Skip to content

MDEV-40916: do not materialise rejected MHNSW neighbours #5593

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
chanztuying wants to merge 2 commits into
base: main
Choose a base branch
from
+17 −6
Open

MDEV-40916: do not materialise rejected MHNSW neighbours #5593

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
148eb78
MDEV-38721: do not materialise rejected MHNSW neighbours
chanztuying Aug 24, 2026
7561e47
Merge branch 'main' into mhnsw-remember
chanztuying Aug 28, 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
23 changes: 17 additions & 6 deletions sql/vector_mhnsw.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1128,6 +1128,12 @@ class VisitedSet
count++;
return v;
}
void remember(FVectorNode *node)
{
// Mark a distance-tested node without materialising a queue entry.
insert(node);
count++;
}

@vuvova vuvova Aug 29, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

move count++ inside insert()

  • you'll have it only in one place
  • you won't need remember() at all, just invoke insert() directly

void insert(const FVectorNode *n)
{
nodes[idx++]= n;
Expand Down Expand Up @@ -1367,22 +1373,27 @@ static int search_layer(MHNSW_param *p, const FVector *target, float threshold,
}
else
{
Visited *v= visited.create(links[i],
links[i]->distance_greater_than(target, furthest_best,
p->mode, &p->acc));
if (v->distance_to_target <= threshold)
float distance= links[i]->distance_greater_than(target, furthest_best,
p->mode, &p->acc);
if (distance <= threshold)

@vuvova vuvova Aug 29, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

better write

  if (distance <= threshold || distance >= furthest_best)

{
visited.remember(links[i]);
continue;
if (v->distance_to_target < furthest_best)
}
if (distance < furthest_best)
{
Visited *v= visited.create(links[i], distance);
candidates.safe_push(v);
if (skip_deleted && v->node->deleted)
continue;
if (v->distance_to_target < best.top()->distance_to_target)
if (distance < best.top()->distance_to_target)
{
best.replace_top(v);
furthest_best= lenient_furthest(best, p->acc.diameter, leniency);
}
}
else
visited.remember(links[i]);
}
}
}
Expand Down