Skip to content

perf: cluster & quad-fit core rework (1/4) #1

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

Closed
bouk wants to merge 10 commits into from
Closed

perf: cluster & quad-fit core rework (1/4) #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
f3aa015
Speed up quad fitting: memoize segment fits, reuse scratch, cut sort …
bouk Jun 10, 2026
a335d94
Speed up gradient clustering: lazy rep lookup, masked buckets, parall…
bouk Jun 10, 2026
64fc38a
Sort cluster points as packed u64 keys
bouk Jun 10, 2026
0c71bb8
Run-driven gradient clusters with chunk-pool point storage
bouk Jun 10, 2026
425fc08
Heads-only union-find and cached threshold buffers
bouk Jun 10, 2026
3b9f803
Batch interior overlap points in gradient clusters
bouk Jun 10, 2026
77fc3ff
Branchless sort merge, lean inline path for window fit errors
bouk Jun 10, 2026
aff3b78
Replace pairwise cluster merge tree with single-pass k-way heap merge
bouk Jun 10, 2026
8315d11
Shrink struct pt to 8 bytes; build sort keys in the slope loop
bouk Jun 10, 2026
8be0535
Sort keys only (lfps reads through key indices); SIMD tile min/max an…
bouk Jun 10, 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
3 changes: 3 additions & 0 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,9 @@ endif()
if(CMAKE_COMPILER_IS_GNUCC OR CMAKE_C_COMPILER_ID MATCHES "Clang")
add_compile_options(-Wall -Wextra)
add_compile_options(-Wpedantic)
# sqrt() & friends never need errno here; this lets them compile to bare
# sqrt instructions (vectorizable) instead of guarded libm calls
add_compile_options(-fno-math-errno)
if(CMAKE_C_COMPILER_ID MATCHES "Clang")
add_compile_options(
-Wno-gnu-zero-variadic-macro-arguments
Expand Down
3 changes: 3 additions & 0 deletions apriltag.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -413,6 +413,9 @@ void apriltag_detector_destroy(apriltag_detector_t *td)
zarray_destroy(td->tag_families);
if (td->cached_uf)
unionfind_destroy(td->cached_uf);
if (td->cached_threshim)
image_u8_destroy(td->cached_threshim);
free(td->cached_tile_bufs);
free(td);
}

Expand Down
5 changes: 5 additions & 0 deletions apriltag.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,6 +192,11 @@ struct apriltag_detector

// Cached unionfind structure (reused across detect calls)
unionfind_t *cached_uf;

// Cached threshold buffers (reused across detect calls)
image_u8_t *cached_threshim;
uint8_t *cached_tile_bufs; // 4 contiguous tw*th tile min/max arrays
int cached_tile_bufs_size;
};

// Represents the detection of a tag. These are returned to the user
Expand Down
Loading