Skip to content

perf: avoid post-pass column reconstruction in NaN preprocessing #6

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
CyrilJl merged 6 commits into from
Mar 20, 2026
Merged

perf: avoid post-pass column reconstruction in NaN preprocessing #6

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
91f0343
Add OptiMask performance tooling
CyrilJl Mar 19, 2026
7d2e7fe
Trial 5: track NaN columns during preprocess
CyrilJl Mar 19, 2026
78c5f71
Record OptiMask trial benchmark results
CyrilJl Mar 19, 2026
6aead48
Update best-branch benchmark summary
CyrilJl Mar 19, 2026
c91b16b
Ignore dev folder on trial 05 branch
CyrilJl Mar 20, 2026
d90aa0f
Print result sizes in speed test
CyrilJl Mar 20, 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
6 changes: 3 additions & 3 deletions optimask/_optimask.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,6 +153,7 @@ def _preprocess(x):
iy, ix = np.empty(m * n, dtype=np.uint32), np.empty(m * n, dtype=np.uint32)
cols_index_mapper = -np.ones(n, dtype=np.int32)
rows_with_nan = np.empty(m, dtype=np.uint32)
cols_with_nan = np.empty(n, dtype=np.uint32)
n_rows_with_nan = 0
n_cols_with_nan = 0
cnt = 0
Expand All @@ -168,6 +169,7 @@ def _preprocess(x):
else:
ix[cnt] = n_cols_with_nan
cols_index_mapper[j] = n_cols_with_nan
cols_with_nan[n_cols_with_nan] = j
n_cols_with_nan += 1
cnt += 1

Expand All @@ -177,9 +179,7 @@ def _preprocess(x):

iy, ix = iy[:cnt], ix[:cnt]
rows_with_nan = rows_with_nan[:n_rows_with_nan]
cols_with_nan = np.flatnonzero(cols_index_mapper >= 0)[
cols_index_mapper[cols_index_mapper >= 0].argsort()
].astype(np.uint32)
cols_with_nan = cols_with_nan[:n_cols_with_nan]
return iy, ix, rows_with_nan, cols_with_nan

def _trial(self, k, rng, m_nan, n_nan, iy, ix, m, n):
Expand Down
8 changes: 4 additions & 4 deletions tests/test_optimask.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -126,14 +126,14 @@ def test_speed(opti_mask_instance):
print("\nVertical arrays")
for _ in range(5):
start = perf_counter()
opti_mask_instance.solve(X=x)
print(f"{1e3 * (perf_counter() - start):.2f}ms")
rows, cols = opti_mask_instance.solve(X=x)
print(f"{1e3 * (perf_counter() - start):.2f}ms rows={rows.size} cols={cols.size}")
x = generate_random(m=1_000, n=100_000, ratio=0.02)
print("Horizontal arrays")
for _ in range(5):
start = perf_counter()
opti_mask_instance.solve(X=x)
print(f"{1e3 * (perf_counter() - start):.2f}ms")
rows, cols = opti_mask_instance.solve(X=x)
print(f"{1e3 * (perf_counter() - start):.2f}ms rows={rows.size} cols={cols.size}")


def test_large_arrays(opti_mask_instance):
Expand Down
Loading