Skip to content

Apply Sum Factorization Again #571

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
jowezarek wants to merge 1 commit into
base: devel
Choose a base branch
from
Open

Apply Sum Factorization Again #571

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
36 changes: 35 additions & 1 deletion examples/performance/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,7 @@ Test cases
Data
----

2025-09-09 17:28:19 (added by Julian O. - ThinkPad T14 on performance mode)
2025-09-09 17:28:19 (added by Julian O. - ThinkPad T14 in performance mode)
----------

| Test case | old assembly | new assembly | old discretization | new discretization |
Expand All @@ -84,3 +84,37 @@ Data
| 3.5 | 0.577 | 0.057 | 1.51 | 1.784 |
| 3.6 | 0.598 | 0.209 | 2.411 | 1.848 |

2026-02-03 09:53:14 (added by Julian O. - ThinkPad T14 in performance mode - after a suspicion that we don't use the newly added SF algorithm by default anymore)
----------

| Test case | old assembly | new assembly | old discretization | new discretization |
| --- | --- | --- | --- | --- |
| 2.1 | 25.953 | 27.642 | 17.806 | 9.641 |
| 2.2 | 2.818 | 3.324 | 6.66 | 3.463 |
| 2.3 | 48.497 | 49.256 | 22.308 | 9.821 |
| 3.1.1 | 0.428 | 0.476 | 2.183 | 1.238 |
| 3.1.2 | 0.983 | 1.336 | 2.323 | 1.2 |
| 3.1.3 | 0.848 | 0.786 | 4.028 | 2.252 |
| 3.2 | 1.676 | 1.369 | 1.513 | 1.662 |
| 3.3 | 0.949 | 0.586 | 1.327 | 1.288 |
| 3.4 | 0.694 | 0.923 | 1.144 | 1.161 |
| 3.5 | 0.612 | 0.558 | 1.564 | 1.511 |
| 3.6 | 1.865 | 0.624 | 2.379 | 1.76 |

2026-02-03 10:37:52 (added by Julian O. - ThinkPad T14 in performance mode - after fixing the above issue)
----------

| Test case | old assembly | new assembly | old discretization | new discretization |
| --- | --- | --- | --- | --- |
| 2.1 | 30.908 | 2.164 | 21.086 | 21.173 |
| 2.2 | 3.193 | 0.581 | 6.229 | 4.364 |
| 2.3 | 43.744 | 3.914 | 21.244 | 23.042 |
| 3.1.1 | 0.242 | 0.043 | 1.713 | 1.798 |
| 3.1.2 | 0.561 | 0.045 | 1.701 | 1.522 |
| 3.1.3 | 0.897 | 0.161 | 4.085 | 3.296 |
| 3.2 | 1.043 | 0.046 | 1.364 | 1.691 |
| 3.3 | 0.57 | 0.082 | 1.045 | 2.024 |
| 3.4 | 1.656 | 0.073 | 1.775 | 2.204 |
| 3.5 | 0.563 | 0.074 | 1.788 | 2.274 |
| 3.6 | 0.768 | 0.211 | 3.307 | 1.865 |

Comment on lines +87 to +120
Copy link
Member

@yguclu yguclu Feb 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am not sure that we should report all of these numbers. In particular, the timings seem to oscillate quite a bit. Maybe one should run these benchmarks a few times and report the average value?

2 changes: 1 addition & 1 deletion examples/performance/compare_3d_matrix_assembly_speed.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -627,7 +627,7 @@ def make_square_torus_geometry_3d(ncells, degree, comm=None):

if mpi_rank == 0:
# Write performance table to MarkDown file
with open('matrix_assembly_speed_log.md', 'a') as f:
with open('README.md', 'a') as f:
f.write(txt)

# Remove temporary folders
Expand Down
5 changes: 4 additions & 1 deletion psydac/api/discretization.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -659,7 +659,10 @@ def discretize(a, *args, **kwargs):
if isinstance(a, sym_BilinearForm):
# Currently sum factorization can only be used for interior domains
from sympde.expr.evaluation import DomainExpression
is_interior_expr = isinstance(kernel_expr, DomainExpression)
if isinstance(kernel_expr, tuple):
is_interior_expr = isinstance(kernel_expr[0], DomainExpression)
Comment on lines +662 to +663
Copy link
Member

@yguclu yguclu Feb 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In practice, is kernel_expr anything else than a tuple? Could it be a list?

else:
is_interior_expr = isinstance(kernel_expr, DomainExpression)

if kwargs.pop('sum_factorization') and is_interior_expr:
return DiscreteBilinearForm_SF(a, kernel_expr, *args, **kwargs)
Expand Down