Skip to content

Commit ade4b4f

Browse files
ErjieWuErjieWu
authored
Merge branch 'develop' into refactor
2 parents 8ded8d7 + 8fc7c42 commit ade4b4f

File tree

455 files changed

+1460
-2106
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

455 files changed

+1460
-2106
lines changed

‎docs/advanced/input_files/input-main.md‎

Lines changed: 4 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -52,7 +52,6 @@
5252
- [lcao\_dr](#lcao_dr)
5353
- [lcao\_rmax](#lcao_rmax)
5454
- [search\_radius](#search_radius)
55-
- [search\_pbc](#search_pbc)
5655
- [bx, by, bz](#bx-by-bz)
5756
- [elpa\_num\_thread](#elpa_num_thread)
5857
- [num\_stream](#num_stream)
@@ -924,12 +923,6 @@ These variables are used to control the numerical atomic orbitals related parame
924923
- **Default**: -1
925924
- **Unit**: Bohr
926925

927-
### search_pbc
928-
929-
- **Type**: Boolean
930-
- **Description**: If True, periodic images will be included in searching for the neighbouring atoms. If False, periodic images will be ignored.
931-
- **Default**: True
932-
933926
### bx, by, bz
934927

935928
- **Type**: Integer
@@ -3586,19 +3579,23 @@ These variables are used to control berry phase and wannier90 interface paramete
35863579
- **Type**: Real
35873580
- **Description**:
35883581
`td_lcut1` is the lower bound of the interval in the length gauge RT-TDDFT, where $x$ is the fractional coordinate:
3582+
35893583
$$
35903584
E(x)=\begin{cases}E_0, & \mathtt{cut1}\leqslant x \leqslant \mathtt{cut2} \\-E_0\left(\dfrac{1}{\mathtt{cut1}+1-\mathtt{cut2}}-1\right), & 0 < x < \mathtt{cut1~~or~~cut2} < x < 1 \end{cases}
35913585
$$
3586+
35923587
- **Default**: 0.05
35933588

35943589
### td_lcut2
35953590

35963591
- **Type**: Real
35973592
- **Description**:
35983593
`td_lcut2` is the upper bound of the interval in the length gauge RT-TDDFT, where $x$ is the fractional coordinate:
3594+
35993595
$$
36003596
E(x)=\begin{cases}E_0, & \mathtt{cut1}\leqslant x \leqslant \mathtt{cut2} \\-E_0\left(\dfrac{1}{\mathtt{cut1}+1-\mathtt{cut2}}-1\right), & 0 < x < \mathtt{cut1~~or~~cut2} < x < 1 \end{cases}
36013597
$$
3598+
36023599
- **Default**: 0.95
36033600

36043601
### td_gauss_freq

‎source/module_base/module_device/memory_op.h‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -133,6 +133,7 @@ struct synchronize_memory_op<FPTYPE, base_device::DEVICE_GPU, base_device::DEVIC
133133
void operator()(FPTYPE* arr_out,
134134
const FPTYPE* arr_in,
135135
const size_t size);
136+
136137
};
137138

138139
template <typename FPTYPE>

‎source/module_base/tool_quit.cpp‎

Lines changed: 14 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -62,6 +62,20 @@ void QUIT(int ret)
6262
void WARNING_QUIT(const std::string &file,const std::string &description)
6363
{
6464
WARNING_QUIT(file, description, 1);
65+
66+
#ifdef __MPI /* if it is MPI run, finalize first, then exit */
67+
std::cout << "Detecting if MPI has been initialized..." << std::endl;
68+
int is_initialized;
69+
MPI_Initialized(&is_initialized);
70+
if (is_initialized) {
71+
std::cout << "Terminating ABACUS with multiprocessing environment." << std::endl;
72+
MPI_Finalize();
73+
}
74+
else{
75+
std::cout << "MPI has not been initialized. Quit normally." << std::endl;
76+
}
77+
/* but seems this is the only correct way to terminate the MPI */
78+
#endif
6579
}
6680

6781
void WARNING_QUIT(const std::string &file,const std::string &description,int ret)

‎source/module_basis/module_pw/CMakeLists.txt‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -58,5 +58,6 @@ if(BUILD_TESTING)
5858
add_subdirectory(test)
5959
add_subdirectory(test_serial)
6060
add_subdirectory(kernels/test)
61+
add_subdirectory(test_gpu)
6162
endif()
6263
endif()

‎source/module_basis/module_pw/module_fft/fft_cuda.cpp‎

Lines changed: 26 additions & 26 deletions
Original file line numberDiff line numberDiff line change
@@ -1,13 +1,12 @@
11
#include "fft_cuda.h"
2+
23
#include "module_base/module_device/memory_op.h"
34
#include "module_hamilt_pw/hamilt_pwdft/global.h"
45

56
namespace ModulePW
67
{
78
template <typename FPTYPE>
8-
void FFT_CUDA<FPTYPE>::initfft(int nx_in,
9-
int ny_in,
10-
int nz_in)
9+
void FFT_CUDA<FPTYPE>::initfft(int nx_in, int ny_in, int nz_in)
1110
{
1211
this->nx = nx_in;
1312
this->ny = ny_in;
@@ -18,9 +17,8 @@ void FFT_CUDA<float>::setupFFT()
1817
{
1918
cufftPlan3d(&c_handle, this->nx, this->ny, this->nz, CUFFT_C2C);
2019
resmem_cd_op()(this->c_auxr_3d, this->nx * this->ny * this->nz);
21-
2220
}
23-
template <>
21+
template <>
2422
void FFT_CUDA<double>::setupFFT()
2523
{
2624
cufftPlan3d(&z_handle, this->nx, this->ny, this->nz, CUFFT_Z2Z);
@@ -66,49 +64,51 @@ void FFT_CUDA<double>::clear()
6664
}
6765

6866
template <>
69-
void FFT_CUDA<float>::fft3D_forward(std::complex<float>* in,
70-
std::complex<float>* out) const
67+
void FFT_CUDA<float>::fft3D_forward(std::complex<float>* in, std::complex<float>* out) const
7168
{
72-
CHECK_CUFFT(cufftExecC2C(this->c_handle,
73-
reinterpret_cast<cufftComplex*>(in),
69+
CHECK_CUFFT(cufftExecC2C(this->c_handle,
70+
reinterpret_cast<cufftComplex*>(in),
7471
reinterpret_cast<cufftComplex*>(out),
7572
CUFFT_FORWARD));
7673
}
7774
template <>
78-
void FFT_CUDA<double>::fft3D_forward(std::complex<double>* in,
79-
std::complex<double>* out) const
75+
void FFT_CUDA<double>::fft3D_forward(std::complex<double>* in, std::complex<double>* out) const
8076
{
81-
CHECK_CUFFT(cufftExecZ2Z(this->z_handle,
77+
CHECK_CUFFT(cufftExecZ2Z(this->z_handle,
8278
reinterpret_cast<cufftDoubleComplex*>(in),
83-
reinterpret_cast<cufftDoubleComplex*>(out),
79+
reinterpret_cast<cufftDoubleComplex*>(out),
8480
CUFFT_FORWARD));
8581
}
8682
template <>
87-
void FFT_CUDA<float>::fft3D_backward(std::complex<float>* in,
88-
std::complex<float>* out) const
83+
void FFT_CUDA<float>::fft3D_backward(std::complex<float>* in, std::complex<float>* out) const
8984
{
90-
CHECK_CUFFT(cufftExecC2C(this->c_handle,
91-
reinterpret_cast<cufftComplex*>(in),
85+
CHECK_CUFFT(cufftExecC2C(this->c_handle,
86+
reinterpret_cast<cufftComplex*>(in),
9287
reinterpret_cast<cufftComplex*>(out),
9388
CUFFT_INVERSE));
9489
}
9590

9691
template <>
97-
void FFT_CUDA<double>::fft3D_backward(std::complex<double>* in,
98-
std::complex<double>* out) const
92+
void FFT_CUDA<double>::fft3D_backward(std::complex<double>* in, std::complex<double>* out) const
9993
{
100-
CHECK_CUFFT(cufftExecZ2Z(this->z_handle,
94+
CHECK_CUFFT(cufftExecZ2Z(this->z_handle,
10195
reinterpret_cast<cufftDoubleComplex*>(in),
102-
reinterpret_cast<cufftDoubleComplex*>(out),
96+
reinterpret_cast<cufftDoubleComplex*>(out),
10397
CUFFT_INVERSE));
10498
}
105-
template <> std::complex<float>*
106-
FFT_CUDA<float>::get_auxr_3d_data() const {return this->c_auxr_3d;}
107-
template <> std::complex<double>*
108-
FFT_CUDA<double>::get_auxr_3d_data() const {return this->z_auxr_3d;}
99+
template <>
100+
std::complex<float>* FFT_CUDA<float>::get_auxr_3d_data() const
101+
{
102+
return this->c_auxr_3d;
103+
}
104+
template <>
105+
std::complex<double>* FFT_CUDA<double>::get_auxr_3d_data() const
106+
{
107+
return this->z_auxr_3d;
108+
}
109109

110110
template FFT_CUDA<float>::FFT_CUDA();
111111
template FFT_CUDA<float>::~FFT_CUDA();
112112
template FFT_CUDA<double>::FFT_CUDA();
113113
template FFT_CUDA<double>::~FFT_CUDA();
114-
}// namespace ModulePW
114+
} // namespace ModulePW

‎source/module_basis/module_pw/pw_basis.cpp‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -43,6 +43,7 @@ PW_Basis:: ~PW_Basis()
4343
if (this->device == "gpu")
4444
{
4545
delmem_int_op()(this->d_is2fftixy);
46+
delmem_int_op()(this->ig2ixyz_gpu);
4647
}
4748
#endif
4849
}
@@ -59,6 +60,7 @@ void PW_Basis::setuptransform()
5960
this->distribute_g();
6061
this->getstartgr();
6162
this->fft_bundle.clear();
63+
6264
if(this->xprime)
6365
{
6466
this->fft_bundle.initfft(this->nx,this->ny,this->nz,this->lix,this->rix,this->nst,this->nplane,this->poolnproc,this->gamma_only, this->xprime);

‎source/module_basis/module_pw/pw_basis.h‎

Lines changed: 13 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -100,10 +100,11 @@ class PW_Basis
100100
//===============================================
101101
public:
102102
#ifdef __MPI
103-
MPI_Comm pool_world;
103+
MPI_Comm pool_world=MPI_COMM_NULL;
104104
#endif
105105

106106
int *ig2isz=nullptr; // map ig to (is, iz).
107+
int *ig2ixyz_gpu = nullptr;
107108
int *istot2ixy=nullptr; // istot2ixy[is]: iy + ix * ny of is^th stick among all sticks.
108109
int *is2fftixy=nullptr, * d_is2fftixy = nullptr; // is2fftixy[is]: iy + ix * ny of is^th stick among sticks on current proc.
109110
int *fftixy2ip=nullptr; // fftixy2ip[iy + ix * fftny]: ip of proc which contains stick on (ix, iy). if no stick: -1
@@ -352,7 +353,10 @@ class PW_Basis
352353
void recip_to_real(TK* in,
353354
TR* out,
354355
const bool add = false,
355-
const typename GetTypeReal<TK>::type factor = 1.0) const;
356+
const typename GetTypeReal<TK>::type factor = 1.0) const
357+
{
358+
this->recip2real_gpu(in,out,add,factor);
359+
};
356360

357361
// template <typename FPTYPE,
358362
// typename Device,
@@ -383,9 +387,7 @@ class PW_Basis
383387
* values in the output array.
384388
* @param factor Optional scaling factor, default value 1.0, applied to the output values.
385389
*/
386-
template <typename TK,
387-
typename TR,
388-
typename Device,
390+
template <typename TR,typename TK,typename Device,
389391
typename std::enable_if<!std::is_same<TK, typename GetTypeReal<TK>::type>::value
390392
&& (std::is_same<TR, typename GetTypeReal<TK>::type>::value || std::is_same<TR, TK>::value)
391393
&& std::is_same<Device, base_device::DEVICE_CPU>::value ,int>::type = 0>
@@ -397,14 +399,17 @@ class PW_Basis
397399
this->real2recip(in, out, add, factor);
398400
}
399401

400-
template <typename TK,typename TR, typename Device,
402+
template <typename TR, typename TK, typename Device,
401403
typename std::enable_if<!std::is_same<TK, typename GetTypeReal<TK>::type>::value
402404
&& (std::is_same<TR, typename GetTypeReal<TK>::type>::value || std::is_same<TR, TK>::value)
403-
&& !std::is_same<Device, base_device::DEVICE_CPU>::value ,int>::type = 0>
405+
&& std::is_same<Device, base_device::DEVICE_GPU>::value ,int>::type = 0>
404406
void real_to_recip(TR* in,
405407
TK* out,
406408
const bool add = false,
407-
const typename GetTypeReal<TK>::type factor = 1.0) const;
409+
const typename GetTypeReal<TK>::type factor = 1.0) const
410+
{
411+
this->real2recip_gpu(in,out,add,factor);
412+
};
408413

409414
protected:
410415
//gather planes and scatter sticks of all processors

‎source/module_basis/module_pw/pw_basis_big.h‎

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -170,7 +170,6 @@ class PW_Basis_Big : public PW_Basis_Sup
170170
MPI_Allreduce(MPI_IN_PLACE, ibox, 3, MPI_INT, MPI_MAX , this->pool_world);
171171
#endif
172172

173-
174173
// Find the minimal FFT box size the factors into the primes (2,3,5,7).
175174
for (int i = 0; i < 3; i++)
176175
{

‎source/module_basis/module_pw/pw_basis_k_big.h‎

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -56,8 +56,7 @@ class PW_Basis_K_Big: public PW_Basis_K
5656
for(int ip = 0 ; ip < this->poolnproc ; ++ip)
5757
{
5858
this->numz[ip] = npbz*this->bz;
59-
if(ip < modbz) { this->numz[ip]+=this->bz;
60-
}
59+
if(ip < modbz) { this->numz[ip]+=this->bz;}
6160
if(ip < this->poolnproc - 1) this->startz[ip+1] = this->startz[ip] + numz[ip];
6261
if(ip == this->poolrank)
6362
{

‎source/module_basis/module_pw/pw_basis_sup.cpp‎

Lines changed: 13 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -114,12 +114,12 @@ void PW_Basis_Sup::distribution_method3(const ModulePW::PW_Basis* pw_rho)
114114
this->count_pw_st(st_length2D, st_bottom2D);
115115
}
116116
#ifdef __MPI
117-
MPI_Bcast(&this->npwtot, 1, MPI_INT, 0, this->pool_world);
118-
MPI_Bcast(&this->nstot, 1, MPI_INT, 0, this->pool_world);
119-
MPI_Bcast(&liy, 1, MPI_INT, 0, this->pool_world);
120-
MPI_Bcast(&riy, 1, MPI_INT, 0, this->pool_world);
121-
MPI_Bcast(&lix, 1, MPI_INT, 0, this->pool_world);
122-
MPI_Bcast(&rix, 1, MPI_INT, 0, this->pool_world);
117+
MPI_Bcast(&this->npwtot, 1, MPI_INT, 0, this->pool_world);
118+
MPI_Bcast(&this->nstot, 1, MPI_INT, 0, this->pool_world);
119+
MPI_Bcast(&liy, 1, MPI_INT, 0, this->pool_world);
120+
MPI_Bcast(&riy, 1, MPI_INT, 0, this->pool_world);
121+
MPI_Bcast(&lix, 1, MPI_INT, 0, this->pool_world);
122+
MPI_Bcast(&rix, 1, MPI_INT, 0, this->pool_world);
123123
#endif
124124
delete[] this->istot2ixy;
125125
this->istot2ixy = new int[this->nstot];
@@ -164,14 +164,14 @@ void PW_Basis_Sup::distribution_method3(const ModulePW::PW_Basis* pw_rho)
164164
}
165165
#endif
166166
}
167-
168167
#ifdef __MPI
169-
MPI_Bcast(st_length2D, this->fftnxy, MPI_INT, 0, this->pool_world);
170-
MPI_Bcast(st_bottom2D, this->fftnxy, MPI_INT, 0, this->pool_world);
171-
MPI_Bcast(this->fftixy2ip, this->fftnxy, MPI_INT, 0, this->pool_world);
172-
MPI_Bcast(this->istot2ixy, this->nstot, MPI_INT, 0, this->pool_world);
173-
MPI_Bcast(this->nst_per, this->poolnproc, MPI_INT, 0, this->pool_world);
174-
MPI_Bcast(this->npw_per, this->poolnproc, MPI_INT, 0, this->pool_world);
168+
169+
MPI_Bcast(st_length2D, this->fftnxy, MPI_INT, 0, this->pool_world);
170+
MPI_Bcast(st_bottom2D, this->fftnxy, MPI_INT, 0, this->pool_world);
171+
MPI_Bcast(this->fftixy2ip, this->fftnxy, MPI_INT, 0, this->pool_world);
172+
MPI_Bcast(this->istot2ixy, this->nstot, MPI_INT, 0, this->pool_world);
173+
MPI_Bcast(this->nst_per, this->poolnproc, MPI_INT, 0, this->pool_world);
174+
MPI_Bcast(this->npw_per, this->poolnproc, MPI_INT, 0, this->pool_world);
175175
#endif
176176
this->npw = this->npw_per[this->poolrank];
177177
this->nst = this->nst_per[this->poolrank];

0 commit comments

Comments
 (0)