Skip to content

More efficient eval_f in RBC on GPU #492

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

Merged
merged 3 commits into from
Oct 8, 2024
Merged

More efficient eval_f in RBC on GPU #492

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7ec7bc5
More efficient `eval_f` in RBC on GPU
brownbaerchen Sep 26, 2024
34aa041
Merge remote-tracking branch 'upstream/master' into RBC_efficiency
brownbaerchen Sep 27, 2024
6826b5e
Merge remote-tracking branch 'upstream/master' into RBC_efficiency
brownbaerchen Oct 8, 2024
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
18 changes: 11 additions & 7 deletions pySDC/implementations/problem_classes/RayleighBenard.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -191,20 +191,24 @@ def eval_f(self, u, *args, **kwargs):
iu, iv, iT, ip = self.index(['u', 'v', 'T', 'p'])

# evaluate implicit terms
f_impl_hat = -(self.base_change @ self.L @ u_hat.flatten()).reshape(u_hat.shape)
if not hasattr(self, '_L_T_base'):
self._L_T_base = self.base_change @ self.L
f_impl_hat = -(self._L_T_base @ u_hat.flatten()).reshape(u_hat.shape)

if self.spectral_space:
f.impl[:] = f_impl_hat
else:
f.impl[:] = self.itransform(f_impl_hat).real

# -------------------------------------------
# treat convection explicitly with dealiasing
Dx_u_hat = self.u_init_forward
for i in [iu, iv, iT]:
Dx_u_hat[i][:] = (Dx @ u_hat[i].flatten()).reshape(Dx_u_hat[i].shape)
Dz_u_hat = self.u_init_forward
for i in [iu, iv, iT]:
Dz_u_hat[i][:] = (Dz @ u_hat[i].flatten()).reshape(Dz_u_hat[i].shape)

# start by computing derivatives
if not hasattr(self, '_Dx_expanded') or not hasattr(self, '_Dz_expanded'):
self._Dx_expanded = self._setup_operator({'u': {'u': Dx}, 'v': {'v': Dx}, 'T': {'T': Dx}, 'p': {}})
self._Dz_expanded = self._setup_operator({'u': {'u': Dz}, 'v': {'v': Dz}, 'T': {'T': Dz}, 'p': {}})
Dx_u_hat = (self._Dx_expanded @ u_hat.flatten()).reshape(u_hat.shape)
Dz_u_hat = (self._Dz_expanded @ u_hat.flatten()).reshape(u_hat.shape)

padding = [self.dealiasing, self.dealiasing]
Dx_u_pad = self.itransform(Dx_u_hat, padding=padding).real
Expand Down