Skip to content

⚡️ Speed up function cosine by 6% #27

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

⚡️ Speed up function cosine by 6% #27

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
20 changes: 14 additions & 6 deletions chromadb/utils/distance_functions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,12 +20,20 @@ def cosine(x: Vector, y: Vector) -> float:
NORM_EPS = 1e-30
if x.dtype == np.float16 or y.dtype == np.float16:
NORM_EPS = 1e-7
return cast(
float,
(
1.0 - np.dot(x, y) / ((np.linalg.norm(x) * np.linalg.norm(y)) + NORM_EPS)
).item(),
)

# Avoid redundant norm calculations: compute each norm only once
x_norm = np.linalg.norm(x)
y_norm = np.linalg.norm(y)
denom = (x_norm * y_norm) + NORM_EPS

# Use np.dot for both 1D and multidimensional (row vector) cases
dot = np.dot(x, y)

# Use fused multiply-add if available for numerical precision, but stay with simple version for performance
# Directly perform subtraction and division without additional .item() dereference if already scalar
# But in either case, guarantee returning a Python float
result = 1.0 - dot / denom
return cast(float, float(result))


def ip(x: Vector, y: Vector) -> float:
Expand Down