portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-04-25 13:15:55 +01:00
Code Releases Activity

Nsimd() and coalesced support

Browse Source
This commit is contained in:
Peter Boyle 2019-05-25 12:44:07 +01:00
parent d8098f1ecd
commit ffde81f22a
2 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Grid/tensors/Tensor_class.h
View File

@ -64,6 +64,8 @@ public:
enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1 }; enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1 };
static constexpr accelerator_inline int Nsimd(void) { return sizeof(vector_type)/sizeof(scalar_type); }
// Scalar no action // Scalar no action
accelerator iScalar() = default; accelerator iScalar() = default;
@ -195,6 +197,8 @@ public:
// get double precision version // get double precision version
typedef iVector<typename GridTypeMapper<vtype>::DoublePrecision, N> DoublePrecision; typedef iVector<typename GridTypeMapper<vtype>::DoublePrecision, N> DoublePrecision;
static constexpr accelerator_inline int Nsimd(void) { return sizeof(vector_type)/sizeof(scalar_type); }
template <class T, typename std::enable_if<!isGridTensor<T>::value, T>::type * = nullptr> template <class T, typename std::enable_if<!isGridTensor<T>::value, T>::type * = nullptr>
accelerator_inline auto operator=(T arg) -> iVector<vtype, N> { accelerator_inline auto operator=(T arg) -> iVector<vtype, N> {
zeroit(*this); zeroit(*this);
@ -304,6 +308,8 @@ public:
enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1 }; enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1 };
static constexpr accelerator_inline int Nsimd(void) { return sizeof(vector_type)/sizeof(scalar_type); }
accelerator_inline iMatrix(const Zero &z) { zeroit(*this); }; accelerator_inline iMatrix(const Zero &z) { zeroit(*this); };
accelerator iMatrix() = default; accelerator iMatrix() = default;

3
Grid/tensors/Tensors.h
View File

@ -48,4 +48,7 @@ Author: neo <cossu@post.kek.jp>
#include <Grid/tensors/Tensor_extract_merge.h> #include <Grid/tensors/Tensor_extract_merge.h>
#include <Grid/tensors/Tensor_logical.h> #include <Grid/tensors/Tensor_logical.h>
// Maybe expect different thread on each SIMD lane
#include <Grid/tensors/Tensor_SIMT.h>
#endif #endif