/*
* This file is part of the GROMACS molecular simulation package.
*
- * Copyright (c) 2014,2015,2016,2017,2018, by the GROMACS development team, led by
+ * Copyright (c) 2014,2015,2016,2017,2018,2019, by the GROMACS development team, led by
* Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
* and including many others, as listed in the AUTHORS file in the
* top-level source directory and at http://www.gromacs.org.
class SimdDouble
{
- public:
- SimdDouble() {}
+public:
+ SimdDouble() {}
- // gcc-4.9 does not recognize that we use the parameter
- SimdDouble(double gmx_unused d) : simdInternal_(vec_splats(d)) {}
+ // gcc-4.9 does not recognize that we use the parameter
+ SimdDouble(double gmx_unused d) : simdInternal_(vec_splats(d)) {}
- // Internal utility constructor to simplify return statements
- SimdDouble(__vector double simd) : simdInternal_(simd) {}
+ // Internal utility constructor to simplify return statements
+ SimdDouble(__vector double simd) : simdInternal_(simd) {}
- __vector double simdInternal_;
+ __vector double simdInternal_;
};
class SimdDInt32
{
- public:
- SimdDInt32() {}
+public:
+ SimdDInt32() {}
- // gcc-4.9 does not recognize that we use the parameter
- SimdDInt32(std::int32_t gmx_unused i) : simdInternal_(vec_splats(i)) {}
+ // gcc-4.9 does not recognize that we use the parameter
+ SimdDInt32(std::int32_t gmx_unused i) : simdInternal_(vec_splats(i)) {}
- // Internal utility constructor to simplify return statements
- SimdDInt32(__vector signed int simd) : simdInternal_(simd) {}
+ // Internal utility constructor to simplify return statements
+ SimdDInt32(__vector signed int simd) : simdInternal_(simd) {}
- __vector signed int simdInternal_;
+ __vector signed int simdInternal_;
};
class SimdDBool
{
- public:
- SimdDBool() {}
+public:
+ SimdDBool() {}
- SimdDBool(bool b) : simdInternal_(reinterpret_cast<__vector vsxBool long long>(vec_splats( b ? 0xFFFFFFFFFFFFFFFFULL : 0))) {}
+ SimdDBool(bool b) :
+ simdInternal_(reinterpret_cast<__vector vsxBool long long>(vec_splats(b ? 0xFFFFFFFFFFFFFFFFULL : 0)))
+ {
+ }
- // Internal utility constructor to simplify return statements
- SimdDBool(__vector vsxBool long long simd) : simdInternal_(simd) {}
+ // Internal utility constructor to simplify return statements
+ SimdDBool(__vector vsxBool long long simd) : simdInternal_(simd) {}
- __vector vsxBool long long simdInternal_;
+ __vector vsxBool long long simdInternal_;
};
class SimdDIBool
{
- public:
- SimdDIBool() {}
+public:
+ SimdDIBool() {}
- SimdDIBool(bool b) : simdInternal_(reinterpret_cast<__vector vsxBool int>(vec_splats( b ? 0xFFFFFFFF : 0))) {}
+ SimdDIBool(bool b) :
+ simdInternal_(reinterpret_cast<__vector vsxBool int>(vec_splats(b ? 0xFFFFFFFF : 0)))
+ {
+ }
- // Internal utility constructor to simplify return statements
- SimdDIBool(__vector vsxBool int simd) : simdInternal_(simd) {}
+ // Internal utility constructor to simplify return statements
+ SimdDIBool(__vector vsxBool int simd) : simdInternal_(simd) {}
- __vector vsxBool int simdInternal_;
+ __vector vsxBool int simdInternal_;
};
// Note that the interfaces we use here have been a mess in xlc;
// currently version 13.1.5 is required.
-static inline SimdDouble gmx_simdcall
-simdLoad(const double *m, SimdDoubleTag = {})
+static inline SimdDouble gmx_simdcall simdLoad(const double* m, SimdDoubleTag = {})
{
- return {
+ return
+ {
#if defined(__ibmxl__)
- vec_ld(0, m)
+ vec_ld(0, m)
#else
-# if __GNUC__ < 7
- *reinterpret_cast<const __vector double *>(m)
-# else
- vec_vsx_ld(0, m)
-# endif
+# if __GNUC__ < 7
+ *reinterpret_cast<const __vector double*>(m)
+# else
+ vec_vsx_ld(0, m)
+# endif
#endif
};
}
-static inline void gmx_simdcall
-store(double *m, SimdDouble a)
+static inline void gmx_simdcall store(double* m, SimdDouble a)
{
#if defined(__ibmxl__)
vec_st(a.simdInternal_, 0, m);
#else
-# if __GNUC__ < 7
- *reinterpret_cast<__vector double *>(m) = a.simdInternal_;
-# else
+# if __GNUC__ < 7
+ *reinterpret_cast<__vector double*>(m) = a.simdInternal_;
+# else
vec_vsx_st(a.simdInternal_, 0, m);
-# endif
+# endif
#endif
}
-static inline SimdDouble gmx_simdcall
-simdLoadU(const double *m, SimdDoubleTag = {})
+static inline SimdDouble gmx_simdcall simdLoadU(const double* m, SimdDoubleTag = {})
{
- return {
+ return
+ {
#if defined(__ibmxl__)
- vec_xl(0, m)
+ vec_xl(0, m)
#else
-# if __GNUC__ < 7
- *reinterpret_cast<const __vector double *>(m)
-# else
- vec_vsx_ld(0, m)
-# endif
+# if __GNUC__ < 7
+ *reinterpret_cast<const __vector double*>(m)
+# else
+ vec_vsx_ld(0, m)
+# endif
#endif
};
}
-static inline void gmx_simdcall
-storeU(double *m, SimdDouble a)
+static inline void gmx_simdcall storeU(double* m, SimdDouble a)
{
#if defined(__ibmxl__)
vec_xst(a.simdInternal_, 0, m);
#else
-# if __GNUC__ < 7
- *reinterpret_cast<__vector double *>(m) = a.simdInternal_;
-# else
+# if __GNUC__ < 7
+ *reinterpret_cast<__vector double*>(m) = a.simdInternal_;
+# else
vec_vsx_st(a.simdInternal_, 0, m);
-# endif
+# endif
#endif
}
-static inline SimdDouble gmx_simdcall
-setZeroD()
+static inline SimdDouble gmx_simdcall setZeroD()
{
- return {
- vec_splats(0.0)
- };
+ return { vec_splats(0.0) };
}
-static inline SimdDInt32 gmx_simdcall
-simdLoad(const std::int32_t * m, SimdDInt32Tag)
+static inline SimdDInt32 gmx_simdcall simdLoad(const std::int32_t* m, SimdDInt32Tag)
{
__vector signed int t0, t1;
const __vector unsigned char perm = { 0, 1, 2, 3, 0, 1, 2, 3, 16, 17, 18, 19, 16, 17, 18, 19 };
- t0 = vec_splats(m[0]);
- t1 = vec_splats(m[1]);
- return {
- vec_perm(t0, t1, perm)
- };
+ t0 = vec_splats(m[0]);
+ t1 = vec_splats(m[1]);
+ return { vec_perm(t0, t1, perm) };
}
// gcc-4.9 does not understand that arguments to vec_extract() are used
-static inline void gmx_simdcall
-store(std::int32_t * m, SimdDInt32 gmx_unused x)
+static inline void gmx_simdcall store(std::int32_t* m, SimdDInt32 gmx_unused x)
{
m[0] = vec_extract(x.simdInternal_, 0);
m[1] = vec_extract(x.simdInternal_, 2);
}
-static inline SimdDInt32 gmx_simdcall
-simdLoadU(const std::int32_t *m, SimdDInt32Tag)
+static inline SimdDInt32 gmx_simdcall simdLoadU(const std::int32_t* m, SimdDInt32Tag)
{
return simdLoad(m, SimdDInt32Tag());
}
-static inline void gmx_simdcall
-storeU(std::int32_t * m, SimdDInt32 a)
+static inline void gmx_simdcall storeU(std::int32_t* m, SimdDInt32 a)
{
return store(m, a);
}
-static inline SimdDInt32 gmx_simdcall
-setZeroDI()
+static inline SimdDInt32 gmx_simdcall setZeroDI()
{
- return {
- vec_splats(static_cast<int>(0))
- };
+ return { vec_splats(static_cast<int>(0)) };
}
// gcc-4.9 does not detect that vec_extract() uses its argument
template<int index>
-static inline std::int32_t gmx_simdcall
-extract(SimdDInt32 gmx_unused a)
+static inline std::int32_t gmx_simdcall extract(SimdDInt32 gmx_unused a)
{
- return vec_extract(a.simdInternal_, 2*index);
+ return vec_extract(a.simdInternal_, 2 * index);
}
-static inline SimdDouble gmx_simdcall
-operator&(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator&(SimdDouble a, SimdDouble b)
{
- return {
- vec_and(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_and(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-andNot(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall andNot(SimdDouble a, SimdDouble b)
{
- return {
- vec_andc(b.simdInternal_, a.simdInternal_)
- };
+ return { vec_andc(b.simdInternal_, a.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-operator|(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator|(SimdDouble a, SimdDouble b)
{
- return {
- vec_or(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_or(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-operator^(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator^(SimdDouble a, SimdDouble b)
{
- return {
- vec_xor(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_xor(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-operator+(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator+(SimdDouble a, SimdDouble b)
{
- return {
- vec_add(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_add(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-operator-(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator-(SimdDouble a, SimdDouble b)
{
- return {
- vec_sub(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_sub(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-operator-(SimdDouble x)
+static inline SimdDouble gmx_simdcall operator-(SimdDouble x)
{
- return {
- -x.simdInternal_
- };
+ return { -x.simdInternal_ };
}
-static inline SimdDouble gmx_simdcall
-operator*(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall operator*(SimdDouble a, SimdDouble b)
{
- return {
- vec_mul(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_mul(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-fma(SimdDouble a, SimdDouble b, SimdDouble c)
+static inline SimdDouble gmx_simdcall fma(SimdDouble a, SimdDouble b, SimdDouble c)
{
- return {
- vec_madd(a.simdInternal_, b.simdInternal_, c.simdInternal_)
- };
+ return { vec_madd(a.simdInternal_, b.simdInternal_, c.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-fms(SimdDouble a, SimdDouble b, SimdDouble c)
+static inline SimdDouble gmx_simdcall fms(SimdDouble a, SimdDouble b, SimdDouble c)
{
- return {
- vec_msub(a.simdInternal_, b.simdInternal_, c.simdInternal_)
- };
+ return { vec_msub(a.simdInternal_, b.simdInternal_, c.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-fnma(SimdDouble a, SimdDouble b, SimdDouble c)
+static inline SimdDouble gmx_simdcall fnma(SimdDouble a, SimdDouble b, SimdDouble c)
{
- return {
- vec_nmsub(a.simdInternal_, b.simdInternal_, c.simdInternal_)
- };
+ return { vec_nmsub(a.simdInternal_, b.simdInternal_, c.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-fnms(SimdDouble a, SimdDouble b, SimdDouble c)
+static inline SimdDouble gmx_simdcall fnms(SimdDouble a, SimdDouble b, SimdDouble c)
{
- return {
- vec_nmadd(a.simdInternal_, b.simdInternal_, c.simdInternal_)
- };
+ return { vec_nmadd(a.simdInternal_, b.simdInternal_, c.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-rsqrt(SimdDouble x)
+static inline SimdDouble gmx_simdcall rsqrt(SimdDouble x)
{
- return {
- vec_rsqrte(x.simdInternal_)
- };
+ return { vec_rsqrte(x.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-rcp(SimdDouble x)
+static inline SimdDouble gmx_simdcall rcp(SimdDouble x)
{
- return {
- vec_re(x.simdInternal_)
- };
+ return { vec_re(x.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-maskAdd(SimdDouble a, SimdDouble b, SimdDBool m)
+static inline SimdDouble gmx_simdcall maskAdd(SimdDouble a, SimdDouble b, SimdDBool m)
{
- return {
- vec_add(a.simdInternal_, vec_and(b.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_)))
- };
+ return { vec_add(a.simdInternal_,
+ vec_and(b.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_))) };
}
-static inline SimdDouble gmx_simdcall
-maskzMul(SimdDouble a, SimdDouble b, SimdDBool m)
+static inline SimdDouble gmx_simdcall maskzMul(SimdDouble a, SimdDouble b, SimdDBool m)
{
SimdDouble prod = a * b;
- return {
- vec_and(prod.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_and(prod.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-maskzFma(SimdDouble a, SimdDouble b, SimdDouble c, SimdDBool m)
+static inline SimdDouble gmx_simdcall maskzFma(SimdDouble a, SimdDouble b, SimdDouble c, SimdDBool m)
{
SimdDouble prod = fma(a, b, c);
- return {
- vec_and(prod.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_and(prod.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-maskzRsqrt(SimdDouble x, SimdDBool m)
+static inline SimdDouble gmx_simdcall maskzRsqrt(SimdDouble x, SimdDBool m)
{
#ifndef NDEBUG
x.simdInternal_ = vec_sel(vec_splats(1.0), x.simdInternal_, m.simdInternal_);
#endif
- return {
- vec_and(vec_rsqrte(x.simdInternal_), reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_and(vec_rsqrte(x.simdInternal_), reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-maskzRcp(SimdDouble x, SimdDBool m)
+static inline SimdDouble gmx_simdcall maskzRcp(SimdDouble x, SimdDBool m)
{
#ifndef NDEBUG
x.simdInternal_ = vec_sel(vec_splats(1.0), x.simdInternal_, m.simdInternal_);
#endif
- return {
- vec_and(vec_re(x.simdInternal_), reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_and(vec_re(x.simdInternal_), reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-abs(SimdDouble x)
+static inline SimdDouble gmx_simdcall abs(SimdDouble x)
{
- return {
- vec_abs( x.simdInternal_ )
- };
+ return { vec_abs(x.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-max(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall max(SimdDouble a, SimdDouble b)
{
- return {
- vec_max(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_max(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-min(SimdDouble a, SimdDouble b)
+static inline SimdDouble gmx_simdcall min(SimdDouble a, SimdDouble b)
{
- return {
- vec_min(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_min(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDouble gmx_simdcall
-round(SimdDouble x)
+static inline SimdDouble gmx_simdcall round(SimdDouble x)
{
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
-// gcc up to at least version 4.9 does not have vec_round() in double precision - use inline asm
+ // gcc up to at least version 4.9 does not have vec_round() in double precision - use inline asm
__vector double res;
- __asm__ ("xvrdpi %x0,%x1" : "=wd" (res) : "wd" (x.simdInternal_));
- return {
- res
- };
+ __asm__("xvrdpi %x0,%x1" : "=wd"(res) : "wd"(x.simdInternal_));
+ return { res };
#else
- return {
- vec_round( x.simdInternal_ )
- };
+ return { vec_round(x.simdInternal_) };
#endif
}
-static inline SimdDouble gmx_simdcall
-trunc(SimdDouble x)
+static inline SimdDouble gmx_simdcall trunc(SimdDouble x)
{
- return {
- vec_trunc( x.simdInternal_ )
- };
+ return { vec_trunc(x.simdInternal_) };
}
-static inline SimdDouble
-frexp(SimdDouble value, SimdDInt32 * exponent)
+static inline SimdDouble frexp(SimdDouble value, SimdDInt32* exponent)
{
- const __vector double exponentMask = reinterpret_cast<__vector double>(vec_splats(0x7FF0000000000000ULL));
+ const __vector double exponentMask =
+ reinterpret_cast<__vector double>(vec_splats(0x7FF0000000000000ULL));
const __vector signed int exponentBias = vec_splats(1022);
const __vector double half = vec_splats(0.5);
__vector signed int iExponent;
- iExponent = reinterpret_cast<__vector signed int>(vec_and(value.simdInternal_, exponentMask));
+ iExponent = reinterpret_cast<__vector signed int>(vec_and(value.simdInternal_, exponentMask));
// The data is in the upper half of each double (corresponding to elements 1 and 3).
// First shift 52-32=20bits, and then permute to swap element 0 with 1 and element 2 with 3
// For big endian they are in opposite order, so then we simply skip the swap.
- iExponent = vec_sr(iExponent, vec_splats(20U));
+ iExponent = vec_sr(iExponent, vec_splats(20U));
#ifndef __BIG_ENDIAN__
- const __vector unsigned char perm = {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11};
- iExponent = vec_perm(iExponent, iExponent, perm);
+ const __vector unsigned char perm = { 4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11 };
+ iExponent = vec_perm(iExponent, iExponent, perm);
#endif
iExponent = vec_sub(iExponent, exponentBias);
exponent->simdInternal_ = iExponent;
- return {
- vec_or(vec_andc(value.simdInternal_, exponentMask), half)
- };
+ return { vec_or(vec_andc(value.simdInternal_, exponentMask), half) };
}
-template <MathOptimization opt = MathOptimization::Safe>
-static inline SimdDouble
-ldexp(SimdDouble value, SimdDInt32 exponent)
+template<MathOptimization opt = MathOptimization::Safe>
+static inline SimdDouble ldexp(SimdDouble value, SimdDInt32 exponent)
{
- const __vector signed int exponentBias = vec_splats(1023);
- __vector signed int iExponent;
+ const __vector signed int exponentBias = vec_splats(1023);
+ __vector signed int iExponent;
#ifdef __BIG_ENDIAN__
- const __vector unsigned char perm = {0, 1, 2, 3, 16, 17, 18, 19, 8, 9, 10, 11, 16, 17, 18, 19};
+ const __vector unsigned char perm = { 0, 1, 2, 3, 16, 17, 18, 19, 8, 9, 10, 11, 16, 17, 18, 19 };
#else
- const __vector unsigned char perm = {16, 17, 18, 19, 0, 1, 2, 3, 16, 17, 18, 19, 8, 9, 10, 11};
+ const __vector unsigned char perm = { 16, 17, 18, 19, 0, 1, 2, 3, 16, 17, 18, 19, 8, 9, 10, 11 };
#endif
iExponent = vec_add(exponent.simdInternal_, exponentBias);
iExponent = vec_sl(iExponent, vec_splats(20U));
iExponent = vec_perm(iExponent, vec_splats(0), perm);
- return {
- vec_mul(value.simdInternal_, reinterpret_cast<__vector double>(iExponent))
- };
+ return { vec_mul(value.simdInternal_, reinterpret_cast<__vector double>(iExponent)) };
}
-static inline double gmx_simdcall
-reduce(SimdDouble x)
+static inline double gmx_simdcall reduce(SimdDouble x)
{
const __vector unsigned char perm = { 8, 9, 10, 11, 12, 13, 14, 15, 0, 1, 2, 3, 4, 5, 6, 7 };
#ifdef __xlC__
/* old xlc version 12 does not understand vec_perm() with double arguments */
- x.simdInternal_ = vec_add(x.simdInternal_,
- reinterpret_cast<__vector double>(vec_perm(reinterpret_cast<__vector signed int>(x.simdInternal_),
- reinterpret_cast<__vector signed int>(x.simdInternal_), perm)));
+ x.simdInternal_ = vec_add(
+ x.simdInternal_, reinterpret_cast<__vector double>(vec_perm(
+ reinterpret_cast<__vector signed int>(x.simdInternal_),
+ reinterpret_cast<__vector signed int>(x.simdInternal_), perm)));
#else
x.simdInternal_ = vec_add(x.simdInternal_, vec_perm(x.simdInternal_, x.simdInternal_, perm));
#endif
return vec_extract(x.simdInternal_, 0);
}
-static inline SimdDBool gmx_simdcall
-operator==(SimdDouble a, SimdDouble b)
+static inline SimdDBool gmx_simdcall operator==(SimdDouble a, SimdDouble b)
{
- return {
- vec_cmpeq(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_cmpeq(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDBool gmx_simdcall
-operator!=(SimdDouble a, SimdDouble b)
+static inline SimdDBool gmx_simdcall operator!=(SimdDouble a, SimdDouble b)
{
- return {
- reinterpret_cast<__vector vsxBool long long>(vec_or(reinterpret_cast<__vector signed int>(vec_cmpgt(a.simdInternal_, b.simdInternal_)),
- reinterpret_cast<__vector signed int>(vec_cmplt(a.simdInternal_, b.simdInternal_))))
- };
+ return { reinterpret_cast<__vector vsxBool long long>(vec_or(
+ reinterpret_cast<__vector signed int>(vec_cmpgt(a.simdInternal_, b.simdInternal_)),
+ reinterpret_cast<__vector signed int>(vec_cmplt(a.simdInternal_, b.simdInternal_)))) };
}
-static inline SimdDBool gmx_simdcall
-operator<(SimdDouble a, SimdDouble b)
+static inline SimdDBool gmx_simdcall operator<(SimdDouble a, SimdDouble b)
{
- return {
- vec_cmplt(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_cmplt(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDBool gmx_simdcall
-operator<=(SimdDouble a, SimdDouble b)
+static inline SimdDBool gmx_simdcall operator<=(SimdDouble a, SimdDouble b)
{
- return {
- vec_cmple(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_cmple(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDBool gmx_simdcall
-testBits(SimdDouble a)
+static inline SimdDBool gmx_simdcall testBits(SimdDouble a)
{
#ifdef __POWER8_VECTOR__
// Power8 VSX has proper support for operations on long long integers
- return {
- vec_cmpgt(reinterpret_cast<__vector unsigned long long>(a.simdInternal_), vec_splats(0ULL))
- };
+ return { vec_cmpgt(reinterpret_cast<__vector unsigned long long>(a.simdInternal_), vec_splats(0ULL)) };
#else
// No support for long long operations.
// Start with comparing 32-bit subfields bitwise by casting to integers
- __vector vsxBool int tmp = vec_cmpgt( reinterpret_cast<__vector unsigned int>(a.simdInternal_), vec_splats(0U));
+ __vector vsxBool int tmp =
+ vec_cmpgt(reinterpret_cast<__vector unsigned int>(a.simdInternal_), vec_splats(0U));
// Shuffle low/high 32-bit fields of tmp into tmp2
- const __vector unsigned char perm = {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11};
- __vector vsxBool int tmp2 = vec_perm(tmp, tmp, perm);
+ const __vector unsigned char perm = { 4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11 };
+ __vector vsxBool int tmp2 = vec_perm(tmp, tmp, perm);
// Return the or:d parts of tmp & tmp2
- return {
- reinterpret_cast<__vector vsxBool long long>(vec_or(tmp, tmp2))
- };
+ return { reinterpret_cast<__vector vsxBool long long>(vec_or(tmp, tmp2)) };
#endif
}
-static inline SimdDBool gmx_simdcall
-operator&&(SimdDBool a, SimdDBool b)
+static inline SimdDBool gmx_simdcall operator&&(SimdDBool a, SimdDBool b)
{
- return {
- reinterpret_cast<__vector vsxBool long long>(vec_and(reinterpret_cast<__vector signed int>(a.simdInternal_), reinterpret_cast<__vector signed int>(b.simdInternal_)))
- };
+ return { reinterpret_cast<__vector vsxBool long long>(
+ vec_and(reinterpret_cast<__vector signed int>(a.simdInternal_),
+ reinterpret_cast<__vector signed int>(b.simdInternal_))) };
}
-static inline SimdDBool gmx_simdcall
-operator||(SimdDBool a, SimdDBool b)
+static inline SimdDBool gmx_simdcall operator||(SimdDBool a, SimdDBool b)
{
- return {
- reinterpret_cast<__vector vsxBool long long>(vec_or(reinterpret_cast<__vector signed int>(a.simdInternal_), reinterpret_cast<__vector signed int>(b.simdInternal_)))
- };
+ return { reinterpret_cast<__vector vsxBool long long>(
+ vec_or(reinterpret_cast<__vector signed int>(a.simdInternal_),
+ reinterpret_cast<__vector signed int>(b.simdInternal_))) };
}
-static inline bool gmx_simdcall
-anyTrue(SimdDBool a)
+static inline bool gmx_simdcall anyTrue(SimdDBool a)
{
- return vec_any_ne(reinterpret_cast<__vector vsxBool int>(a.simdInternal_), reinterpret_cast<__vector vsxBool int>(vec_splats(0)));
+ return vec_any_ne(reinterpret_cast<__vector vsxBool int>(a.simdInternal_),
+ reinterpret_cast<__vector vsxBool int>(vec_splats(0)));
}
-static inline SimdDouble gmx_simdcall
-selectByMask(SimdDouble a, SimdDBool m)
+static inline SimdDouble gmx_simdcall selectByMask(SimdDouble a, SimdDBool m)
{
- return {
- vec_and(a.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_and(a.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-selectByNotMask(SimdDouble a, SimdDBool m)
+static inline SimdDouble gmx_simdcall selectByNotMask(SimdDouble a, SimdDBool m)
{
- return {
- vec_andc(a.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_))
- };
+ return { vec_andc(a.simdInternal_, reinterpret_cast<__vector double>(m.simdInternal_)) };
}
-static inline SimdDouble gmx_simdcall
-blend(SimdDouble a, SimdDouble b, SimdDBool sel)
+static inline SimdDouble gmx_simdcall blend(SimdDouble a, SimdDouble b, SimdDBool sel)
{
- return {
- vec_sel(a.simdInternal_, b.simdInternal_, sel.simdInternal_)
- };
+ return { vec_sel(a.simdInternal_, b.simdInternal_, sel.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator&(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator&(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_and(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_and(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-andNot(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall andNot(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_andc(b.simdInternal_, a.simdInternal_)
- };
+ return { vec_andc(b.simdInternal_, a.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator|(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator|(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_or(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_or(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator^(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator^(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_xor(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_xor(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator+(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator+(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_add(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_add(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator-(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator-(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_sub(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_sub(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-operator*(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDInt32 gmx_simdcall operator*(SimdDInt32 a, SimdDInt32 b)
{
- return {
- a.simdInternal_ * b.simdInternal_
- };
+ return { a.simdInternal_ * b.simdInternal_ };
}
-static inline SimdDIBool gmx_simdcall
-operator==(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDIBool gmx_simdcall operator==(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_cmpeq(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_cmpeq(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDIBool gmx_simdcall
-testBits(SimdDInt32 a)
+static inline SimdDIBool gmx_simdcall testBits(SimdDInt32 a)
{
- return {
- vec_cmpgt( reinterpret_cast<__vector unsigned int>(a.simdInternal_), vec_splats(0U))
- };
+ return { vec_cmpgt(reinterpret_cast<__vector unsigned int>(a.simdInternal_), vec_splats(0U)) };
}
-static inline SimdDIBool gmx_simdcall
-operator<(SimdDInt32 a, SimdDInt32 b)
+static inline SimdDIBool gmx_simdcall operator<(SimdDInt32 a, SimdDInt32 b)
{
- return {
- vec_cmplt(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_cmplt(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDIBool gmx_simdcall
-operator&&(SimdDIBool a, SimdDIBool b)
+static inline SimdDIBool gmx_simdcall operator&&(SimdDIBool a, SimdDIBool b)
{
- return {
- vec_and(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_and(a.simdInternal_, b.simdInternal_) };
}
-static inline SimdDIBool gmx_simdcall
-operator||(SimdDIBool a, SimdDIBool b)
+static inline SimdDIBool gmx_simdcall operator||(SimdDIBool a, SimdDIBool b)
{
- return {
- vec_or(a.simdInternal_, b.simdInternal_)
- };
+ return { vec_or(a.simdInternal_, b.simdInternal_) };
}
-static inline bool gmx_simdcall
-anyTrue(SimdDIBool a)
+static inline bool gmx_simdcall anyTrue(SimdDIBool a)
{
return vec_any_ne(a.simdInternal_, reinterpret_cast<__vector vsxBool int>(vec_splats(0)));
}
-static inline SimdDInt32 gmx_simdcall
-selectByMask(SimdDInt32 a, SimdDIBool m)
+static inline SimdDInt32 gmx_simdcall selectByMask(SimdDInt32 a, SimdDIBool m)
{
- return {
- vec_and(a.simdInternal_, reinterpret_cast<__vector signed int>(m.simdInternal_))
- };
+ return { vec_and(a.simdInternal_, reinterpret_cast<__vector signed int>(m.simdInternal_)) };
}
-static inline SimdDInt32 gmx_simdcall
-selectByNotMask(SimdDInt32 a, SimdDIBool m)
+static inline SimdDInt32 gmx_simdcall selectByNotMask(SimdDInt32 a, SimdDIBool m)
{
- return {
- vec_andc(a.simdInternal_, reinterpret_cast<__vector signed int>(m.simdInternal_))
- };
+ return { vec_andc(a.simdInternal_, reinterpret_cast<__vector signed int>(m.simdInternal_)) };
}
-static inline SimdDInt32 gmx_simdcall
-blend(SimdDInt32 a, SimdDInt32 b, SimdDIBool sel)
+static inline SimdDInt32 gmx_simdcall blend(SimdDInt32 a, SimdDInt32 b, SimdDIBool sel)
{
- return {
- vec_sel(a.simdInternal_, b.simdInternal_, sel.simdInternal_)
- };
+ return { vec_sel(a.simdInternal_, b.simdInternal_, sel.simdInternal_) };
}
-static inline SimdDInt32 gmx_simdcall
-cvttR2I(SimdDouble a)
+static inline SimdDInt32 gmx_simdcall cvttR2I(SimdDouble a)
{
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
-// gcc up to at least version 6.1 is missing intrinsics for converting double to/from int - use inline asm
- const __vector unsigned char perm = {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11};
+ // gcc up to at least version 6.1 is missing intrinsics for converting double to/from int - use inline asm
+ const __vector unsigned char perm = { 4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11 };
__vector double ix;
- __asm__ ("xvcvdpsxws %x0,%x1" : "=wa" (ix) : "wd" (a.simdInternal_));
+ __asm__("xvcvdpsxws %x0,%x1" : "=wa"(ix) : "wd"(a.simdInternal_));
- return {
- reinterpret_cast<__vector signed int>(vec_perm(ix, ix, perm))
- };
+ return { reinterpret_cast<__vector signed int>(vec_perm(ix, ix, perm)) };
#else
- return {
- vec_cts(a.simdInternal_, 0)
- };
+ return { vec_cts(a.simdInternal_, 0) };
#endif
}
-static inline SimdDInt32 gmx_simdcall
-cvtR2I(SimdDouble a)
+static inline SimdDInt32 gmx_simdcall cvtR2I(SimdDouble a)
{
return cvttR2I(round(a));
}
-static inline SimdDouble gmx_simdcall
-cvtI2R(SimdDInt32 a)
+static inline SimdDouble gmx_simdcall cvtI2R(SimdDInt32 a)
{
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
-// gcc up to at least version 4.9 is missing intrinsics for converting double to/from int - use inline asm
- __vector double x;
-#ifndef __BIG_ENDIAN__
- const __vector unsigned char perm = {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11};
- a.simdInternal_ = vec_perm(a.simdInternal_, a.simdInternal_, perm);
-#endif
+ // gcc up to at least version 4.9 is missing intrinsics for converting double to/from int - use inline asm
+ __vector double x;
+# ifndef __BIG_ENDIAN__
+ const __vector unsigned char perm = { 4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11 };
+ a.simdInternal_ = vec_perm(a.simdInternal_, a.simdInternal_, perm);
+# endif
- __asm__ ("xvcvsxwdp %x0,%x1" : "=wd" (x) : "wa" (a.simdInternal_));
+ __asm__("xvcvsxwdp %x0,%x1" : "=wd"(x) : "wa"(a.simdInternal_));
- return {
- x
- };
+ return { x };
#else
- return {
- vec_ctd(a.simdInternal_, 0)
- };
+ return { vec_ctd(a.simdInternal_, 0) };
#endif
}
-static inline SimdDIBool gmx_simdcall
-cvtB2IB(SimdDBool a)
+static inline SimdDIBool gmx_simdcall cvtB2IB(SimdDBool a)
{
- return {
- reinterpret_cast<__vector vsxBool int>(a.simdInternal_)
- };
+ return { reinterpret_cast<__vector vsxBool int>(a.simdInternal_) };
}
-static inline SimdDBool gmx_simdcall
-cvtIB2B(SimdDIBool a)
+static inline SimdDBool gmx_simdcall cvtIB2B(SimdDIBool a)
{
- return {
- reinterpret_cast<__vector vsxBool long long>(a.simdInternal_)
- };
+ return { reinterpret_cast<__vector vsxBool long long>(a.simdInternal_) };
}
-static inline void gmx_simdcall
-cvtF2DD(SimdFloat f, SimdDouble *d0, SimdDouble *d1)
+static inline void gmx_simdcall cvtF2DD(SimdFloat f, SimdDouble* d0, SimdDouble* d1)
{
__vector float fA, fB;
- fA = vec_mergeh(f.simdInternal_, f.simdInternal_); /* 0011 */
- fB = vec_mergel(f.simdInternal_, f.simdInternal_); /* 2233 */
+ fA = vec_mergeh(f.simdInternal_, f.simdInternal_); /* 0011 */
+ fB = vec_mergel(f.simdInternal_, f.simdInternal_); /* 2233 */
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
// gcc-4.9 is missing double-to-float/float-to-double conversions.
- __asm__ ("xvcvspdp %x0,%x1" : "=wd" (d0->simdInternal_) : "wf" (fA));
- __asm__ ("xvcvspdp %x0,%x1" : "=wd" (d1->simdInternal_) : "wf" (fB));
+ __asm__("xvcvspdp %x0,%x1" : "=wd"(d0->simdInternal_) : "wf"(fA));
+ __asm__("xvcvspdp %x0,%x1" : "=wd"(d1->simdInternal_) : "wf"(fB));
#else
- d0->simdInternal_ = vec_cvf(fA); /* 01 */
- d1->simdInternal_ = vec_cvf(fB); /* 23 */
+ d0->simdInternal_ = vec_cvf(fA); /* 01 */
+ d1->simdInternal_ = vec_cvf(fB); /* 23 */
#endif
}
-static inline SimdFloat gmx_simdcall
-cvtDD2F(SimdDouble d0, SimdDouble d1)
+static inline SimdFloat gmx_simdcall cvtDD2F(SimdDouble d0, SimdDouble d1)
{
__vector float fA, fB, fC, fD, fE;
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
// gcc-4.9 is missing double-to-float/float-to-double conversions.
- __asm__ ("xvcvdpsp %x0,%x1" : "=wf" (fA) : "wd" (d0.simdInternal_));
- __asm__ ("xvcvdpsp %x0,%x1" : "=wf" (fB) : "wd" (d1.simdInternal_));
+ __asm__("xvcvdpsp %x0,%x1" : "=wf"(fA) : "wd"(d0.simdInternal_));
+ __asm__("xvcvdpsp %x0,%x1" : "=wf"(fB) : "wd"(d1.simdInternal_));
#else
- fA = vec_cvf(d0.simdInternal_); /* 0x1x */
- fB = vec_cvf(d1.simdInternal_); /* 2x3x */
+ fA = vec_cvf(d0.simdInternal_); /* 0x1x */
+ fB = vec_cvf(d1.simdInternal_); /* 2x3x */
#endif
- fC = vec_mergeh(fA, fB); /* 02xx */
- fD = vec_mergel(fA, fB); /* 13xx */
- fE = vec_mergeh(fC, fD); /* 0123 */
- return {
- fE
- };
+ fC = vec_mergeh(fA, fB); /* 02xx */
+ fD = vec_mergel(fA, fB); /* 13xx */
+ fE = vec_mergeh(fC, fD); /* 0123 */
+ return { fE };
}
-static inline SimdDouble gmx_simdcall
-copysign(SimdDouble x, SimdDouble y)
+static inline SimdDouble gmx_simdcall copysign(SimdDouble x, SimdDouble y)
{
#if defined(__GNUC__) && !defined(__ibmxl__) && !defined(__xlC__)
__vector double res;
- __asm__ ("xvcpsgndp %x0,%x1,%x2" : "=wd" (res) : "wd" (y.simdInternal_), "wd" (x.simdInternal_));
- return {
- res
- };
+ __asm__("xvcpsgndp %x0,%x1,%x2" : "=wd"(res) : "wd"(y.simdInternal_), "wd"(x.simdInternal_));
+ return { res };
#else
- return {
- vec_cpsgn(y.simdInternal_, x.simdInternal_)
- };
+ return { vec_cpsgn(y.simdInternal_, x.simdInternal_) };
#endif
}
-} // namespace gmx
+} // namespace gmx
#endif // GMX_SIMD_IMPLEMENTATION_IBM_VSX_SIMD_DOUBLE_H
biod.pnpi.spb.ru / alexxy / gromacs.git / blobdiff