 +1179,7 @@ gmx_simd_atan2_f(gmx_simd_float_t y, gmx_simd_float_t x)
  * For \f$\beta r \geq 7206\f$ the return value can be inf or NaN.
  *
  */
-static gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_pmecorrF_f(gmx_simd_float_t z2)
 {
     const gmx_simd_float_t  FN6      = gmx_simd_set1_f(-1.7357322914161492954e-8f);
@@ -1259,7 +1259,7 @@ gmx_simd_pmecorrF_f(gmx_simd_float_t z2)
  * when added to \f$1/r\f$ the error will be insignificant.
  * For \f$\beta r \geq 7142\f$ the return value can be inf or NaN.
  */
-static gmx_simd_float_t
+static gmx_inline gmx_simd_float_t gmx_simdcall
 gmx_simd_pmecorrV_f(gmx_simd_float_t z2)
 {
     const gmx_simd_float_t  VN6      = gmx_simd_set1_f(1.9296833005951166339e-8f);
@@ -1315,7 +1315,7 @@ gmx_simd_pmecorrV_f(gmx_simd_float_t z2)
  *
  * \copydetails gmx_simd_sum4_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_sum4_d(gmx_simd_double_t a, gmx_simd_double_t b,
                 gmx_simd_double_t c, gmx_simd_double_t d)
 {
@@ -1334,7 +1334,7 @@ gmx_simd_sum4_d(gmx_simd_double_t a, gmx_simd_double_t b,
  * with the exception that negative zero is not considered to be negative
  * on architectures where \ref GMX_SIMD_HAVE_LOGICAL is not set.
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_xor_sign_d(gmx_simd_double_t a, gmx_simd_double_t b)
 {
 #ifdef GMX_SIMD_HAVE_LOGICAL
@@ -1348,7 +1348,7 @@ gmx_simd_xor_sign_d(gmx_simd_double_t a, gmx_simd_double_t b)
  *
  * \copydetails gmx_simd_rsqrt_iter_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_rsqrt_iter_d(gmx_simd_double_t lu, gmx_simd_double_t x)
 {
 #ifdef GMX_SIMD_HAVE_FMA
@@ -1363,7 +1363,7 @@ gmx_simd_rsqrt_iter_d(gmx_simd_double_t lu, gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_invsqrt_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_invsqrt_d(gmx_simd_double_t x)
 {
     gmx_simd_double_t lu = gmx_simd_rsqrt_d(x);
@@ -1386,7 +1386,7 @@ gmx_simd_invsqrt_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_invsqrt_pair_f
  */
-static gmx_inline void
+static gmx_inline void gmx_simdcall
 gmx_simd_invsqrt_pair_d(gmx_simd_double_t x0,    gmx_simd_double_t x1,
                         gmx_simd_double_t *out0, gmx_simd_double_t *out1)
 {
@@ -1426,7 +1426,7 @@ gmx_simd_invsqrt_pair_d(gmx_simd_double_t x0,    gmx_simd_double_t x1,
  *
  * \copydetails gmx_simd_rcp_iter_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_rcp_iter_d(gmx_simd_double_t lu, gmx_simd_double_t x)
 {
     return gmx_simd_mul_d(lu, gmx_simd_fnmadd_d(lu, x, gmx_simd_set1_d(2.0)));
@@ -1436,7 +1436,7 @@ gmx_simd_rcp_iter_d(gmx_simd_double_t lu, gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_inv_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_inv_d(gmx_simd_double_t x)
 {
     gmx_simd_double_t lu = gmx_simd_rcp_d(x);
@@ -1459,7 +1459,7 @@ gmx_simd_inv_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_sqrt_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_sqrt_d(gmx_simd_double_t x)
 {
     gmx_simd_dbool_t   mask;
@@ -1474,7 +1474,7 @@ gmx_simd_sqrt_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_log_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_log_d(gmx_simd_double_t x)
 {
     const gmx_simd_double_t  half       = gmx_simd_set1_d(0.5);
@@ -1519,7 +1519,7 @@ gmx_simd_log_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_exp2_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_exp2_d(gmx_simd_double_t x)
 {
     const gmx_simd_double_t  arglimit      = gmx_simd_set1_d(1022.0);
@@ -1565,7 +1565,7 @@ gmx_simd_exp2_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_exp_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_exp_d(gmx_simd_double_t x)
 {
     const gmx_simd_double_t  argscale      = gmx_simd_set1_d(1.44269504088896340735992468100);
@@ -1618,7 +1618,7 @@ gmx_simd_exp_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_erf_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_erf_d(gmx_simd_double_t x)
 {
     /* Coefficients for minimax approximation of erf(x)=x*(CAoffset + P(x^2)/Q(x^2)) in range [-0.75,0.75] */
@@ -1803,7 +1803,7 @@ gmx_simd_erf_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_erfc_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_erfc_d(gmx_simd_double_t x)
 {
     /* Coefficients for minimax approximation of erf(x)=x*(CAoffset + P(x^2)/Q(x^2)) in range [-0.75,0.75] */
@@ -1988,7 +1988,7 @@ gmx_simd_erfc_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_sincos_f
  */
-static gmx_inline void
+static gmx_inline void gmx_simdcall
 gmx_simd_sincos_d(gmx_simd_double_t x, gmx_simd_double_t *sinval, gmx_simd_double_t *cosval)
 {
     /* Constants to subtract Pi/4*x from y while minimizing precision loss */
@@ -2115,7 +2115,7 @@ gmx_simd_sincos_d(gmx_simd_double_t x, gmx_simd_double_t *sinval, gmx_simd_doubl
  *
  * \copydetails gmx_simd_sin_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_sin_d(gmx_simd_double_t x)
 {
     gmx_simd_double_t s, c;
@@ -2127,7 +2127,7 @@ gmx_simd_sin_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_cos_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_cos_d(gmx_simd_double_t x)
 {
     gmx_simd_double_t s, c;
@@ -2139,7 +2139,7 @@ gmx_simd_cos_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_tan_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_tan_d(gmx_simd_double_t x)
 {
     const gmx_simd_double_t  argred0         = gmx_simd_set1_d(2*0.78539816290140151978);
@@ -2230,7 +2230,7 @@ gmx_simd_tan_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_asin_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_asin_d(gmx_simd_double_t x)
 {
     /* Same algorithm as cephes library */
@@ -2357,7 +2357,7 @@ gmx_simd_asin_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_acos_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_acos_d(gmx_simd_double_t x)
 {
     const gmx_simd_double_t one        = gmx_simd_set1_d(1.0);
@@ -2390,7 +2390,7 @@ gmx_simd_acos_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_atan_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_atan_d(gmx_simd_double_t x)
 {
     /* Same algorithm as cephes library */
@@ -2475,7 +2475,7 @@ gmx_simd_atan_d(gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_atan2_f
  */
-static gmx_inline gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_atan2_d(gmx_simd_double_t y, gmx_simd_double_t x)
 {
     const gmx_simd_double_t pi          = gmx_simd_set1_d(M_PI);
@@ -2507,7 +2507,7 @@ gmx_simd_atan2_d(gmx_simd_double_t y, gmx_simd_double_t x)
  *
  * \copydetails gmx_simd_pmecorrF_f
  */
-static gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_pmecorrF_d(gmx_simd_double_t z2)
 {
     const gmx_simd_double_t  FN10     = gmx_simd_set1_d(-8.0072854618360083154e-14);
@@ -2564,7 +2564,7 @@ gmx_simd_pmecorrF_d(gmx_simd_double_t z2)
  *
  * \copydetails gmx_simd_pmecorrV_f
  */
-static gmx_simd_double_t
+static gmx_inline gmx_simd_double_t gmx_simdcall
 gmx_simd_pmecorrV_d(gmx_simd_double_t z2)
 {
     const gmx_simd_double_t  VN9      = gmx_simd_set1_d(-9.3723776169321855475e-13);
@@ -2633,7 +2633,7 @@ gmx_simd_pmecorrV_d(gmx_simd_double_t z2)
  *
  * \copydetails gmx_simd_sum4_f
  */
-static gmx_inline gmx_simd4_float_t
+static gmx_inline gmx_simd4_float_t gmx_simdcall
 gmx_simd4_sum4_f(gmx_simd4_float_t a, gmx_simd4_float_t b,
                  gmx_simd4_float_t c, gmx_simd4_float_t d)
 {
@@ -2644,7 +2644,7 @@ gmx_simd4_sum4_f(gmx_simd4_float_t a, gmx_simd4_float_t b,
  *
  * \copydetails gmx_simd_rsqrt_iter_f
  */
-static gmx_inline gmx_simd4_float_t
+static gmx_inline gmx_simd4_float_t gmx_simdcall
 gmx_simd4_rsqrt_iter_f(gmx_simd4_float_t lu, gmx_simd4_float_t x)
 {
 #    ifdef GMX_SIMD_HAVE_FMA
@@ -2658,7 +2658,7 @@ gmx_simd4_rsqrt_iter_f(gmx_simd4_float_t lu, gmx_simd4_float_t x)
  *
  * \copydetails gmx_simd_invsqrt_f
  */
-static gmx_inline gmx_simd4_float_t
+static gmx_inline gmx_simd4_float_t gmx_simdcall
 gmx_simd4_invsqrt_f(gmx_simd4_float_t x)
 {
     gmx_simd4_float_t lu = gmx_simd4_rsqrt_f(x);
@@ -2688,7 +2688,7 @@ gmx_simd4_invsqrt_f(gmx_simd4_float_t x)
  *
  * \copydetails gmx_simd_sum4_f
  */
-static gmx_inline gmx_simd4_double_t
+static gmx_inline gmx_simd4_double_t gmx_simdcall
 gmx_simd4_sum4_d(gmx_simd4_double_t a, gmx_simd4_double_t b,
                  gmx_simd4_double_t c, gmx_simd4_double_t d)
 {
@@ -2699,7 +2699,7 @@ gmx_simd4_sum4_d(gmx_simd4_double_t a, gmx_simd4_double_t b,
  *
  * \copydetails gmx_simd_rsqrt_iter_f
  */
-static gmx_inline gmx_simd4_double_t
+static gmx_inline gmx_simd4_double_t gmx_simdcall
 gmx_simd4_rsqrt_iter_d(gmx_simd4_double_t lu, gmx_simd4_double_t x)
 {
 #ifdef GMX_SIMD_HAVE_FMA
@@ -2713,7 +2713,7 @@ gmx_simd4_rsqrt_iter_d(gmx_simd4_double_t lu, gmx_simd4_double_t x)
  *
  * \copydetails gmx_simd_invsqrt_f
  */
-static gmx_inline gmx_simd4_double_t
+static gmx_inline gmx_simd4_double_t gmx_simdcall
 gmx_simd4_invsqrt_d(gmx_simd4_double_t x)
 {
     gmx_simd4_double_t lu = gmx_simd4_rsqrt_d(x);