Skip to content

Fix potential Data race in OpenMP #7345

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
mohanchen wants to merge 7 commits into
base: develop
Choose a base branch
from
Open

Fix potential Data race in OpenMP #7345

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
35e3e2a
fix data race
May 15, 2026
1ff5516
fix potential data race
May 15, 2026
a3b0622
fix data race in pw
May 15, 2026
5087b2d
fix potential data race
May 15, 2026
1755bfb
update format
May 15, 2026
23d40e3
fix data race in fft
May 15, 2026
76948b9
fix
May 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 22 additions & 22 deletions source/source_base/math_integral.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,7 @@ void Integral::Simpson_Integral

// simpson's rule integrator for function stored on the
// radial logarithmic mesh
// routine assumes that mesh is an odd number so run check
// routine assumes that mesh is an odd number so run check
if (mesh % 2 == 0)
{
std::cout << "\n error in subroutine simpson ";
Expand Down Expand Up @@ -83,21 +83,21 @@ void Integral::Simpson_Integral
*/
// simpson's rule integrator for function stored on the
// radial logarithmic mesh
// routine assumes that mesh is an odd number so run check
// routine assumes that mesh is an odd number so run check
assert(mesh&1);

asum = 0.00;
const size_t end = mesh-2;
const size_t end = mesh-2;
for( size_t i=1; i!=end; i+=2 )
{
const double f1 = func[i]*rab[i];
asum += f1 + f1 + func[i+1]*rab[i+1];
const double f1 = func[i]*rab[i];
asum += f1 + f1 + func[i+1]*rab[i+1];
}
const double f1 = func[mesh-2]*rab[mesh-2];
asum += f1+f1;
asum += asum;
asum += func[0]*rab[0] + func[mesh-1]*rab[mesh-1];
asum /= 3.0;
const double f1 = func[mesh-2]*rab[mesh-2];
asum += f1+f1;
asum += asum;
asum += func[0]*rab[0] + func[mesh-1]*rab[mesh-1];
asum /= 3.0;
return;
}// end subroutine simpson

Expand All @@ -124,21 +124,21 @@ void Integral::Simpson_Integral
*/
// simpson's rule integrator for function stored on the
// radial logarithmic mesh
// routine assumes that mesh is an odd number so run check
// routine assumes that mesh is an odd number so run check
assert(mesh&1);

asum = 0.00;
const size_t end = mesh-2;
const size_t end = mesh-2;
for(size_t i=1; i!=end; i+=2 )
{
const double f1 = func[i];
asum += f1 + f1 + func[i+1];
const double f1 = func[i];
asum += f1 + f1 + func[i+1];
}
const double f1 = func[mesh-2];
asum += f1+f1;
asum += asum;
asum += func[0] + func[mesh-1];
asum *= dr/3.0;
const double f1 = func[mesh-2];
asum += f1+f1;
asum += asum;
asum += func[0] + func[mesh-1];
asum *= dr/3.0;
return;
}// end subroutine simpson

Expand Down Expand Up @@ -220,10 +220,10 @@ void Integral::Simpson_Integral_alltoinf

const double asum_all = asum[mesh-1];
for (int i = 0;i < mesh; ++i)
{
{
asum[i] = asum_all - asum[i];
}
return;
}
return;
}

double Integral::simpson(const int n, const double* const f, const double dx)
Expand Down
114 changes: 80 additions & 34 deletions source/source_base/module_fft/fft_cpu.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -344,60 +344,86 @@ void FFT_CPU<double>::clear()
template <>
void FFT_CPU<double>::fftxyfor(std::complex<double>* in, std::complex<double>* out) const
{
int npy = this->nplane * this->ny;
const int npy = this->nplane * this->ny;
const int nx = this->nx;
const int lixy = this->lixy;
const int rixy = this->rixy;
const int nplane = this->nplane;
const fftw_plan planyfor = this->planyfor;
const fftw_plan planxfor1 = this->planxfor1;
const fftw_plan planxfor2 = this->planxfor2;
if (this->xprime)
{

fftw_execute_dft(this->planxfor1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(planxfor1, (fftw_complex*)in, (fftw_complex*)out);
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->lixy + 1; ++i)
#endif
for (int i = 0; i < lixy + 1; ++i)
{
fftw_execute_dft(this->planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
#ifdef _OPENMP
#pragma omp parallel for
for (int i = rixy; i < this->nx; ++i)
#endif
for (int i = rixy; i < nx; ++i)
{
fftw_execute_dft(this->planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
}
else
{
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->nx; ++i)
#endif
for (int i = 0; i < nx; ++i)
{
fftw_execute_dft(this->planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planyfor, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
fftw_execute_dft(this->planxfor1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(this->planxfor2, (fftw_complex*)&in[rixy * nplane], (fftw_complex*)&out[rixy * nplane]);
fftw_execute_dft(planxfor1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(planxfor2, (fftw_complex*)&in[rixy * nplane], (fftw_complex*)&out[rixy * nplane]);
}
}

template <>
void FFT_CPU<double>::fftxybac(std::complex<double>* in,std::complex<double>* out) const
{
int npy = this->nplane * this->ny;
const int npy = this->nplane * this->ny;
const int nx = this->nx;
const int lixy = this->lixy;
const int rixy = this->rixy;
const int nplane = this->nplane;
const fftw_plan planybac = this->planybac;
const fftw_plan planxbac1 = this->planxbac1;
const fftw_plan planxbac2 = this->planxbac2;
if (this->xprime)
{
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->lixy + 1; ++i)
#endif
for (int i = 0; i < lixy + 1; ++i)
{
fftw_execute_dft(this->planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
#ifdef _OPENMP
#pragma omp parallel for
for (int i = rixy; i < this->nx; ++i)
#endif
for (int i = rixy; i < nx; ++i)
{
fftw_execute_dft(this->planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
fftw_execute_dft(this->planxbac1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(planxbac1, (fftw_complex*)in, (fftw_complex*)out);
}
else
{
fftw_execute_dft(this->planxbac1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(this->planxbac2, (fftw_complex*)&in[rixy * nplane], (fftw_complex*)&out[rixy * nplane]);
fftw_execute_dft(planxbac1, (fftw_complex*)in, (fftw_complex*)out);
fftw_execute_dft(planxbac2, (fftw_complex*)&in[rixy * nplane], (fftw_complex*)&out[rixy * nplane]);
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->nx; ++i)
#endif
for (int i = 0; i < nx; ++i)
{
fftw_execute_dft(this->planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&out[i * npy]);
}
}
}
Expand All @@ -417,47 +443,67 @@ void FFT_CPU<double>::fftzbac(std::complex<double>* in, std::complex<double>* ou
template <>
void FFT_CPU<double>::fftxyr2c(double* in, std::complex<double>* out) const
{
int npy = this->nplane * this->ny;
const int npy = this->nplane * this->ny;
const int nx = this->nx;
const int lixy = this->lixy;
const fftw_plan planxr2c = this->planxr2c;
const fftw_plan planyfor = this->planyfor;
const fftw_plan planyr2c = this->planyr2c;
const fftw_plan planxfor1 = this->planxfor1;
if (this->xprime)
{
fftw_execute_dft_r2c(this->planxr2c, in, (fftw_complex*)out);
fftw_execute_dft_r2c(planxr2c, in, (fftw_complex*)out);
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->lixy + 1; ++i)
#endif
for (int i = 0; i < lixy + 1; ++i)
{
fftw_execute_dft(this->planyfor, (fftw_complex*)&out[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft(planyfor, (fftw_complex*)&out[i * npy], (fftw_complex*)&out[i * npy]);
}
}
else
{
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->nx; ++i)
#endif
for (int i = 0; i < nx; ++i)
{
fftw_execute_dft_r2c(this->planyr2c, &in[i * npy], (fftw_complex*)&out[i * npy]);
fftw_execute_dft_r2c(planyr2c, &in[i * npy], (fftw_complex*)&out[i * npy]);
}
fftw_execute_dft(this->planxfor1, (fftw_complex*)out, (fftw_complex*)out);
fftw_execute_dft(planxfor1, (fftw_complex*)out, (fftw_complex*)out);
}
}

template <>
void FFT_CPU<double>::fftxyc2r(std::complex<double> *in,double *out) const
{
int npy = this->nplane * this->ny;
const int npy = this->nplane * this->ny;
const int nx = this->nx;
const int lixy = this->lixy;
const fftw_plan planybac = this->planybac;
const fftw_plan planxc2r = this->planxc2r;
const fftw_plan planxbac1 = this->planxbac1;
const fftw_plan planyc2r = this->planyc2r;
if (this->xprime)
{
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->lixy + 1; ++i)
#endif
for (int i = 0; i < lixy + 1; ++i)
{
fftw_execute_dft(this->planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&in[i * npy]);
fftw_execute_dft(planybac, (fftw_complex*)&in[i * npy], (fftw_complex*)&in[i * npy]);
}
fftw_execute_dft_c2r(this->planxc2r, (fftw_complex*)in, out);
fftw_execute_dft_c2r(planxc2r, (fftw_complex*)in, out);
}
else
{
fftw_execute_dft(this->planxbac1, (fftw_complex*)in, (fftw_complex*)in);
fftw_execute_dft(planxbac1, (fftw_complex*)in, (fftw_complex*)in);
#ifdef _OPENMP
#pragma omp parallel for
for (int i = 0; i < this->nx; ++i)
#endif
for (int i = 0; i < nx; ++i)
{
fftw_execute_dft_c2r(this->planyc2r, (fftw_complex*)&in[i * npy], &out[i * npy]);
fftw_execute_dft_c2r(planyc2r, (fftw_complex*)&in[i * npy], &out[i * npy]);
}
}
}
Expand Down
Loading
Loading