liuyu/pgvector
1
0
Fork 0
mirror of https://github.com/pgvector/pgvector.git synced 2026-07-01 18:21:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

Speed up ivfflat build: use float instead of double for dot product (#180)

Browse Source 
calculation

On ARM this makes CPU using vector multiply-add instruction (fmadd)
instead of vector multiplication + conversion to double + addition
(fmul + fcvt + fadd) at each vector dimension.

Output of distance functions and calculations that are are done once
per vector pair are left double as this don't make speed difference
and for compatibility.
This commit is contained in:
Pavel Borisov
2023-07-18 23:55:38 +04:00
committed by GitHub
parent f4c28b1c06
commit 3950bc3dc6
1 changed files with 16 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
src/vector.c
View File

@@ -530,8 +530,8 @@ l2_distance(PG_FUNCTION_ARGS)
Vector *b = PG_GETARG_VECTOR_P(1);
float *ax = a->x;
float *bx = b->x;
double distance = 0.0;
double diff;
float distance = 0.0;
float diff;
CheckDims(a, b);
@@ -542,7 +542,7 @@ l2_distance(PG_FUNCTION_ARGS)
distance += diff * diff;
}
PG_RETURN_FLOAT8(sqrt(distance));
PG_RETURN_FLOAT8(sqrt((double)distance));
}
/*
@@ -557,8 +557,8 @@ vector_l2_squared_distance(PG_FUNCTION_ARGS)
Vector *b = PG_GETARG_VECTOR_P(1);
float *ax = a->x;
float *bx = b->x;
double distance = 0.0;
double diff;
float distance = 0.0;
float diff;
CheckDims(a, b);
@@ -569,7 +569,7 @@ vector_l2_squared_distance(PG_FUNCTION_ARGS)
distance += diff * diff;
}
PG_RETURN_FLOAT8(distance);
PG_RETURN_FLOAT8((double)distance);
}
/*
@@ -583,7 +583,7 @@ inner_product(PG_FUNCTION_ARGS)
Vector *b = PG_GETARG_VECTOR_P(1);
float *ax = a->x;
float *bx = b->x;
double distance = 0.0;
float distance = 0.0;
CheckDims(a, b);
@@ -591,7 +591,7 @@ inner_product(PG_FUNCTION_ARGS)
for (int i = 0; i < a->dim; i++)
distance += ax[i] * bx[i];
PG_RETURN_FLOAT8(distance);
PG_RETURN_FLOAT8((double)distance);
}
/*
@@ -605,7 +605,7 @@ vector_negative_inner_product(PG_FUNCTION_ARGS)
Vector *b = PG_GETARG_VECTOR_P(1);
float *ax = a->x;
float *bx = b->x;
double distance = 0.0;
float distance = 0.0;
CheckDims(a, b);
@@ -613,7 +613,7 @@ vector_negative_inner_product(PG_FUNCTION_ARGS)
for (int i = 0; i < a->dim; i++)
distance += ax[i] * bx[i];
PG_RETURN_FLOAT8(distance * -1);
PG_RETURN_FLOAT8((double)distance * -1);
}
/*
@@ -656,14 +656,16 @@ vector_spherical_distance(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
Vector *b = PG_GETARG_VECTOR_P(1);
double distance = 0.0;
float dp = 0.0;
double distance;
CheckDims(a, b);
/* Auto-vectorized */
for (int i = 0; i < a->dim; i++)
distance += a->x[i] * b->x[i];
dp += a->x[i] * b->x[i];
distance = (double) dp;
/* Prevent NaN with acos with loss of precision */
if (distance > 1)
distance = 1;
@@ -716,13 +718,13 @@ vector_norm(PG_FUNCTION_ARGS)
{
Vector *a = PG_GETARG_VECTOR_P(0);
float *ax = a->x;
double norm = 0.0;
float norm = 0.0;
/* Auto-vectorized */
for (int i = 0; i < a->dim; i++)
norm += ax[i] * ax[i];
PG_RETURN_FLOAT8(sqrt(norm));
PG_RETURN_FLOAT8(sqrt((double)norm));
}
/*