#include "gromacs/simd/vector_operations.h"
#include "gromacs/utility/basedefinitions.h"
#ifdef CALC_COUL_EWALD
-#include "gromacs/math/utilities.h"
+# include "gromacs/math/utilities.h"
#endif
#include "config.h"
#ifndef GMX_SIMD_J_UNROLL_SIZE
-#error "Need to define GMX_SIMD_J_UNROLL_SIZE before including the 4xn kernel common header file"
+# error "Need to define GMX_SIMD_J_UNROLL_SIZE before including the 4xn kernel common header file"
#endif
-#define UNROLLI 4
-#define UNROLLJ (GMX_SIMD_REAL_WIDTH/GMX_SIMD_J_UNROLL_SIZE)
+#define UNROLLI 4
+#define UNROLLJ (GMX_SIMD_REAL_WIDTH / GMX_SIMD_J_UNROLL_SIZE)
static_assert(UNROLLI == c_nbnxnCpuIClusterSize, "UNROLLI should match the i-cluster size");
/* The stride of all the atom data arrays is max(UNROLLI,unrollj) */
#if GMX_SIMD_REAL_WIDTH >= UNROLLI
-#define STRIDE (GMX_SIMD_REAL_WIDTH/GMX_SIMD_J_UNROLL_SIZE)
+# define STRIDE (GMX_SIMD_REAL_WIDTH / GMX_SIMD_J_UNROLL_SIZE)
#else
-#define STRIDE (UNROLLI)
+# define STRIDE (UNROLLI)
#endif
#if !defined GMX_NBNXN_SIMD_2XNN && !defined GMX_NBNXN_SIMD_4XN
-#error "Must define an NBNxN kernel flavour before including NBNxN kernel utility functions"
+# error "Must define an NBNxN kernel flavour before including NBNxN kernel utility functions"
#endif
// We use the FDV0 tables for width==4 (when we can load it in one go), or if we don't have any unaligned loads
#if GMX_SIMD_REAL_WIDTH == 4 || !GMX_SIMD_HAVE_GATHER_LOADU_BYSIMDINT_TRANSPOSE_REAL
-#define TAB_FDV0
+# define TAB_FDV0
#endif
#ifdef UNROLLJ
/* Add energy register to possibly multiple terms in the energy array */
-static inline void add_ener_grp(gmx::SimdReal e_S, real *v, const int *offset_jj)
+static inline void add_ener_grp(gmx::SimdReal e_S, real* v, const int* offset_jj)
{
using namespace gmx;
int jj;
* the rapidly increases number of combinations of energy groups.
* We add to a temporary buffer for 1 i-group vs 2 j-groups.
*/
- for (jj = 0; jj < (UNROLLJ/2); jj++)
+ for (jj = 0; jj < (UNROLLJ / 2); jj++)
{
SimdReal v_S;
- v_S = load<SimdReal>(v+offset_jj[jj]+jj*GMX_SIMD_REAL_WIDTH);
- store(v+offset_jj[jj]+jj*GMX_SIMD_REAL_WIDTH, v_S + e_S);
+ v_S = load<SimdReal>(v + offset_jj[jj] + jj * GMX_SIMD_REAL_WIDTH);
+ store(v + offset_jj[jj] + jj * GMX_SIMD_REAL_WIDTH, v_S + e_S);
}
}
#endif
#if GMX_SIMD_HAVE_INT32_LOGICAL
-typedef gmx::SimdInt32 SimdBitMask;
+typedef gmx::SimdInt32 SimdBitMask;
#else
-typedef gmx::SimdReal SimdBitMask;
+typedef gmx::SimdReal SimdBitMask;
#endif
-static inline void gmx_simdcall
-gmx_load_simd_4xn_interactions(int excl,
- SimdBitMask gmx_unused filter_S0,
- SimdBitMask gmx_unused filter_S1,
- SimdBitMask gmx_unused filter_S2,
- SimdBitMask gmx_unused filter_S3,
- const real gmx_unused *simd_interaction_array,
- gmx::SimdBool *interact_S0,
- gmx::SimdBool *interact_S1,
- gmx::SimdBool *interact_S2,
- gmx::SimdBool *interact_S3)
+static inline void gmx_simdcall gmx_load_simd_4xn_interactions(int excl,
+ SimdBitMask gmx_unused filter_S0,
+ SimdBitMask gmx_unused filter_S1,
+ SimdBitMask gmx_unused filter_S2,
+ SimdBitMask gmx_unused filter_S3,
+ const real gmx_unused* simd_interaction_array,
+ gmx::SimdBool* interact_S0,
+ gmx::SimdBool* interact_S1,
+ gmx::SimdBool* interact_S2,
+ gmx::SimdBool* interact_S3)
{
using namespace gmx;
#if GMX_SIMD_HAVE_INT32_LOGICAL
/* Load integer interaction mask */
SimdInt32 mask_pr_S(excl);
- *interact_S0 = cvtIB2B(testBits( mask_pr_S & filter_S0 ));
- *interact_S1 = cvtIB2B(testBits( mask_pr_S & filter_S1 ));
- *interact_S2 = cvtIB2B(testBits( mask_pr_S & filter_S2 ));
- *interact_S3 = cvtIB2B(testBits( mask_pr_S & filter_S3 ));
+ *interact_S0 = cvtIB2B(testBits(mask_pr_S & filter_S0));
+ *interact_S1 = cvtIB2B(testBits(mask_pr_S & filter_S1));
+ *interact_S2 = cvtIB2B(testBits(mask_pr_S & filter_S2));
+ *interact_S3 = cvtIB2B(testBits(mask_pr_S & filter_S3));
#elif GMX_SIMD_HAVE_LOGICAL
- union
- {
-#if GMX_DOUBLE
+ union {
+# if GMX_DOUBLE
std::int64_t i;
-#else
+# else
std::int32_t i;
-#endif
+# endif
real r;
} conv;
conv.i = excl;
- SimdReal mask_pr_S(conv.r);
+ SimdReal mask_pr_S(conv.r);
- *interact_S0 = testBits( mask_pr_S & filter_S0 );
- *interact_S1 = testBits( mask_pr_S & filter_S1 );
- *interact_S2 = testBits( mask_pr_S & filter_S2 );
- *interact_S3 = testBits( mask_pr_S & filter_S3 );
+ *interact_S0 = testBits(mask_pr_S & filter_S0);
+ *interact_S1 = testBits(mask_pr_S & filter_S1);
+ *interact_S2 = testBits(mask_pr_S & filter_S2);
+ *interact_S3 = testBits(mask_pr_S & filter_S3);
#else
// Neither real or integer bitwise logical operations supported.
// Load masks from memory instead.
- SimdReal zero = setZero();
- *interact_S0 = ( zero < load<SimdReal>( simd_interaction_array + GMX_SIMD_REAL_WIDTH*((excl >> (0 * UNROLLJ)) & 0xF) ) );
- *interact_S1 = ( zero < load<SimdReal>( simd_interaction_array + GMX_SIMD_REAL_WIDTH*((excl >> (1 * UNROLLJ)) & 0xF) ) );
- *interact_S2 = ( zero < load<SimdReal>( simd_interaction_array + GMX_SIMD_REAL_WIDTH*((excl >> (2 * UNROLLJ)) & 0xF) ) );
- *interact_S3 = ( zero < load<SimdReal>( simd_interaction_array + GMX_SIMD_REAL_WIDTH*((excl >> (3 * UNROLLJ)) & 0xF) ) );
+ SimdReal zero = setZero();
+ *interact_S0 = (zero < load<SimdReal>(simd_interaction_array
+ + GMX_SIMD_REAL_WIDTH * ((excl >> (0 * UNROLLJ)) & 0xF)));
+ *interact_S1 = (zero < load<SimdReal>(simd_interaction_array
+ + GMX_SIMD_REAL_WIDTH * ((excl >> (1 * UNROLLJ)) & 0xF)));
+ *interact_S2 = (zero < load<SimdReal>(simd_interaction_array
+ + GMX_SIMD_REAL_WIDTH * ((excl >> (2 * UNROLLJ)) & 0xF)));
+ *interact_S3 = (zero < load<SimdReal>(simd_interaction_array
+ + GMX_SIMD_REAL_WIDTH * ((excl >> (3 * UNROLLJ)) & 0xF)));
#endif
}
biod.pnpi.spb.ru / alexxy / gromacs.git / blobdiff