Skip to content

Use appropriate return type for byte vector specific cosineBody* functions #15293

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
merged 1 commit into from
Oct 7, 2025
Merged

Use appropriate return type for byte vector specific cosineBody* functions #15293

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
14 changes: 7 additions & 7 deletions lucene/core/src/java25/org/apache/lucene/internal/vectorization/PanamaVectorUtilSupport.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -700,7 +700,7 @@ private static float cosineBody(ByteVectorLoader a, ByteVectorLoader b) {

// only vectorize if we'll at least enter the loop a single time
if (a.length() >= 16) {
final float[] ret;
final int[] ret;
if (VECTOR_BITSIZE >= 512) {
i += BYTE_SPECIES.loopBound(a.length());
ret = cosineBody512(a, b, i);
Expand Down Expand Up @@ -729,7 +729,7 @@ private static float cosineBody(ByteVectorLoader a, ByteVectorLoader b) {
}

/** vectorized cosine body (512 bit vectors) */
private static float[] cosineBody512(ByteVectorLoader a, ByteVectorLoader b, int limit) {
private static int[] cosineBody512(ByteVectorLoader a, ByteVectorLoader b, int limit) {
IntVector accSum = IntVector.zero(INT_SPECIES);
IntVector accNorm1 = IntVector.zero(INT_SPECIES);
IntVector accNorm2 = IntVector.zero(INT_SPECIES);
Expand All @@ -753,13 +753,13 @@ private static float[] cosineBody512(ByteVectorLoader a, ByteVectorLoader b, int
accSum = accSum.add(prod32);
}
// reduce
return new float[] {
return new int[] {
accSum.reduceLanes(ADD), accNorm1.reduceLanes(ADD), accNorm2.reduceLanes(ADD)
};
}

/** vectorized cosine body (256 bit vectors) */
private static float[] cosineBody256(ByteVectorLoader a, ByteVectorLoader b, int limit) {
private static int[] cosineBody256(ByteVectorLoader a, ByteVectorLoader b, int limit) {
IntVector accSum = IntVector.zero(IntVector.SPECIES_256);
IntVector accNorm1 = IntVector.zero(IntVector.SPECIES_256);
IntVector accNorm2 = IntVector.zero(IntVector.SPECIES_256);
Expand All @@ -778,13 +778,13 @@ private static float[] cosineBody256(ByteVectorLoader a, ByteVectorLoader b, int
accSum = accSum.add(prod32);
}
// reduce
return new float[] {
return new int[] {
accSum.reduceLanes(ADD), accNorm1.reduceLanes(ADD), accNorm2.reduceLanes(ADD)
};
}

/** vectorized cosine body (128 bit vectors) */
private static float[] cosineBody128(ByteVectorLoader a, ByteVectorLoader b, int limit) {
private static int[] cosineBody128(ByteVectorLoader a, ByteVectorLoader b, int limit) {
IntVector accSum = IntVector.zero(IntVector.SPECIES_128);
IntVector accNorm1 = IntVector.zero(IntVector.SPECIES_128);
IntVector accNorm2 = IntVector.zero(IntVector.SPECIES_128);
Expand All @@ -805,7 +805,7 @@ private static float[] cosineBody128(ByteVectorLoader a, ByteVectorLoader b, int
accSum = accSum.add(prod16.convertShape(S2I, IntVector.SPECIES_128, 0));
}
// reduce
return new float[] {
return new int[] {
accSum.reduceLanes(ADD), accNorm1.reduceLanes(ADD), accNorm2.reduceLanes(ADD)
};
}
Expand Down
Loading