Skip to content

Clamp DiskANN int8 quantization to avoid float->int8 UB #311

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

Clamp DiskANN int8 quantization to avoid float->int8 UB #311

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
8 changes: 7 additions & 1 deletion sqlite-vec-diskann.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,7 +157,13 @@ int diskann_quantize_vector(
case VEC0_DISKANN_QUANTIZER_INT8: {
f32 step = (1.0f - (-1.0f)) / 255.0f;
for (size_t i = 0; i < dimensions; i++) {
((i8 *)out)[i] = (i8)(((src[i] - (-1.0f)) / step) - 128.0f);
// Saturate before the cast: converting a float outside [-128, 127] to a
// signed integer type is undefined behavior in C (and silently wraps to
// garbage). Inputs are expected to be normalized to [-1, 1]; clamp so
// out-of-range values quantize to the nearest representable int8.
const f32 raw = ((src[i] - (-1.0f)) / step) - 128.0f;
const f32 q = raw < -128.0f ? -128.0f : (raw > 127.0f ? 127.0f : raw);
((i8 *)out)[i] = (i8)q;
}
return SQLITE_OK;
}
Expand Down