Skip to content

GitLab

elpa
elpa
Commit c9455ca8 authored by Andreas Marek's avatar Andreas Marek
Browse files
Options

Template for prepare_matrix

parent c0d57a28
Hide whitespace changes
Inline Side-by-side
Showing with 192 additions and 362 deletions
Makefile.am
View file @ c9455ca8
......@@ -490,6 +490,7 @@ libelpatest@SUFFIX@_la_SOURCES = \
test/shared/setup_mpi.F90 \
test/shared/blacs_infrastructure.F90 \
test/shared/prepare_matrix.F90 \
test/shared/prepare_matrix_template.X90 \
test/shared/mod_output_types.F90 \
test/shared/mod_from_c.F90 \
test/shared/call_elpa1.c
......@@ -1018,6 +1019,7 @@ EXTRA_DIST = \
src/elpa2/GPU/ev_tridi_band_gpu_c_v2_complex_template.Xcu \
src/elpa2/GPU/ev_tridi_band_gpu_c_v2_real_template.Xcu \
src/GPU/cuUtils_template.Xcu \
test/shared/prepare_matrix_template.X90\
nvcc_wrap \
test_project/Makefile.am \
test_project/autogen.sh \
......
generated_headers.am
View file @ c9455ca8
......@@ -16,7 +16,7 @@ config-f90.h: config.h
elpa/elpa_generated.h: $(top_srcdir)/src/elpa_c_interface.F90 | elpa
$(call extract_interface,!c>)
test/shared/generated.h: $(wildcard $(top_srcdir)/test/shared/*.F90) | test/shared
test/shared/generated.h: $(wildcard $(top_srcdir)/test/shared/*.*90) | test/shared
$(call extract_interface,!c>)
elpa/elpa_generated_fortran_interfaces.h: $(wildcard $(top_srcdir)/src/elpa2/kernels/*.c) $(wildcard $(top_srcdir)/src/elpa2/kernels/*.s) | elpa
......
test/shared/prepare_matrix.F90
View file @ c9455ca8
......@@ -64,380 +64,43 @@ module mod_prepare_matrix
#endif
contains
#define DOUBLE_PRECISION_COMPLEX 1
subroutine prepare_matrix_complex_double(na, myid, sc_desc, a, z, as)
#define COMPLEXCASE 1
#define DOUBLE_PRECISION 1
#include "../../src/precision_macros.h"
#include "prepare_matrix_template.X90"
#undef DOUBLE_PRECISION
#undef COMPLEXCASE
use precision
implicit none
integer(kind=ik), intent(in) :: myid, na, sc_desc(:)
complex(kind=ck8), intent(inout) :: z(:,:), a(:,:), as(:,:)
complex(kind=ck8), parameter :: CONE = (1.0_rk8, 0.0_rk8)
real(kind=rk8) :: xr(size(a,dim=1), size(a,dim=2))
integer, allocatable :: iseed(:)
integer :: n
! for getting a hermitian test matrix A we get a random matrix Z
! and calculate A = Z + Z**H
! we want different random numbers on every process
! (otherwise A might get rank deficient):
call random_seed(size=n)
allocate(iseed(n))
iseed(:) = myid
call random_seed(put=iseed)
call random_number(xr)
z(:,:) = xr(:,:)
call RANDOM_NUMBER(xr)
z(:,:) = z(:,:) + (0.0_rk8,1.0_rk8)*xr(:,:)
a(:,:) = z(:,:)
if (myid == 0) then
print '(a)','| Random matrix block has been set up. (only processor 0 confirms this step)'
endif
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_COMPLEX
call pztranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#else
call pctranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#endif
#else /* WITH_MPI */
a = a + transpose(conjg(z))
#endif /* WITH_MPI */
if (myid == 0) then
print '(a)','| Random matrix block has been symmetrized'
endif
! save original matrix A for later accuracy checks
as = a
deallocate(iseed)
end subroutine
#ifdef DOUBLE_PRECISION_COMPLEX
!c> void prepare_matrix_complex_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex double *a, complex double *z, complex double *as);
#else
!c> void prepare_matrix_complex_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex *a, complex *z, complex *as);
#endif
#ifdef DOUBLE_PRECISION_COMPLEX
subroutine prepare_matrix_complex_wrapper_double&
#else
subroutine prepare_matrix_complex_wrapper_single&
#endif
(na, myid, na_rows, na_cols, sc_desc, a, z, as) &
#ifdef DOUBLE_PRECISION_COMPLEX
bind(C, name="prepare_matrix_complex_double_f")
#else
bind(C, name="prepare_matrix_complex_single_f")
#endif
use iso_c_binding
implicit none
integer(kind=c_int) , value :: myid, na, na_rows, na_cols
integer(kind=c_int) :: sc_desc(1:9)
#ifdef DOUBLE_PRECISION_COMPLEX
complex(kind=c_double) :: &
#else
complex(kind=c_float) :: &
#endif
z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
call prepare_matrix_complex_double(na, myid, sc_desc, a, z, as)
end subroutine
#ifdef WANT_SINGLE_PRECISION_COMPLEX
#undef DOUBLE_PRECISION_COMPLEX
subroutine prepare_matrix_complex_single(na, myid, sc_desc, a, z, as)
use precision
implicit none
integer(kind=ik), intent(in) :: myid, na, sc_desc(:)
complex(kind=ck4), intent(inout) :: z(:,:), a(:,:), as(:,:)
complex(kind=ck4), parameter :: CONE = (1.0_rk4, 0.0_rk4)
real(kind=rk4) :: xr(size(a,dim=1), size(a,dim=2))
integer, allocatable :: iseed(:)
integer :: n
! for getting a hermitian test matrix A we get a random matrix Z
! and calculate A = Z + Z**H
! we want different random numbers on every process
! (otherwise A might get rank deficient):
call random_seed(size=n)
allocate(iseed(n))
iseed(:) = myid
call random_seed(put=iseed)
call random_number(xr)
z(:,:) = xr(:,:)
call random_number(xr)
z(:,:) = z(:,:) + (0.0_rk4,1.0_rk4)*xr(:,:)
a(:,:) = z(:,:)
if (myid == 0) then
print '(a)','| Random matrix block has been set up. (only processor 0 confirms this step)'
endif
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_COMPLEX
call pztranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#else
call pctranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#endif
#else /* WITH_MPI */
a = a + transpose(conjg(z))
#endif /* WITH_MPI */
if (myid == 0) then
print '(a)','| Random matrix block has been symmetrized'
endif
! save original matrix A for later accuracy checks
as = a
deallocate(iseed)
end subroutine
#ifdef DOUBLE_PRECISION_COMPLEX
!c> void prepare_matrix_complex_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex double *a, complex double *z, complex double *as);
#else
!c> void prepare_matrix_complex_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex *a, complex *z, complex *as);
#endif
subroutine prepare_matrix_complex_wrapper_single(na, myid, na_rows, na_cols, sc_desc, a, z, as) &
#ifdef DOUBLE_PRECISION_COMPLEX
bind(C, name="prepare_matrix_complex_double_f")
#else
bind(C, name="prepare_matrix_complex_single_f")
#endif
use iso_c_binding
implicit none
integer(kind=c_int) , value :: myid, na, na_rows, na_cols
integer(kind=c_int) :: sc_desc(1:9)
#ifdef DOUBLE_PRECISION_COMPLEX
complex(kind=c_double) :: &
#else
complex(kind=c_float) :: &
#endif
z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
call prepare_matrix_complex_single(na, myid, sc_desc, a, z, as)
end subroutine
#define COMPLEXCASE 1
#define SINGLE_PRECISION 1
#include "../../src/precision_macros.h"
#include "prepare_matrix_template.X90"
#undef SINGLE_PRECISION
#undef COMPLEXCASE
#endif /* WANT_SINGLE_PRECISION_COMPLEX */
#define DOUBLE_PRECISION_REAL 1
subroutine prepare_matrix_real_double(na, myid, sc_desc, a, z, as)
use precision
implicit none
integer(kind=ik), intent(in) :: myid, na, sc_desc(:)
real(kind=rk8), intent(inout) :: z(:,:), a(:,:), as(:,:)
integer, allocatable :: iseed(:)
integer :: n
! for getting a hermitian test matrix A we get a random matrix Z
! and calculate A = Z + Z**H
! we want different random numbers on every process
! (otherwise A might get rank deficient):
call random_seed(size=n)
allocate(iseed(n))
iseed(:) = myid
call random_seed(put=iseed)
call random_number(z)
a(:,:) = z(:,:)
if (myid == 0) then
print '(a)','| Random matrix block has been set up. (only processor 0 confirms this step)'
endif
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_REAL
call pdtran(na, na, 1.0_rk8, z, 1, 1, sc_desc, 1.0_rk8, a, 1, 1, sc_desc) ! A = A + Z**T
#else
call pstran(na, na, 1.0_rk4, z, 1, 1, sc_desc, 1.0_rk4, a, 1, 1, sc_desc) ! A = A + Z**T
#endif
#else /* WITH_MPI */
a = a + transpose(z)
#endif /* WITH_MPI */
if (myid == 0) then
print '(a)','| Random matrix block has been symmetrized'
endif
! save original matrix A for later accuracy checks
as = a
deallocate(iseed)
end subroutine
#ifdef DOUBLE_PRECISION_REAL
!c> void prepare_matrix_real_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> double *a, double *z, double *as);
#else
!c> void prepare_matrix_real_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> float *a, float *z, float *as);
#endif
#ifdef DOUBLE_PRECISION_REAL
subroutine prepare_matrix_real_wrapper_double&
#else
subroutine prepare_matrix_real_wrapper_single&
#endif
(na, myid, na_rows, na_cols, sc_desc, a, z, as) &
#ifdef DOUBLE_PRECISION_REAL
bind(C, name="prepare_matrix_real_double_f")
#else
bind(C, name="prepare_matrix_real_single_f")
#endif
use iso_c_binding
implicit none
integer(kind=c_int) , value :: myid, na, na_rows, na_cols
integer(kind=c_int) :: sc_desc(1:9)
#ifdef DOUBLE_PRECISION_REAL
real(kind=c_double) :: z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#else
real(kind=c_float) :: z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#endif
call prepare_matrix_real_double(na, myid, sc_desc, a, z, as)
end subroutine
#define REALCASE 1
#define DOUBLE_PRECISION 1
#include "../../src/precision_macros.h"
#include "prepare_matrix_template.X90"
#undef DOUBLE_PRECISION
#undef REALCASE
#ifdef WANT_SINGLE_PRECISION_REAL
#undef DOUBLE_PRECISION_REAL
subroutine prepare_matrix_real_single(na, myid, sc_desc, a, z, as)
use precision
implicit none
integer(kind=ik), intent(in) :: myid, na, sc_desc(:)
real(kind=rk4), intent(inout) :: z(:,:), a(:,:), as(:,:)
integer, allocatable :: iseed(:)
integer :: n
! for getting a hermitian test matrix A we get a random matrix Z
! and calculate A = Z + Z**H
! we want different random numbers on every process
! (otherwise A might get rank deficient):
call random_seed(size=n)
allocate(iseed(n))
iseed(:) = myid
call random_seed(put=iseed)
call random_number(z)
a(:,:) = z(:,:)
if (myid == 0) then
print '(a)','| Random matrix block has been set up. (only processor 0 confirms this step)'
endif
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_REAL
call pdtran(na, na, 1.0_rk8, z, 1, 1, sc_desc, 1.0_rk8, a, 1, 1, sc_desc) ! A = A + Z**T
#else
call pstran(na, na, 1.0_rk4, z, 1, 1, sc_desc, 1.0_rk4, a, 1, 1, sc_desc) ! A = A + Z**T
#endif
#else /* WITH_MPI */
a = a + transpose(z)
#endif /* WITH_MPI */
if (myid == 0) then
print '(a)','| Random matrix block has been symmetrized'
endif
! save original matrix A for later accuracy checks
as = a
deallocate(iseed)
end subroutine
#ifdef DOUBLE_PRECISION_REAL
!c> void prepare_matrix_real_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> double *a, double *z, double *as);
#else
!c> void prepare_matrix_real_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> float *a, float *z, float *as);
#endif
#ifdef DOUBLE_PRECISION_REAL
subroutine prepare_matrix_real_wrapper_double&
#else
subroutine prepare_matrix_real_wrapper_single&
#endif
(na, myid, na_rows, na_cols, sc_desc, a, z, as) &
#ifdef DOUBLE_PRECISION_REAL
bind(C, name="prepare_matrix_real_double_f")
#else
bind(C, name="prepare_matrix_real_single_f")
#endif
use iso_c_binding
implicit none
integer(kind=c_int) , value :: myid, na, na_rows, na_cols
integer(kind=c_int) :: sc_desc(1:9)
#ifdef DOUBLE_PRECISION_REAL
real(kind=c_double) :: z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#else
real(kind=c_float) :: z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#endif
call prepare_matrix_real_single(na, myid, sc_desc, a, z, as)
end subroutine
#define REALCASE 1
#define SINGLE_PRECISION 1
#include "../../src/precision_macros.h"
#include "prepare_matrix_template.X90"
#undef SINGLE_PRECISION
#undef REALCASE
#endif /* WANT_SINGLE_PRECISION_REAL */
......
test/shared/prepare_matrix_template.X90 0 → 100644
View file @ c9455ca8
subroutine prepare_matrix_&
&MATH_DATATYPE&
&_&
&PRECISION&
& (na, myid, sc_desc, a, z, as)
use precision
implicit none
integer(kind=ik), intent(in) :: myid, na, sc_desc(:)
#if REALCASE == 1
real(kind=C_DATATYPE_KIND), intent(inout) :: z(:,:), a(:,:), as(:,:)
#endif /* REALCASE */
#if COMPLEXCASE == 1
complex(kind=C_DATATYPE_KIND), intent(inout) :: z(:,:), a(:,:), as(:,:)
#ifdef DOUBLE_PRECISION_COMPLEX
complex(kind=C_DATATYPE_KIND), parameter :: CONE = (1.0_rk8, 0.0_rk8)
#else
complex(kind=C_DATATYPE_KIND), parameter :: CONE = (1.0_rk4, 0.0_rk4)
#endif
real(kind=C_DATATYPE_KIND) :: xr(size(a,dim=1), size(a,dim=2))
#endif /* COMPLEXCASE */
integer, allocatable :: iseed(:)
integer :: n
! for getting a hermitian test matrix A we get a random matrix Z
! and calculate A = Z + Z**H
! we want different random numbers on every process
! (otherwise A might get rank deficient):
call random_seed(size=n)
allocate(iseed(n))
iseed(:) = myid
call random_seed(put=iseed)
#if REALCASE == 1
call random_number(z)
a(:,:) = z(:,:)
#endif /* REALCASE */
#if COMPLEXCASE == 1
call random_number(xr)
z(:,:) = xr(:,:)
call RANDOM_NUMBER(xr)
#ifdef DOUBLE_PRECISION_COMPLEX
z(:,:) = z(:,:) + (0.0_rk8,1.0_rk8)*xr(:,:)
#else
z(:,:) = z(:,:) + (0.0_rk4,1.0_rk4)*xr(:,:)
#endif
a(:,:) = z(:,:)
#endif /* COMPLEXCASE */
if (myid == 0) then
print '(a)','| Random matrix block has been set up. (only processor 0 confirms this step)'
endif
#if REALCASE == 1
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_REAL
call pdtran(na, na, 1.0_rk8, z, 1, 1, sc_desc, 1.0_rk8, a, 1, 1, sc_desc) ! A = A + Z**T
#else
call pstran(na, na, 1.0_rk4, z, 1, 1, sc_desc, 1.0_rk4, a, 1, 1, sc_desc) ! A = A + Z**T
#endif
#else /* WITH_MPI */
a = a + transpose(z)
#endif /* WITH_MPI */
#endif /* REALCASE */
#if COMPLEXCASE == 1
#ifdef WITH_MPI
#ifdef DOUBLE_PRECISION_COMPLEX
call pztranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#else
call pctranc(na, na, CONE, z, 1, 1, sc_desc, CONE, a, 1, 1, sc_desc) ! A = A + Z**H
#endif
#else /* WITH_MPI */
a = a + transpose(conjg(z))
#endif /* WITH_MPI */
#endif /* COMPLEXCASE */
if (myid == 0) then
print '(a)','| Random matrix block has been symmetrized'
endif
! save original matrix A for later accuracy checks
as = a
deallocate(iseed)
end subroutine
#if REALCASE == 1
#ifdef DOUBLE_PRECISION_REAL
!c> void prepare_matrix_real_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> double *a, double *z, double *as);
#else
!c> void prepare_matrix_real_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> float *a, float *z, float *as);
#endif
#endif /* REALCASE */
#if COMPLEXCASE == 1
#ifdef DOUBLE_PRECISION_COMPLEX
!c> void prepare_matrix_complex_double_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex double *a, complex double *z, complex double *as);
#else
!c> void prepare_matrix_complex_single_f(int na, int myid, int na_rows, int na_cols,
!c> int sc_desc[9],
!c> complex *a, complex *z, complex *as);
#endif
#endif /* COMPLEXCASE */
subroutine prepare_matrix_&
&MATH_DATATYPE&
&_wrapper_&
&PRECISION&
& (na, myid, na_rows, na_cols, sc_desc, a, z, as) &
bind(C, name="prepare_matrix_&
&MATH_DATATYPE&
&_&
&PRECISION&
&_f")
use iso_c_binding
implicit none
integer(kind=c_int) , value :: myid, na, na_rows, na_cols
integer(kind=c_int) :: sc_desc(1:9)
#if REALCASE == 1
real(kind=C_DATATYPE_KIND) :: z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#endif
#if COMPLEXCASE == 1
complex(kind=C_DATATYPE_KIND) :: &
z(1:na_rows,1:na_cols), a(1:na_rows,1:na_cols), &
as(1:na_rows,1:na_cols)
#endif
call prepare_matrix_&
&MATH_DATATYPE&
&_&
&PRECISION&
& (na, myid, sc_desc, a, z, as)
end subroutine
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment