/*
* This file is part of the GROMACS molecular simulation package.
*
- * Copyright (c) 2014,2015, by the GROMACS development team, led by
+ * Copyright (c) 2014,2015,2019, 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.
/****************************************************
* DOUBLE PRECISION SIMD IMPLEMENTATION *
****************************************************/
-#define SimdDouble _fjsp_v2r8
-#define simdLoadD _fjsp_load_v2r8
-#define simdLoad1D(m) _fjsp_set_v2r8((*m), (*m))
-#define simdSet1D(a) _fjsp_set_v2r8(a, a)
-#define simdStoreD _fjsp_store_v2r8
-#define simdLoadUD simdLoadD
+#define SimdDouble _fjsp_v2r8
+#define simdLoadD _fjsp_load_v2r8
+#define simdLoad1D(m) _fjsp_set_v2r8((*m), (*m))
+#define simdSet1D(a) _fjsp_set_v2r8(a, a)
+#define simdStoreD _fjsp_store_v2r8
+#define simdLoadUD simdLoadD
/* No unaligned store of SimdDouble */
-#define simdSetZeroD _fjsp_setzero_v2r8
-#define simdAddD _fjsp_add_v2r8
-#define simdSubD _fjsp_sub_v2r8
-#define simdMulD _fjsp_mul_v2r8
-#define simdFmaddD(a, b, c) _fjsp_madd_v2r8(a, b, c)
-#define simdFmsubD(a, b, c) _fjsp_msub_v2r8(a, b, c)
-#define simdFnmaddD(a, b, c) _fjsp_nmsub_v2r8(a, b, c)
-#define simdFnmsubD(a, b, c) _fjsp_nmadd_v2r8(a, b, c)
-#define simdAndD _fjsp_and_v2r8
-#define simdAndNotD _fjsp_andnot1_v2r8
-#define simdOrD _fjsp_or_v2r8
-#define simdXorD _fjsp_xor_v2r8
-#define simdRsqrtD(x) _fjsp_rsqrta_v2r8(x)
-#define simdRcpD(x) _fjsp_rcpa_v2r8(x)
-#define simdAbsD(x) _fjsp_abs_v2r8(x)
-#define simdNegD(x) _fjsp_neg_v2r8(x)
-#define simdMaxD _fjsp_max_v2r8
-#define simdMinD _fjsp_min_v2r8
-#define simdRoundD(x) simdCvtI2D(simdCvtD2I(x))
-#define simdTruncD(x) simdCvtI2D(simdCvttD2I(x))
-#define simdFractionD(x) simdSubD(x, simdTruncD(x))
-#define simdGetExponentD simdGetExponentD_sparc64_hpc_ace
-#define simdGetMantissaD simdGetMantissaD_sparc64_hpc_ace
-#define simdSetExponentD simdSetExponentD_sparc64_hpc_ace
+#define simdSetZeroD _fjsp_setzero_v2r8
+#define simdAddD _fjsp_add_v2r8
+#define simdSubD _fjsp_sub_v2r8
+#define simdMulD _fjsp_mul_v2r8
+#define simdFmaddD(a, b, c) _fjsp_madd_v2r8(a, b, c)
+#define simdFmsubD(a, b, c) _fjsp_msub_v2r8(a, b, c)
+#define simdFnmaddD(a, b, c) _fjsp_nmsub_v2r8(a, b, c)
+#define simdFnmsubD(a, b, c) _fjsp_nmadd_v2r8(a, b, c)
+#define simdAndD _fjsp_and_v2r8
+#define simdAndNotD _fjsp_andnot1_v2r8
+#define simdOrD _fjsp_or_v2r8
+#define simdXorD _fjsp_xor_v2r8
+#define simdRsqrtD(x) _fjsp_rsqrta_v2r8(x)
+#define simdRcpD(x) _fjsp_rcpa_v2r8(x)
+#define simdAbsD(x) _fjsp_abs_v2r8(x)
+#define simdNegD(x) _fjsp_neg_v2r8(x)
+#define simdMaxD _fjsp_max_v2r8
+#define simdMinD _fjsp_min_v2r8
+#define simdRoundD(x) simdCvtI2D(simdCvtD2I(x))
+#define simdTruncD(x) simdCvtI2D(simdCvttD2I(x))
+#define simdFractionD(x) simdSubD(x, simdTruncD(x))
+#define simdGetExponentD simdGetExponentD_sparc64_hpc_ace
+#define simdGetMantissaD simdGetMantissaD_sparc64_hpc_ace
+#define simdSetExponentD simdSetExponentD_sparc64_hpc_ace
/* integer datatype corresponding to double: SimdDInt32 */
-#define SimdDInt32 _fjsp_v2r8
-#define simdLoadDI(m) simdLoadDI_sparc64_hpc_ace(m)
-#define simdSet1DI(i) simdSet1DI_sparc64_hpc_ace(i)
-#define simdStoreDI(m, x) simdStoreDI_sparc64_hpc_ace(m, x)
-#define simdLoadUDI simdLoadDI
+#define SimdDInt32 _fjsp_v2r8
+#define simdLoadDI(m) simdLoadDI_sparc64_hpc_ace(m)
+#define simdSet1DI(i) simdSet1DI_sparc64_hpc_ace(i)
+#define simdStoreDI(m, x) simdStoreDI_sparc64_hpc_ace(m, x)
+#define simdLoadUDI simdLoadDI
/* No unaligned store of SimdDInt32 */
-#define simdSetZeroDI _fjsp_setzero_v2r8
-#define simdCvtD2I simdCvtD2I_sparc64_hpc_ace
-#define simdCvttD2I _fjsp_dtox_v2r8
-#define simdCvtI2D _fjsp_xtod_v2r8
-#define simdExtractDI simdExtractDI_sparc64_hpc_ace
+#define simdSetZeroDI _fjsp_setzero_v2r8
+#define simdCvtD2I simdCvtD2I_sparc64_hpc_ace
+#define simdCvttD2I _fjsp_dtox_v2r8
+#define simdCvtI2D _fjsp_xtod_v2r8
+#define simdExtractDI simdExtractDI_sparc64_hpc_ace
/* Integer logical ops on SimdDInt32 */
-#define simdSlliDI simdSlliDI_sparc64_hpc_ace
-#define simdSrliDI simdSrliDI_sparc64_hpc_ace
-#define simdAndDI _fjsp_and_v2r8
-#define simdAndNotDI _fjsp_andnot1_v2r8
-#define simdOrDI _fjsp_or_v2r8
-#define simdXorDI _fjsp_xor_v2r8
+#define simdSlliDI simdSlliDI_sparc64_hpc_ace
+#define simdSrliDI simdSrliDI_sparc64_hpc_ace
+#define simdAndDI _fjsp_and_v2r8
+#define simdAndNotDI _fjsp_andnot1_v2r8
+#define simdOrDI _fjsp_or_v2r8
+#define simdXorDI _fjsp_xor_v2r8
/* Integer arithmetic ops on integer datatype corresponding to double */
/* Boolean & comparison operations on SimdDouble */
-#define SimdDBool _fjsp_v2r8
-#define simdCmpEqD _fjsp_cmpeq_v2r8
-#define simdCmpLtD _fjsp_cmplt_v2r8
-#define simdCmpLeD _fjsp_cmple_v2r8
-#define simdAndDB _fjsp_and_v2r8
-#define simdOrDB _fjsp_or_v2r8
-#define simdAnyTrueDB gmx_simd_anytrue_d_sparc64_hpc_ace
-#define simdMaskD _fjsp_and_v2r8
-#define simdMaskNotD(a, sel) _fjsp_andnot1_v2r8(sel, a)
+#define SimdDBool _fjsp_v2r8
+#define simdCmpEqD _fjsp_cmpeq_v2r8
+#define simdCmpLtD _fjsp_cmplt_v2r8
+#define simdCmpLeD _fjsp_cmple_v2r8
+#define simdAndDB _fjsp_and_v2r8
+#define simdOrDB _fjsp_or_v2r8
+#define simdAnyTrueDB gmx_simd_anytrue_d_sparc64_hpc_ace
+#define simdMaskD _fjsp_and_v2r8
+#define simdMaskNotD(a, sel) _fjsp_andnot1_v2r8(sel, a)
#define simdBlendD(a, b, sel) _fjsp_selmov_v2r8(b, a, sel)
-#define simdReduceD(a) simdReduceD_sparc64_hpc_ace(a)
+#define simdReduceD(a) simdReduceD_sparc64_hpc_ace(a)
/* No boolean & comparison operations on SimdDInt32 */
/* Float/double conversion */
-#define simdCvtF2D(f) (f)
-#define simdCvtD2F(d) (d)
+#define simdCvtF2D(f) (f)
+#define simdCvtD2F(d) (d)
/****************************************************
* DOUBLE PRECISION IMPLEMENTATION HELPER FUNCTIONS *
****************************************************/
-static inline SimdDInt32
-simdLoadDI_sparc64_hpc_ace(const int *m)
+static inline SimdDInt32 simdLoadDI_sparc64_hpc_ace(const int* m)
{
- union
- {
- _fjsp_v2r8 simd;
- long long int i[2];
- }
- conv;
+ union {
+ _fjsp_v2r8 simd;
+ long long int i[2];
+ } conv;
conv.i[0] = m[0];
conv.i[1] = m[1];
- return _fjsp_load_v2r8( (double *) &(conv.simd) );
+ return _fjsp_load_v2r8((double*)&(conv.simd));
}
-static inline void
-simdStoreDI_sparc64_hpc_ace(int *m, SimdDInt32 x)
+static inline void simdStoreDI_sparc64_hpc_ace(int* m, SimdDInt32 x)
{
- union
- {
- _fjsp_v2r8 simd;
- long long int i[2];
- }
- conv;
+ union {
+ _fjsp_v2r8 simd;
+ long long int i[2];
+ } conv;
- _fjsp_store_v2r8( (double *) &(conv.simd), x );
+ _fjsp_store_v2r8((double*)&(conv.simd), x);
m[0] = conv.i[0];
m[1] = conv.i[1];
}
-static inline SimdDInt32
-simdSet1DI_sparc64_hpc_ace(int i)
+static inline SimdDInt32 simdSet1DI_sparc64_hpc_ace(int i)
{
- union
- {
- _fjsp_v2r8 simd;
- long long int i[2];
- }
- conv;
+ union {
+ _fjsp_v2r8 simd;
+ long long int i[2];
+ } conv;
conv.i[0] = i;
conv.i[1] = i;
- return _fjsp_load_v2r8( (double *) &(conv.simd) );
+ return _fjsp_load_v2r8((double*)&(conv.simd));
}
-static inline int
-simdExtractDI_sparc64_hpc_ace(SimdDInt32 x, int i)
+static inline int simdExtractDI_sparc64_hpc_ace(SimdDInt32 x, int i)
{
long long int res;
/* This conditional should be optimized away at compile time */
if (i == 0)
{
- _fjsp_storel_v2r8((double *)&res, x);
+ _fjsp_storel_v2r8((double*)&res, x);
}
else
{
- _fjsp_storeh_v2r8((double *)&res, x);
+ _fjsp_storeh_v2r8((double*)&res, x);
}
return (int)res;
}
-static inline SimdDInt32
-simdSlliDI_sparc64_hpc_ace(SimdDInt32 x, int i)
+static inline SimdDInt32 simdSlliDI_sparc64_hpc_ace(SimdDInt32 x, int i)
{
- _fjsp_v2i8 ix = *((_fjsp_v2i8 *)&x);
- ix = _fjsp_slli_v2i8(ix, i);
- x = *((_fjsp_v2r8 *)&ix);
+ _fjsp_v2i8 ix = *((_fjsp_v2i8*)&x);
+ ix = _fjsp_slli_v2i8(ix, i);
+ x = *((_fjsp_v2r8*)&ix);
return x;
}
-static inline SimdDInt32
-simdSrliDI_sparc64_hpc_ace(SimdDInt32 x, int i)
+static inline SimdDInt32 simdSrliDI_sparc64_hpc_ace(SimdDInt32 x, int i)
{
- _fjsp_v2i8 ix = *((_fjsp_v2i8 *)&x);
- ix = _fjsp_srli_v2i8(ix, i);
- x = *((_fjsp_v2r8 *)&ix);
+ _fjsp_v2i8 ix = *((_fjsp_v2i8*)&x);
+ ix = _fjsp_srli_v2i8(ix, i);
+ x = *((_fjsp_v2r8*)&ix);
return x;
}
-static inline SimdDInt32
-simdCvtD2I_sparc64_hpc_ace(SimdDouble x)
+static inline SimdDInt32 simdCvtD2I_sparc64_hpc_ace(SimdDouble x)
{
_fjsp_v2r8 signbit = _fjsp_set_v2r8(-0.0, -0.0);
_fjsp_v2r8 half = _fjsp_set_v2r8(0.5, 0.5);
return _fjsp_dtox_v2r8(x);
}
-static inline int
-gmx_simd_anytrue_d_sparc64_hpc_ace(SimdDBool x)
+static inline int gmx_simd_anytrue_d_sparc64_hpc_ace(SimdDBool x)
{
long long int i;
x = _fjsp_or_v2r8(x, _fjsp_unpackhi_v2r8(x, x));
- _fjsp_storel_v2r8((double *)&i, x);
+ _fjsp_storel_v2r8((double*)&i, x);
return (i != 0LL);
}
-static inline double
-simdReduceD_sparc64_hpc_ace(SimdDouble x)
+static inline double simdReduceD_sparc64_hpc_ace(SimdDouble x)
{
double d;
x = _fjsp_add_v2r8(x, _fjsp_unpackhi_v2r8(x, x));
}
-static inline SimdDouble
-simdGetExponentD_sparc64_hpc_ace(SimdDouble x)
+static inline SimdDouble simdGetExponentD_sparc64_hpc_ace(SimdDouble x)
{
/* HPC-ACE cannot cast _fjsp_v2r8 to _fjsp_v4i4, so to perform shifts we
* would need to store and reload. Since we are only operating on two
* numbers it is likely more efficient to do the operations directly on
* normal registers.
*/
- const std::int64_t expmask = 0x7ff0000000000000LL;
- const std::int64_t expbias = 1023LL;
+ const std::int64_t expmask = 0x7ff0000000000000LL;
+ const std::int64_t expbias = 1023LL;
- union
- {
- _fjsp_v2r8 simd;
- long long int i[2];
- }
- conv;
+ union {
+ _fjsp_v2r8 simd;
+ long long int i[2];
+ } conv;
- _fjsp_store_v2r8( (double *) &conv.simd, x);
+ _fjsp_store_v2r8((double*)&conv.simd, x);
conv.i[0] = ((conv.i[0] & expmask) >> 52) - expbias;
conv.i[1] = ((conv.i[1] & expmask) >> 52) - expbias;
- x = _fjsp_load_v2r8( (double *) &conv.simd);
+ x = _fjsp_load_v2r8((double*)&conv.simd);
return _fjsp_xtod_v2r8(x);
}
-static inline SimdDouble
-simdGetMantissaD_sparc64_hpc_ace(SimdDouble x)
+static inline SimdDouble simdGetMantissaD_sparc64_hpc_ace(SimdDouble x)
{
- std::int64_t mantmask[2] = {0x000fffffffffffffLL, 0x000fffffffffffffLL};
- SimdDouble one = _fjsp_set_v2r8(1.0, 1.0);
+ std::int64_t mantmask[2] = { 0x000fffffffffffffLL, 0x000fffffffffffffLL };
+ SimdDouble one = _fjsp_set_v2r8(1.0, 1.0);
- x = _fjsp_and_v2r8(x, _fjsp_load_v2r8((double *)mantmask));
+ x = _fjsp_and_v2r8(x, _fjsp_load_v2r8((double*)mantmask));
return _fjsp_or_v2r8(x, one);
}
-static inline SimdDouble
-simdSetExponentD_sparc64_hpc_ace(SimdDouble x)
+static inline SimdDouble simdSetExponentD_sparc64_hpc_ace(SimdDouble x)
{
- const std::int64_t expbias = 1023;
- union
- {
- _fjsp_v2r8 simd;
- long long int i[2];
- }
- conv;
+ const std::int64_t expbias = 1023;
+ union {
+ _fjsp_v2r8 simd;
+ long long int i[2];
+ } conv;
- _fjsp_store_v2r8( (double *) &conv.simd, simdCvtD2I_sparc64_hpc_ace(x));
+ _fjsp_store_v2r8((double*)&conv.simd, simdCvtD2I_sparc64_hpc_ace(x));
conv.i[0] = (conv.i[0] + expbias) << 52;
conv.i[1] = (conv.i[1] + expbias) << 52;
- return _fjsp_load_v2r8( (double *) &conv.simd);
+ return _fjsp_load_v2r8((double*)&conv.simd);
}
#endif /* GMX_SIMD_IMPL_SPARC64_HPC_ACE_SIMD_DOUBLE_H */
biod.pnpi.spb.ru / alexxy / gromacs.git / blobdiff