Compare commits

..

15 Commits

Author SHA1 Message Date
Andrew Kane
421116c8cd Improved name [skip ci] 2024-09-24 11:48:13 -07:00
Andrew Kane
3623c64d82 Added comment [skip ci] 2024-09-24 11:46:13 -07:00
Andrew Kane
18be1fda24 Fixed lookup table [skip ci] 2024-09-24 01:25:20 -07:00
Andrew Kane
e10a5c90df Fixed code [skip ci] 2024-09-24 00:36:23 -07:00
Andrew Kane
3b0976b725 Fixed rounding [skip ci] 2024-09-24 00:24:02 -07:00
Andrew Kane
724bde5020 Improved code [skip ci] 2024-09-23 22:07:30 -07:00
Andrew Kane
261ba4a0d9 Added test for IVFFlat [skip ci] 2024-09-23 18:57:13 -07:00
Andrew Kane
4293bc439e Exclude zero vectors for cosine distance to be consistent with other types [skip ci] 2024-09-23 18:55:10 -07:00
Andrew Kane
044f0c6441 Improved precision for cosine distance [skip ci] 2024-09-23 18:53:23 -07:00
Andrew Kane
22efa8ec22 Improved precision for cosine distance [skip ci] 2024-09-23 18:50:10 -07:00
Andrew Kane
0a2803130f Added support for IVFFlat [skip ci] 2024-09-23 18:40:49 -07:00
Andrew Kane
6105a01881 Added more tests [skip ci] 2024-09-23 18:33:08 -07:00
Andrew Kane
9925b92d73 Added more minivec tests [skip ci] 2024-09-23 18:29:17 -07:00
Andrew Kane
e49bae9dc3 Fixed CI 2024-09-23 18:24:21 -07:00
Andrew Kane
eea2c44fae Added minivec type 2024-09-23 18:12:26 -07:00
47 changed files with 3679 additions and 1000 deletions

View File

@@ -1,9 +1,6 @@
## 0.8.0 (unreleased)
- Added support for inline filtering with IVFFlat
- Added casts for arrays to `sparsevec`
- Improved cost estimation
- Improved performance of HNSW inserts and on-disk index builds
- Reduced memory usage for HNSW index scans
- Dropped support for Postgres 12

View File

@@ -1,4 +1,4 @@
ARG PG_MAJOR=17
ARG PG_MAJOR=16
FROM postgres:$PG_MAJOR
ARG PG_MAJOR

View File

@@ -4,8 +4,8 @@ EXTVERSION = 0.7.4
MODULE_big = vector
DATA = $(wildcard sql/*--*--*.sql)
DATA_built = sql/$(EXTENSION)--$(EXTVERSION).sql
OBJS = src/bitutils.o src/bitvec.o src/halfutils.o src/halfvec.o src/hnsw.o src/hnswbuild.o src/hnswinsert.o src/hnswscan.o src/hnswutils.o src/hnswvacuum.o src/ivfbuild.o src/ivfflat.o src/ivfinsert.o src/ivfkmeans.o src/ivfscan.o src/ivfutils.o src/ivfvacuum.o src/sparsevec.o src/vector.o
HEADERS = src/halfvec.h src/sparsevec.h src/vector.h
OBJS = src/bitutils.o src/bitvec.o src/halfutils.o src/halfvec.o src/hnsw.o src/hnswbuild.o src/hnswinsert.o src/hnswscan.o src/hnswutils.o src/hnswvacuum.o src/ivfbuild.o src/ivfflat.o src/ivfinsert.o src/ivfkmeans.o src/ivfscan.o src/ivfutils.o src/minivec.o src/ivfvacuum.o src/sparsevec.o src/vector.o
HEADERS = src/halfvec.h src/minivec.h src/sparsevec.h src/vector.h
TESTS = $(wildcard test/sql/*.sql)
REGRESS = $(patsubst test/sql/%.sql,%,$(TESTS))
@@ -66,7 +66,7 @@ dist:
git archive --format zip --prefix=$(EXTENSION)-$(EXTVERSION)/ --output dist/$(EXTENSION)-$(EXTVERSION).zip master
# for Docker
PG_MAJOR ?= 17
PG_MAJOR ?= 16
.PHONY: docker

View File

@@ -2,10 +2,10 @@ EXTENSION = vector
EXTVERSION = 0.7.4
DATA_built = sql\$(EXTENSION)--$(EXTVERSION).sql
OBJS = src\bitutils.obj src\bitvec.obj src\halfutils.obj src\halfvec.obj src\hnsw.obj src\hnswbuild.obj src\hnswinsert.obj src\hnswscan.obj src\hnswutils.obj src\hnswvacuum.obj src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\sparsevec.obj src\vector.obj
HEADERS = src\halfvec.h src\sparsevec.h src\vector.h
OBJS = src\bitutils.obj src\bitvec.obj src\halfutils.obj src\halfvec.obj src\hnsw.obj src\hnswbuild.obj src\hnswinsert.obj src\hnswscan.obj src\hnswutils.obj src\hnswvacuum.obj src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\minivec.obj src\sparsevec.obj src\vector.obj
HEADERS = src\halfvec.h src\minivec.h src\sparsevec.h src\vector.h
REGRESS = bit btree cast copy halfvec hnsw_bit hnsw_halfvec hnsw_sparsevec hnsw_vector ivfflat_bit ivfflat_halfvec ivfflat_vector sparsevec vector_type
REGRESS = bit btree cast copy halfvec hnsw_bit hnsw_halfvec hnsw_sparsevec hnsw_vector ivfflat_bit ivfflat_halfvec ivfflat_vector minivec sparsevec vector_type
REGRESS_OPTS = --inputdir=test --load-extension=$(EXTENSION)
# For /arch flags

View File

@@ -52,8 +52,6 @@ nmake /F Makefile.win
nmake /F Makefile.win install
```
Note: Postgres 17 is not supported yet due to an upstream issue
See the [installation notes](#installation-notes---windows) if you run into issues
You can also install it with [Docker](#docker) or [conda-forge](#conda-forge).
@@ -102,8 +100,6 @@ Or add a vector column to an existing table
ALTER TABLE items ADD COLUMN embedding vector(3);
```
Also supports [half-precision](#half-precision-vectors), [binary](#binary-vectors), and [sparse](#sparse-vectors) vectors
Insert vectors
```sql
@@ -149,8 +145,6 @@ Supported distance functions are:
- `<#>` - (negative) inner product
- `<=>` - cosine distance
- `<+>` - L1 distance (added in 0.7.0)
- `<~>` - Hamming distance (binary vectors, added in 0.7.0)
- `<%>` - Jaccard distance (binary vectors, added in 0.7.0)
Get the nearest neighbors to a row
@@ -439,12 +433,6 @@ Create an index on one [or more](https://www.postgresql.org/docs/current/indexes
CREATE INDEX ON items (category_id);
```
Or a composite IVFFlat index for approximate search (added in 0.8.0)
```sql
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops, category_id) WITH (lists = 100);
```
Or a [partial index](https://www.postgresql.org/docs/current/indexes-partial.html) on the vector column for approximate search
```sql
@@ -946,6 +934,37 @@ Function | Description | Added
avg(halfvec) → halfvec | average | 0.7.0
sum(halfvec) → halfvec | sum | 0.7.0
### Minivec Type
Each mini vector takes `dimensions + 8` bytes of storage. Each element is a E4M3 8-bit floating-point number, and all elements must be finite (no `NaN`). Mini vectors can have up to 16,000 dimensions.
### Minivec Operators
Operator | Description | Added
--- | --- | ---
\+ | element-wise addition | 0.8.0
\- | element-wise subtraction | 0.8.0
\* | element-wise multiplication | 0.8.0
\|\| | concatenate | 0.8.0
<-> | Euclidean distance | 0.8.0
<#> | negative inner product | 0.8.0
<=> | cosine distance | 0.8.0
<+> | taxicab distance | 0.8.0
### Minivec Functions
Function | Description | Added
--- | --- | ---
binary_quantize(minivec) → bit | binary quantize | 0.8.0
cosine_distance(minivec, minivec) → double precision | cosine distance | 0.8.0
inner_product(minivec, minivec) → double precision | inner product | 0.8.0
l1_distance(minivec, minivec) → double precision | taxicab distance | 0.8.0
l2_distance(minivec, minivec) → double precision | Euclidean distance | 0.8.0
l2_norm(minivec) → double precision | Euclidean norm | 0.8.0
l2_normalize(minivec) → minivec | Normalize with Euclidean norm | 0.8.0
subvector(minivec, integer, integer) → minivec | subvector | 0.8.0
vector_dims(minivec) → integer | number of dimensions | 0.8.0
### Bit Type
Each bit vector takes `dimensions / 8 + 8` bytes of storage. See the [Postgres docs](https://www.postgresql.org/docs/current/datatype-bit.html) for more info.
@@ -995,7 +1014,7 @@ l2_normalize(sparsevec) → sparsevec | Normalize with Euclidean norm | 0.7.0
If your machine has multiple Postgres installations, specify the path to [pg_config](https://www.postgresql.org/docs/current/app-pgconfig.html) with:
```sh
export PG_CONFIG=/Library/PostgreSQL/17/bin/pg_config
export PG_CONFIG=/Library/PostgreSQL/16/bin/pg_config
```
Then re-run the installation instructions (run `make clean` before `make` if needed). If `sudo` is needed for `make install`, use:
@@ -1006,11 +1025,11 @@ sudo --preserve-env=PG_CONFIG make install
A few common paths on Mac are:
- EDB installer - `/Library/PostgreSQL/17/bin/pg_config`
- Homebrew (arm64) - `/opt/homebrew/opt/postgresql@17/bin/pg_config`
- Homebrew (x86-64) - `/usr/local/opt/postgresql@17/bin/pg_config`
- EDB installer - `/Library/PostgreSQL/16/bin/pg_config`
- Homebrew (arm64) - `/opt/homebrew/opt/postgresql@16/bin/pg_config`
- Homebrew (x86-64) - `/usr/local/opt/postgresql@16/bin/pg_config`
Note: Replace `17` with your Postgres server version
Note: Replace `16` with your Postgres server version
### Missing Header
@@ -1019,10 +1038,10 @@ If compilation fails with `fatal error: postgres.h: No such file or directory`,
For Ubuntu and Debian, use:
```sh
sudo apt install postgresql-server-dev-17
sudo apt install postgresql-server-dev-16
```
Note: Replace `17` with your Postgres server version
Note: Replace `16` with your Postgres server version
### Missing SDK
@@ -1055,17 +1074,17 @@ If installation fails with `Access is denied`, re-run the installation instructi
Get the [Docker image](https://hub.docker.com/r/pgvector/pgvector) with:
```sh
docker pull pgvector/pgvector:pg17
docker pull pgvector/pgvector:pg16
```
This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `17` with your Postgres server version, and run it the same way).
This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `16` with your Postgres server version, and run it the same way).
You can also build the image manually:
```sh
git clone --branch v0.7.4 https://github.com/pgvector/pgvector.git
cd pgvector
docker build --pull --build-arg PG_MAJOR=17 -t myuser/pgvector .
docker build --pull --build-arg PG_MAJOR=16 -t myuser/pgvector .
```
### Homebrew
@@ -1076,7 +1095,7 @@ With Homebrew Postgres, you can use:
brew install pgvector
```
Note: This only adds it to the `postgresql@17` and `postgresql@14` formulas
Note: This only adds it to the `postgresql@14` formula
### PGXN
@@ -1091,22 +1110,22 @@ pgxn install vector
Debian and Ubuntu packages are available from the [PostgreSQL APT Repository](https://wiki.postgresql.org/wiki/Apt). Follow the [setup instructions](https://wiki.postgresql.org/wiki/Apt#Quickstart) and run:
```sh
sudo apt install postgresql-17-pgvector
sudo apt install postgresql-16-pgvector
```
Note: Replace `17` with your Postgres server version
Note: Replace `16` with your Postgres server version
### Yum
RPM packages are available from the [PostgreSQL Yum Repository](https://yum.postgresql.org/). Follow the [setup instructions](https://www.postgresql.org/download/linux/redhat/) for your distribution and run:
```sh
sudo yum install pgvector_17
sudo yum install pgvector_16
# or
sudo dnf install pgvector_17
sudo dnf install pgvector_16
```
Note: Replace `17` with your Postgres server version
Note: Replace `16` with your Postgres server version
### pkg

View File

@@ -1,6 +1,8 @@
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
\echo Use "ALTER EXTENSION vector UPDATE TO '0.8.0'" to load this file. \quit
-- TODO minivec functions
CREATE FUNCTION array_to_sparsevec(integer[], integer, boolean) RETURNS sparsevec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
@@ -24,5 +26,3 @@ CREATE CAST (double precision[] AS sparsevec)
CREATE CAST (numeric[] AS sparsevec)
WITH FUNCTION array_to_sparsevec(numeric[], integer, boolean) AS ASSIGNMENT;
-- TODO add ivfflat attributes

View File

@@ -266,12 +266,18 @@ COMMENT ON ACCESS METHOD hnsw IS 'hnsw index access method';
CREATE FUNCTION ivfflat_halfvec_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
CREATE FUNCTION ivfflat_minivec_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
CREATE FUNCTION ivfflat_bit_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
CREATE FUNCTION hnsw_halfvec_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
CREATE FUNCTION hnsw_minivec_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
CREATE FUNCTION hnsw_bit_support(internal) RETURNS internal
AS 'MODULE_PATHNAME' LANGUAGE C;
@@ -647,6 +653,295 @@ CREATE OPERATOR CLASS halfvec_l1_ops
FUNCTION 1 l1_distance(halfvec, halfvec),
FUNCTION 3 hnsw_halfvec_support(internal);
-- minivec type
CREATE TYPE minivec;
CREATE FUNCTION minivec_in(cstring, oid, integer) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_out(minivec) RETURNS cstring
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_typmod_in(cstring[]) RETURNS integer
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_recv(internal, oid, integer) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_send(minivec) RETURNS bytea
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE TYPE minivec (
INPUT = minivec_in,
OUTPUT = minivec_out,
TYPMOD_IN = minivec_typmod_in,
RECEIVE = minivec_recv,
SEND = minivec_send,
STORAGE = external
);
-- minivec functions
CREATE FUNCTION l2_distance(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME', 'minivec_l2_distance' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION inner_product(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME', 'minivec_inner_product' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION cosine_distance(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME', 'minivec_cosine_distance' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION l1_distance(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME', 'minivec_l1_distance' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION vector_dims(minivec) RETURNS integer
AS 'MODULE_PATHNAME', 'minivec_vector_dims' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION l2_norm(minivec) RETURNS float8
AS 'MODULE_PATHNAME', 'minivec_l2_norm' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION l2_normalize(minivec) RETURNS minivec
AS 'MODULE_PATHNAME', 'minivec_l2_normalize' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION binary_quantize(minivec) RETURNS bit
AS 'MODULE_PATHNAME', 'minivec_binary_quantize' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION subvector(minivec, int, int) RETURNS minivec
AS 'MODULE_PATHNAME', 'minivec_subvector' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
-- minivec private functions
CREATE FUNCTION minivec_add(minivec, minivec) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_sub(minivec, minivec) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_mul(minivec, minivec) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_concat(minivec, minivec) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_lt(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_le(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_eq(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_ne(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_ge(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_gt(minivec, minivec) RETURNS bool
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_cmp(minivec, minivec) RETURNS int4
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_l2_squared_distance(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_negative_inner_product(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_spherical_distance(minivec, minivec) RETURNS float8
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
-- minivec cast functions
CREATE FUNCTION minivec(minivec, integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_to_vector(minivec, integer, boolean) RETURNS vector
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION vector_to_minivec(vector, integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION array_to_minivec(integer[], integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION array_to_minivec(real[], integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION array_to_minivec(double precision[], integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION array_to_minivec(numeric[], integer, boolean) RETURNS minivec
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
CREATE FUNCTION minivec_to_float4(minivec, integer, boolean) RETURNS real[]
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
-- minivec casts
CREATE CAST (minivec AS minivec)
WITH FUNCTION minivec(minivec, integer, boolean) AS IMPLICIT;
CREATE CAST (minivec AS vector)
WITH FUNCTION minivec_to_vector(minivec, integer, boolean) AS ASSIGNMENT;
CREATE CAST (vector AS minivec)
WITH FUNCTION vector_to_minivec(vector, integer, boolean) AS IMPLICIT;
CREATE CAST (minivec AS real[])
WITH FUNCTION minivec_to_float4(minivec, integer, boolean) AS ASSIGNMENT;
CREATE CAST (integer[] AS minivec)
WITH FUNCTION array_to_minivec(integer[], integer, boolean) AS ASSIGNMENT;
CREATE CAST (real[] AS minivec)
WITH FUNCTION array_to_minivec(real[], integer, boolean) AS ASSIGNMENT;
CREATE CAST (double precision[] AS minivec)
WITH FUNCTION array_to_minivec(double precision[], integer, boolean) AS ASSIGNMENT;
CREATE CAST (numeric[] AS minivec)
WITH FUNCTION array_to_minivec(numeric[], integer, boolean) AS ASSIGNMENT;
-- minivec operators
CREATE OPERATOR <-> (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = l2_distance,
COMMUTATOR = '<->'
);
CREATE OPERATOR <#> (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_negative_inner_product,
COMMUTATOR = '<#>'
);
CREATE OPERATOR <=> (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = cosine_distance,
COMMUTATOR = '<=>'
);
CREATE OPERATOR <+> (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = l1_distance,
COMMUTATOR = '<+>'
);
CREATE OPERATOR + (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_add,
COMMUTATOR = +
);
CREATE OPERATOR - (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_sub
);
CREATE OPERATOR * (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_mul,
COMMUTATOR = *
);
CREATE OPERATOR || (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_concat
);
CREATE OPERATOR < (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_lt,
COMMUTATOR = > , NEGATOR = >= ,
RESTRICT = scalarltsel, JOIN = scalarltjoinsel
);
CREATE OPERATOR <= (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_le,
COMMUTATOR = >= , NEGATOR = > ,
RESTRICT = scalarlesel, JOIN = scalarlejoinsel
);
CREATE OPERATOR = (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_eq,
COMMUTATOR = = , NEGATOR = <> ,
RESTRICT = eqsel, JOIN = eqjoinsel
);
CREATE OPERATOR <> (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_ne,
COMMUTATOR = <> , NEGATOR = = ,
RESTRICT = eqsel, JOIN = eqjoinsel
);
CREATE OPERATOR >= (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_ge,
COMMUTATOR = <= , NEGATOR = < ,
RESTRICT = scalargesel, JOIN = scalargejoinsel
);
CREATE OPERATOR > (
LEFTARG = minivec, RIGHTARG = minivec, PROCEDURE = minivec_gt,
COMMUTATOR = < , NEGATOR = <= ,
RESTRICT = scalargtsel, JOIN = scalargtjoinsel
);
-- minivec op classes
CREATE OPERATOR CLASS minivec_ops
DEFAULT FOR TYPE minivec USING btree AS
OPERATOR 1 < ,
OPERATOR 2 <= ,
OPERATOR 3 = ,
OPERATOR 4 >= ,
OPERATOR 5 > ,
FUNCTION 1 minivec_cmp(minivec, minivec);
CREATE OPERATOR CLASS minivec_l2_ops
FOR TYPE minivec USING ivfflat AS
OPERATOR 1 <-> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 minivec_l2_squared_distance(minivec, minivec),
FUNCTION 3 l2_distance(minivec, minivec),
FUNCTION 5 ivfflat_minivec_support(internal);
CREATE OPERATOR CLASS minivec_ip_ops
FOR TYPE minivec USING ivfflat AS
OPERATOR 1 <#> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 minivec_negative_inner_product(minivec, minivec),
FUNCTION 3 minivec_spherical_distance(minivec, minivec),
FUNCTION 4 l2_norm(minivec),
FUNCTION 5 ivfflat_minivec_support(internal);
CREATE OPERATOR CLASS minivec_cosine_ops
FOR TYPE minivec USING ivfflat AS
OPERATOR 1 <=> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 cosine_distance(minivec, minivec),
FUNCTION 2 l2_norm(minivec),
FUNCTION 3 minivec_spherical_distance(minivec, minivec),
FUNCTION 4 l2_norm(minivec),
FUNCTION 5 ivfflat_minivec_support(internal);
CREATE OPERATOR CLASS minivec_l2_ops
FOR TYPE minivec USING hnsw AS
OPERATOR 1 <-> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 minivec_l2_squared_distance(minivec, minivec),
FUNCTION 3 hnsw_minivec_support(internal);
CREATE OPERATOR CLASS minivec_ip_ops
FOR TYPE minivec USING hnsw AS
OPERATOR 1 <#> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 minivec_negative_inner_product(minivec, minivec),
FUNCTION 3 hnsw_minivec_support(internal);
CREATE OPERATOR CLASS minivec_cosine_ops
FOR TYPE minivec USING hnsw AS
OPERATOR 1 <=> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 cosine_distance(minivec, minivec),
FUNCTION 2 l2_norm(minivec),
FUNCTION 3 hnsw_minivec_support(internal);
CREATE OPERATOR CLASS minivec_l1_ops
FOR TYPE minivec USING hnsw AS
OPERATOR 1 <+> (minivec, minivec) FOR ORDER BY float_ops,
FUNCTION 1 l1_distance(minivec, minivec),
FUNCTION 3 hnsw_minivec_support(internal);
-- bit functions
CREATE FUNCTION hamming_distance(bit, bit) RETURNS float8
@@ -916,13 +1211,3 @@ CREATE OPERATOR CLASS sparsevec_l1_ops
OPERATOR 1 <+> (sparsevec, sparsevec) FOR ORDER BY float_ops,
FUNCTION 1 l1_distance(sparsevec, sparsevec),
FUNCTION 3 hnsw_sparsevec_support(internal);
-- ivfflat attributes
CREATE OPERATOR CLASS vector_integer_ops
DEFAULT FOR TYPE integer USING ivfflat AS
OPERATOR 2 < ,
OPERATOR 3 <= ,
OPERATOR 4 = ,
OPERATOR 5 >= ,
OPERATOR 6 > ;

View File

@@ -12,7 +12,6 @@
#include "utils/float.h"
#include "utils/guc.h"
#include "utils/selfuncs.h"
#include "utils/spccache.h"
#if PG_VERSION_NUM < 150000
#define MarkGUCPrefixReserved(x) EmitWarningsOnPlaceholders(x)
@@ -100,9 +99,7 @@ hnswcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
{
GenericCosts costs;
int m;
double ratio;
double startupPages;
double spc_seq_page_cost;
int entryLevel;
Relation index;
/* Never use index without order */
@@ -118,71 +115,21 @@ hnswcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
MemSet(&costs, 0, sizeof(costs));
genericcostestimate(root, path, loop_count, &costs);
index = index_open(path->indexinfo->indexoid, NoLock);
HnswGetMetaPageInfo(index, &m, NULL);
index_close(index, NoLock);
/*
* HNSW cost estimation follows a formula that accounts for the total
* number of tuples indexed combined with the parameters that most
* influence the duration of the index scan, namely: m - the number of
* tuples that are scanned in each step of the HNSW graph traversal
* ef_search - which influences the total number of steps taken at layer 0
*
* The source of the vector data can impact how many steps it takes to
* converge on the set of vectors to return to the executor. Currently, we
* use a hardcoded scaling factor (HNSWScanScalingFactor) to help
* influence that, but this could later become a configurable parameter
* based on the cost estimations.
*
* The tuple estimator formula is below:
*
* numIndexTuples = entryLevel * m + layer0TuplesMax * layer0Selectivity
*
* "entryLevel * m" represents the floor of tuples we need to scan to get
* to layer 0 (L0).
*
* "layer0TuplesMax" is the estimated total number of tuples we'd scan at
* L0 if we weren't discarding already visited tuples as part of the scan.
*
* "layer0Selectivity" estimates the percentage of tuples that are scanned
* at L0, accounting for previously visited tuples, multiplied by the
* "scalingFactor" (currently hardcoded).
*/
if (path->indexinfo->tuples > 0)
{
double scalingFactor = 0.55;
int entryLevel = (int) (log(path->indexinfo->tuples) * HnswGetMl(m));
int layer0TuplesMax = HnswGetLayerM(m, 0) * hnsw_ef_search;
double layer0Selectivity = scalingFactor * log(path->indexinfo->tuples) / (log(m) * (1 + log(hnsw_ef_search)));
/* Approximate entry level */
entryLevel = (int) -log(1.0 / path->indexinfo->tuples) * HnswGetMl(m);
ratio = (entryLevel * m + layer0TuplesMax * layer0Selectivity) / path->indexinfo->tuples;
/* TODO Improve estimate of visited tuples (currently underestimates) */
/* Account for number of tuples (or entry level), m, and ef_search */
costs.numIndexTuples = (entryLevel + 2) * m;
if (ratio > 1)
ratio = 1;
}
else
ratio = 1;
genericcostestimate(root, path, loop_count, &costs);
get_tablespace_page_costs(path->indexinfo->reltablespace, NULL, &spc_seq_page_cost);
/* Startup cost is cost before returning the first row */
costs.indexStartupCost = costs.indexTotalCost * ratio;
/* Adjust cost if needed since TOAST not included in seq scan cost */
startupPages = costs.numIndexPages * ratio;
if (startupPages > path->indexinfo->rel->pages && ratio < 0.5)
{
/* Change all page cost from random to sequential */
costs.indexStartupCost -= startupPages * (costs.spc_random_page_cost - spc_seq_page_cost);
/* Remove cost of extra pages */
costs.indexStartupCost -= (startupPages - path->indexinfo->rel->pages) * spc_seq_page_cost;
}
*indexStartupCost = costs.indexStartupCost;
/* Use total cost since most work happens before first tuple is returned */
*indexStartupCost = costs.indexTotalCost;
*indexTotalCost = costs.indexTotalCost;
*indexSelectivity = costs.indexSelectivity;
*indexCorrelation = costs.indexCorrelation;

View File

@@ -88,9 +88,6 @@
/* Ensure fits on page and in uint8 */
#define HnswGetMaxLevel(m) Min(((BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(HnswPageOpaqueData)) - offsetof(HnswNeighborTupleData, indextids) - sizeof(ItemIdData)) / (sizeof(ItemPointerData)) / (m)) - 2, 255)
#define HnswGetSearchCandidate(membername, ptr) pairingheap_container(HnswSearchCandidate, membername, ptr)
#define HnswGetSearchCandidateConst(membername, ptr) pairingheap_const_container(HnswSearchCandidate, membername, ptr)
#define HnswGetValue(base, element) PointerGetDatum(HnswPtrAccess(base, (element)->value))
#if PG_VERSION_NUM < 140005
@@ -163,7 +160,7 @@ typedef struct HnswSearchCandidate
pairingheap_node c_node;
pairingheap_node w_node;
HnswElementPtr element;
double distance;
float distance;
} HnswSearchCandidate;
/* HNSW index options */
@@ -188,8 +185,8 @@ typedef struct HnswGraph
/* Allocations state */
LWLock allocatorLock;
Size memoryUsed;
Size memoryTotal;
long memoryUsed;
long memoryTotal;
/* Flushed state */
LWLock flushLock;
@@ -240,18 +237,6 @@ typedef struct HnswTypeInfo
void (*checkValue) (Pointer v);
} HnswTypeInfo;
typedef struct HnswSupport
{
FmgrInfo *procinfo;
FmgrInfo *normprocinfo;
Oid collation;
} HnswSupport;
typedef struct HnswQuery
{
Datum value;
} HnswQuery;
typedef struct HnswBuildState
{
/* Info */
@@ -271,7 +256,9 @@ typedef struct HnswBuildState
double reltuples;
/* Support functions */
HnswSupport support;
FmgrInfo *procinfo;
FmgrInfo *normprocinfo;
Oid collation;
/* Variables */
HnswGraph graphData;
@@ -338,19 +325,6 @@ typedef struct HnswNeighborTupleData
typedef HnswNeighborTupleData * HnswNeighborTuple;
typedef union
{
struct pointerhash_hash *pointers;
struct offsethash_hash *offsets;
struct tidhash_hash *tids;
} visited_hash;
typedef union
{
HnswElement element;
ItemPointerData indextid;
} HnswUnvisited;
typedef struct HnswScanOpaqueData
{
const HnswTypeInfo *typeInfo;
@@ -359,7 +333,9 @@ typedef struct HnswScanOpaqueData
MemoryContext tmpCtx;
/* Support functions */
HnswSupport support;
FmgrInfo *procinfo;
FmgrInfo *normprocinfo;
Oid collation;
} HnswScanOpaqueData;
typedef HnswScanOpaqueData * HnswScanOpaque;
@@ -377,7 +353,8 @@ typedef struct HnswVacuumState
int efConstruction;
/* Support functions */
HnswSupport support;
FmgrInfo *procinfo;
Oid collation;
/* Variables */
struct tidhash_hash *deleted;
@@ -393,33 +370,30 @@ typedef struct HnswVacuumState
int HnswGetM(Relation index);
int HnswGetEfConstruction(Relation index);
FmgrInfo *HnswOptionalProcInfo(Relation index, uint16 procnum);
void HnswInitSupport(HnswSupport * support, Relation index);
Datum HnswNormValue(const HnswTypeInfo * typeInfo, Oid collation, Datum value);
bool HnswCheckNorm(HnswSupport * support, Datum value);
bool HnswCheckNorm(FmgrInfo *procinfo, Oid collation, Datum value);
Buffer HnswNewBuffer(Relation index, ForkNumber forkNum);
void HnswInitPage(Buffer buf, Page page);
void HnswInit(void);
List *HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation index, HnswSupport * support, int m, bool inserting, HnswElement skipElement);
List *HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement);
HnswElement HnswGetEntryPoint(Relation index);
void HnswGetMetaPageInfo(Relation index, int *m, HnswElement * entryPoint);
void *HnswAlloc(HnswAllocator * allocator, Size size);
HnswElement HnswInitElement(char *base, ItemPointer tid, int m, double ml, int maxLevel, HnswAllocator * alloc);
HnswElement HnswInitElementFromBlock(BlockNumber blkno, OffsetNumber offno);
void HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, HnswSupport * support, int m, int efConstruction, bool existing);
HnswSearchCandidate *HnswEntryCandidate(char *base, HnswElement em, HnswQuery * q, Relation rel, HnswSupport * support, bool loadVec);
void HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, FmgrInfo *procinfo, Oid collation, int m, int efConstruction, bool existing);
HnswSearchCandidate *HnswEntryCandidate(char *base, HnswElement em, Datum q, Relation rel, FmgrInfo *procinfo, Oid collation, bool loadVec);
void HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, BlockNumber insertPage, ForkNumber forkNum, bool building);
void HnswSetNeighborTuple(char *base, HnswNeighborTuple ntup, HnswElement e, int m);
void HnswAddHeapTid(HnswElement element, ItemPointer heaptid);
HnswNeighborArray *HnswInitNeighborArray(int lm, HnswAllocator * allocator);
void HnswInitNeighbors(char *base, HnswElement element, int m, HnswAllocator * alloc);
bool HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPointer heaptid, bool building);
void HnswUpdateNeighborsOnDisk(Relation index, HnswSupport * support, HnswElement e, int m, bool checkExisting, bool building);
bool HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, bool building);
void HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building);
void HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec);
void HnswLoadElement(HnswElement element, double *distance, HnswQuery * q, Relation index, HnswSupport * support, bool loadVec, double *maxDistance);
bool HnswFormIndexValue(Datum *out, Datum *values, bool *isnull, const HnswTypeInfo * typeInfo, HnswSupport * support);
void HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec, float *maxDistance);
void HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element);
void HnswUpdateConnection(char *base, HnswNeighborArray * neighbors, HnswElement newElement, float distance, int lm, int *updateIdx, Relation index, HnswSupport * support);
bool HnswLoadNeighborTids(HnswElement element, ItemPointerData *indextids, Relation index, int m, int lm, int lc);
void HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation);
void HnswLoadNeighbors(HnswElement element, Relation index, int m);
void HnswInitLockTranche(void);
const HnswTypeInfo *HnswGetTypeInfo(Relation index);
PGDLLEXPORT void HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc);

View File

@@ -366,7 +366,7 @@ AddElementInMemory(char *base, HnswGraph * graph, HnswElement element)
* Update neighbors
*/
static void
UpdateNeighborsInMemory(char *base, HnswSupport * support, HnswElement e, int m)
UpdateNeighborsInMemory(char *base, FmgrInfo *procinfo, Oid collation, HnswElement e, int m)
{
for (int lc = e->level; lc >= 0; lc--)
{
@@ -388,7 +388,7 @@ UpdateNeighborsInMemory(char *base, HnswSupport * support, HnswElement e, int m)
Assert(neighborElement);
LWLockAcquire(&neighborElement->lock, LW_EXCLUSIVE);
HnswUpdateConnection(base, HnswGetNeighbors(base, neighborElement, lc), e, hc->distance, lm, NULL, NULL, support);
HnswUpdateConnection(base, e, hc, lm, lc, NULL, NULL, procinfo, collation);
LWLockRelease(&neighborElement->lock);
}
}
@@ -398,7 +398,7 @@ UpdateNeighborsInMemory(char *base, HnswSupport * support, HnswElement e, int m)
* Update graph in memory
*/
static void
UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, HnswBuildState * buildstate)
UpdateGraphInMemory(FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement entryPoint, HnswBuildState * buildstate)
{
HnswGraph *graph = buildstate->graph;
char *base = buildstate->hnswarea;
@@ -411,7 +411,7 @@ UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, int efCon
AddElementInMemory(base, graph, element);
/* Update neighbors */
UpdateNeighborsInMemory(base, support, element, m);
UpdateNeighborsInMemory(base, procinfo, collation, element, m);
/* Update entry point if needed (already have lock) */
if (entryPoint == NULL || element->level > entryPoint->level)
@@ -424,8 +424,9 @@ UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, int efCon
static void
InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
{
FmgrInfo *procinfo = buildstate->procinfo;
Oid collation = buildstate->collation;
HnswGraph *graph = buildstate->graph;
HnswSupport *support = &buildstate->support;
HnswElement entryPoint;
LWLock *entryLock = &graph->entryLock;
LWLock *entryWaitLock = &graph->entryWaitLock;
@@ -457,10 +458,10 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
}
/* Find neighbors for element */
HnswFindElementNeighbors(base, element, entryPoint, NULL, support, m, efConstruction, false);
HnswFindElementNeighbors(base, element, entryPoint, NULL, procinfo, collation, m, efConstruction, false);
/* Update graph in memory */
UpdateGraphInMemory(support, element, m, efConstruction, entryPoint, buildstate);
UpdateGraphInMemory(procinfo, collation, element, m, efConstruction, entryPoint, buildstate);
/* Release entry lock */
LWLockRelease(entryLock);
@@ -472,19 +473,30 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
static bool
InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, HnswBuildState * buildstate)
{
const HnswTypeInfo *typeInfo = buildstate->typeInfo;
HnswGraph *graph = buildstate->graph;
HnswElement element;
HnswAllocator *allocator = &buildstate->allocator;
HnswSupport *support = &buildstate->support;
Size valueSize;
Pointer valuePtr;
LWLock *flushLock = &graph->flushLock;
char *base = buildstate->hnswarea;
Datum value;
/* Form index value */
if (!HnswFormIndexValue(&value, values, isnull, buildstate->typeInfo, support))
return false;
/* Detoast once for all calls */
Datum value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
/* Check value */
if (typeInfo->checkValue != NULL)
typeInfo->checkValue(DatumGetPointer(value));
/* Normalize if needed */
if (buildstate->normprocinfo != NULL)
{
if (!HnswCheckNorm(buildstate->normprocinfo, buildstate->collation, value))
return false;
value = HnswNormValue(typeInfo, buildstate->collation, value);
}
/* Get datum size */
valueSize = VARSIZE_ANY(DatumGetPointer(value));
@@ -497,7 +509,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
{
LWLockRelease(flushLock);
return HnswInsertTupleOnDisk(index, support, value, heaptid, true);
return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, true);
}
/*
@@ -529,7 +541,7 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid, Hn
LWLockRelease(flushLock);
return HnswInsertTupleOnDisk(index, support, value, heaptid, true);
return HnswInsertTupleOnDisk(index, value, values, isnull, heaptid, true);
}
/* Ok, we can proceed to allocate the element */
@@ -595,7 +607,7 @@ BuildCallback(Relation index, ItemPointer tid, Datum *values,
* Initialize the graph
*/
static void
InitGraph(HnswGraph * graph, char *base, Size memoryTotal)
InitGraph(HnswGraph * graph, char *base, long memoryTotal)
{
/* Initialize the lock tranche if needed */
HnswInitLockTranche();
@@ -692,9 +704,11 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
buildstate->indtuples = 0;
/* Get support functions */
HnswInitSupport(&buildstate->support, index);
buildstate->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
buildstate->normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
buildstate->collation = index->rd_indcollation[0];
InitGraph(&buildstate->graphData, NULL, (Size) maintenance_work_mem * 1024L);
InitGraph(&buildstate->graphData, NULL, maintenance_work_mem * 1024L);
buildstate->graph = &buildstate->graphData;
buildstate->ml = HnswGetMl(buildstate->m);
buildstate->maxLevel = HnswGetMaxLevel(buildstate->m);

View File

@@ -334,107 +334,6 @@ AddElementOnDisk(Relation index, HnswElement e, int m, BlockNumber insertPage, B
*updatedInsertPage = newInsertPage;
}
/*
* Load neighbors
*/
static HnswNeighborArray *
HnswLoadNeighbors(HnswElement element, Relation index, int m, int lm, int lc)
{
char *base = NULL;
HnswNeighborArray *neighbors = HnswInitNeighborArray(lm, NULL);
ItemPointerData indextids[HNSW_MAX_M * 2];
if (!HnswLoadNeighborTids(element, indextids, index, m, lm, lc))
return neighbors;
for (int i = 0; i < lm; i++)
{
ItemPointer indextid = &indextids[i];
HnswElement e;
HnswCandidate *hc;
if (!ItemPointerIsValid(indextid))
break;
e = HnswInitElementFromBlock(ItemPointerGetBlockNumber(indextid), ItemPointerGetOffsetNumber(indextid));
hc = &neighbors->items[neighbors->length++];
HnswPtrStore(base, hc->element, e);
}
return neighbors;
}
/*
* Load elements for insert
*/
static void
LoadElementsForInsert(HnswNeighborArray * neighbors, HnswQuery * q, int *idx, Relation index, HnswSupport * support)
{
char *base = NULL;
for (int i = 0; i < neighbors->length; i++)
{
HnswCandidate *hc = &neighbors->items[i];
HnswElement element = HnswPtrAccess(base, hc->element);
double distance;
HnswLoadElement(element, &distance, q, index, support, true, NULL);
hc->distance = distance;
/* Prune element if being deleted */
if (element->heaptidsLength == 0)
{
*idx = i;
break;
}
}
}
/*
* Get update index
*/
static int
GetUpdateIndex(HnswElement element, HnswElement newElement, float distance, int m, int lm, int lc, Relation index, HnswSupport * support, MemoryContext updateCtx)
{
char *base = NULL;
int idx = -1;
HnswNeighborArray *neighbors;
MemoryContext oldCtx = MemoryContextSwitchTo(updateCtx);
/*
* Get latest neighbors since they may have changed. Do not lock yet since
* selecting neighbors can take time. Could use optimistic locking to
* retry if another update occurs before getting exclusive lock.
*/
neighbors = HnswLoadNeighbors(element, index, m, lm, lc);
/*
* Could improve performance for vacuuming by checking neighbors against
* list of elements being deleted to find index. It's important to exclude
* already deleted elements for this since they can be replaced at any
* time.
*/
if (neighbors->length < lm)
idx = -2;
else
{
HnswQuery q;
q.value = HnswGetValue(base, element);
LoadElementsForInsert(neighbors, &q, &idx, index, support);
if (idx == -1)
HnswUpdateConnection(base, neighbors, newElement, distance, lm, &idx, index, support);
}
MemoryContextSwitchTo(oldCtx);
MemoryContextReset(updateCtx);
return idx;
}
/*
* Check if connection already exists
*/
@@ -455,94 +354,14 @@ ConnectionExists(HnswElement e, HnswNeighborTuple ntup, int startIdx, int lm)
return false;
}
/*
* Update neighbor
*/
static void
UpdateNeighborOnDisk(HnswElement element, HnswElement newElement, int idx, int m, int lm, int lc, Relation index, bool checkExisting, bool building)
{
Buffer buf;
Page page;
GenericXLogState *state;
HnswNeighborTuple ntup;
int startIdx;
OffsetNumber offno = element->neighborOffno;
/* Register page */
buf = ReadBuffer(index, element->neighborPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
/* Get tuple */
ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, offno));
/* Calculate index for update */
startIdx = (element->level - lc) * m;
/* Check for existing connection */
if (checkExisting && ConnectionExists(newElement, ntup, startIdx, lm))
idx = -1;
else if (idx == -2)
{
/* Find free offset if still exists */
/* TODO Retry updating connections if not */
for (int j = 0; j < lm; j++)
{
if (!ItemPointerIsValid(&ntup->indextids[startIdx + j]))
{
idx = startIdx + j;
break;
}
}
}
else
idx += startIdx;
/* Make robust to issues */
if (idx >= 0 && idx < ntup->count)
{
ItemPointer indextid = &ntup->indextids[idx];
/* Update neighbor on the buffer */
ItemPointerSet(indextid, newElement->blkno, newElement->offno);
/* Commit */
if (building)
MarkBufferDirty(buf);
else
GenericXLogFinish(state);
}
else if (!building)
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
}
/*
* Update neighbors
*/
void
HnswUpdateNeighborsOnDisk(Relation index, HnswSupport * support, HnswElement e, int m, bool checkExisting, bool building)
HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building)
{
char *base = NULL;
/* Use separate memory context to improve performance for larger vectors */
MemoryContext updateCtx = GenerationContextCreate(CurrentMemoryContext,
"Hnsw insert update context",
#if PG_VERSION_NUM >= 150000
128 * 1024, 128 * 1024,
#endif
128 * 1024);
for (int lc = e->level; lc >= 0; lc--)
{
int lm = HnswGetLayerM(m, lc);
@@ -551,20 +370,96 @@ HnswUpdateNeighborsOnDisk(Relation index, HnswSupport * support, HnswElement e,
for (int i = 0; i < neighbors->length; i++)
{
HnswCandidate *hc = &neighbors->items[i];
Buffer buf;
Page page;
GenericXLogState *state;
HnswNeighborTuple ntup;
int idx = -1;
int startIdx;
HnswElement neighborElement = HnswPtrAccess(base, hc->element);
int idx;
OffsetNumber offno = neighborElement->neighborOffno;
idx = GetUpdateIndex(neighborElement, e, hc->distance, m, lm, lc, index, support, updateCtx);
/*
* Get latest neighbors since they may have changed. Do not lock
* yet since selecting neighbors can take time. Could use
* optimistic locking to retry if another update occurs before
* getting exclusive lock.
*/
HnswLoadNeighbors(neighborElement, index, m);
/*
* Could improve performance for vacuuming by checking neighbors
* against list of elements being deleted to find index. It's
* important to exclude already deleted elements for this since
* they can be replaced at any time.
*/
/* Select neighbors */
HnswUpdateConnection(NULL, e, hc, lm, lc, &idx, index, procinfo, collation);
/* New element was not selected as a neighbor */
if (idx == -1)
continue;
UpdateNeighborOnDisk(neighborElement, e, idx, m, lm, lc, index, checkExisting, building);
/* Register page */
buf = ReadBuffer(index, neighborElement->neighborPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (building)
{
state = NULL;
page = BufferGetPage(buf);
}
else
{
state = GenericXLogStart(index);
page = GenericXLogRegisterBuffer(state, buf, 0);
}
/* Get tuple */
ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, offno));
/* Calculate index for update */
startIdx = (neighborElement->level - lc) * m;
/* Check for existing connection */
if (checkExisting && ConnectionExists(e, ntup, startIdx, lm))
idx = -1;
else if (idx == -2)
{
/* Find free offset if still exists */
/* TODO Retry updating connections if not */
for (int j = 0; j < lm; j++)
{
if (!ItemPointerIsValid(&ntup->indextids[startIdx + j]))
{
idx = startIdx + j;
break;
}
}
}
else
idx += startIdx;
/* Make robust to issues */
if (idx >= 0 && idx < ntup->count)
{
ItemPointer indextid = &ntup->indextids[idx];
/* Update neighbor on the buffer */
ItemPointerSet(indextid, e->blkno, e->offno);
/* Commit */
if (building)
MarkBufferDirty(buf);
else
GenericXLogFinish(state);
}
else if (!building)
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
}
}
MemoryContextDelete(updateCtx);
}
/*
@@ -654,7 +549,7 @@ FindDuplicateOnDisk(Relation index, HnswElement element, bool building)
* Update graph on disk
*/
static void
UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, bool building)
UpdateGraphOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement entryPoint, bool building)
{
BlockNumber newInsertPage = InvalidBlockNumber;
@@ -670,7 +565,7 @@ UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, in
HnswUpdateMetaPage(index, 0, NULL, newInsertPage, MAIN_FORKNUM, building);
/* Update neighbors */
HnswUpdateNeighborsOnDisk(index, support, element, m, false, building);
HnswUpdateNeighborsOnDisk(index, procinfo, collation, element, m, false, building);
/* Update entry point if needed */
if (entryPoint == NULL || element->level > entryPoint->level)
@@ -681,12 +576,14 @@ UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, in
* Insert a tuple into the index
*/
bool
HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPointer heaptid, bool building)
HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull, ItemPointer heap_tid, bool building)
{
HnswElement entryPoint;
HnswElement element;
int m;
int efConstruction = HnswGetEfConstruction(index);
FmgrInfo *procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
Oid collation = index->rd_indcollation[0];
LOCKMODE lockmode = ShareLock;
char *base = NULL;
@@ -701,7 +598,7 @@ HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPo
HnswGetMetaPageInfo(index, &m, &entryPoint);
/* Create an element */
element = HnswInitElement(base, heaptid, m, HnswGetMl(m), HnswGetMaxLevel(m), NULL);
element = HnswInitElement(base, heap_tid, m, HnswGetMl(m), HnswGetMaxLevel(m), NULL);
HnswPtrStore(base, element->value, DatumGetPointer(value));
/* Prevent concurrent inserts when likely updating entry point */
@@ -719,10 +616,10 @@ HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPo
}
/* Find neighbors for element */
HnswFindElementNeighbors(base, element, entryPoint, index, support, m, efConstruction, false);
HnswFindElementNeighbors(base, element, entryPoint, index, procinfo, collation, m, efConstruction, false);
/* Update graph on disk */
UpdateGraphOnDisk(index, support, element, m, efConstruction, entryPoint, building);
UpdateGraphOnDisk(index, procinfo, collation, element, m, efConstruction, entryPoint, building);
/* Release lock */
UnlockPage(index, HNSW_UPDATE_LOCK, lockmode);
@@ -734,19 +631,31 @@ HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPo
* Insert a tuple into the index
*/
static void
HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heaptid)
HnswInsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid)
{
Datum value;
const HnswTypeInfo *typeInfo = HnswGetTypeInfo(index);
HnswSupport support;
FmgrInfo *normprocinfo;
Oid collation = index->rd_indcollation[0];
HnswInitSupport(&support, index);
/* Detoast once for all calls */
value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
/* Form index value */
if (!HnswFormIndexValue(&value, values, isnull, typeInfo, &support))
return;
/* Check value */
if (typeInfo->checkValue != NULL)
typeInfo->checkValue(DatumGetPointer(value));
HnswInsertTupleOnDisk(index, &support, value, heaptid, false);
/* Normalize if needed */
normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
if (normprocinfo != NULL)
{
if (!HnswCheckNorm(normprocinfo, collation, value))
return;
value = HnswNormValue(typeInfo, collation, value);
}
HnswInsertTupleOnDisk(index, value, values, isnull, heap_tid, false);
}
/*

View File

@@ -11,19 +11,17 @@
* Algorithm 5 from paper
*/
static List *
GetScanItems(IndexScanDesc scan, Datum value)
GetScanItems(IndexScanDesc scan, Datum q)
{
HnswScanOpaque so = (HnswScanOpaque) scan->opaque;
Relation index = scan->indexRelation;
HnswSupport *support = &so->support;
FmgrInfo *procinfo = so->procinfo;
Oid collation = so->collation;
List *ep;
List *w;
int m;
HnswElement entryPoint;
char *base = NULL;
HnswQuery q;
q.value = value;
/* Get m and entry point */
HnswGetMetaPageInfo(index, &m, &entryPoint);
@@ -31,15 +29,15 @@ GetScanItems(IndexScanDesc scan, Datum value)
if (entryPoint == NULL)
return NIL;
ep = list_make1(HnswEntryCandidate(base, entryPoint, &q, index, support, false));
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, procinfo, collation, false));
for (int lc = entryPoint->level; lc >= 1; lc--)
{
w = HnswSearchLayer(base, &q, ep, 1, lc, index, support, m, false, NULL);
w = HnswSearchLayer(base, q, ep, 1, lc, index, procinfo, collation, m, false, NULL);
ep = w;
}
return HnswSearchLayer(base, &q, ep, hnsw_ef_search, 0, index, support, m, false, NULL);
return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
}
/*
@@ -62,8 +60,8 @@ GetScanValue(IndexScanDesc scan)
Assert(!VARATT_IS_EXTENDED(DatumGetPointer(value)));
/* Normalize if needed */
if (so->support.normprocinfo != NULL)
value = HnswNormValue(so->typeInfo, so->support.collation, value);
if (so->normprocinfo != NULL)
value = HnswNormValue(so->typeInfo, so->collation, value);
}
return value;
@@ -88,7 +86,9 @@ hnswbeginscan(Relation index, int nkeys, int norderbys)
ALLOCSET_DEFAULT_SIZES);
/* Set support functions */
HnswInitSupport(&so->support, index);
so->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
so->normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
so->collation = index->rd_indcollation[0];
scan->opaque = so;
@@ -168,8 +168,8 @@ hnswgettuple(IndexScanDesc scan, ScanDirection dir)
while (list_length(so->w) > 0)
{
char *base = NULL;
HnswSearchCandidate *sc = llast(so->w);
HnswElement element = HnswPtrAccess(base, sc->element);
HnswSearchCandidate *hc = llast(so->w);
HnswElement element = HnswPtrAccess(base, hc->element);
ItemPointer heaptid;
/* Move to next element if no valid heap TIDs */

View File

@@ -100,6 +100,19 @@ hash_offset(Size offset)
#define SH_DEFINE
#include "lib/simplehash.h"
typedef union
{
pointerhash_hash *pointers;
offsethash_hash *offsets;
tidhash_hash *tids;
} visited_hash;
typedef union
{
HnswElement element;
ItemPointerData indextid;
} HnswUnvisited;
/*
* Get the max number of connections in an upper layer for each element in the index
*/
@@ -140,23 +153,15 @@ HnswOptionalProcInfo(Relation index, uint16 procnum)
return index_getprocinfo(index, 1, procnum);
}
/*
* Init support functions
*/
void
HnswInitSupport(HnswSupport * support, Relation index)
{
support->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
support->collation = index->rd_indcollation[0];
support->normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
}
/*
* Normalize value
*/
Datum
HnswNormValue(const HnswTypeInfo * typeInfo, Oid collation, Datum value)
{
if (!typeInfo->normalize)
return value;
return DirectFunctionCall1Coll(typeInfo->normalize, collation, value);
}
@@ -164,9 +169,9 @@ HnswNormValue(const HnswTypeInfo * typeInfo, Oid collation, Datum value)
* Check if non-zero norm
*/
bool
HnswCheckNorm(HnswSupport * support, Datum value)
HnswCheckNorm(FmgrInfo *procinfo, Oid collation, Datum value)
{
return DatumGetFloat8(FunctionCall1Coll(support->normprocinfo, support->collation, value)) > 0;
return DatumGetFloat8(FunctionCall1Coll(procinfo, collation, value)) > 0;
}
/*
@@ -195,7 +200,7 @@ HnswInitPage(Buffer buf, Page page)
/*
* Allocate a neighbor array
*/
HnswNeighborArray *
static HnswNeighborArray *
HnswInitNeighborArray(int lm, HnswAllocator * allocator)
{
HnswNeighborArray *a = HnswAlloc(allocator, HNSW_NEIGHBOR_ARRAY_SIZE(lm));
@@ -392,33 +397,6 @@ HnswUpdateMetaPage(Relation index, int updateEntry, HnswElement entryPoint, Bloc
UnlockReleaseBuffer(buf);
}
/*
* Form index value
*/
bool
HnswFormIndexValue(Datum *out, Datum *values, bool *isnull, const HnswTypeInfo * typeInfo, HnswSupport * support)
{
/* Detoast once for all calls */
Datum value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
/* Check value */
if (typeInfo->checkValue != NULL)
typeInfo->checkValue(DatumGetPointer(value));
/* Normalize if needed */
if (support->normprocinfo != NULL)
{
if (!HnswCheckNorm(support, value))
return false;
value = HnswNormValue(typeInfo, support->collation, value);
}
*out = value;
return true;
}
/*
* Set element tuple, except for neighbor info
*/
@@ -474,6 +452,69 @@ HnswSetNeighborTuple(char *base, HnswNeighborTuple ntup, HnswElement e, int m)
ntup->count = idx;
}
/*
* Load neighbors from page
*/
static void
LoadNeighborsFromPage(HnswElement element, Relation index, Page page, int m)
{
char *base = NULL;
HnswNeighborTuple ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, element->neighborOffno));
int neighborCount = (element->level + 2) * m;
Assert(HnswIsNeighborTuple(ntup));
HnswInitNeighbors(base, element, m, NULL);
/* Ensure expected neighbors */
if (ntup->count != neighborCount)
return;
for (int i = 0; i < neighborCount; i++)
{
HnswElement e;
int level;
HnswCandidate *hc;
ItemPointer indextid;
HnswNeighborArray *neighbors;
indextid = &ntup->indextids[i];
if (!ItemPointerIsValid(indextid))
continue;
e = HnswInitElementFromBlock(ItemPointerGetBlockNumber(indextid), ItemPointerGetOffsetNumber(indextid));
/* Calculate level based on offset */
level = element->level - i / m;
if (level < 0)
level = 0;
neighbors = HnswGetNeighbors(base, element, level);
hc = &neighbors->items[neighbors->length++];
HnswPtrStore(base, hc->element, e);
}
}
/*
* Load neighbors
*/
void
HnswLoadNeighbors(HnswElement element, Relation index, int m)
{
Buffer buf;
Page page;
buf = ReadBuffer(index, element->neighborPage);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
LoadNeighborsFromPage(element, index, page, m);
UnlockReleaseBuffer(buf);
}
/*
* Load an element from a tuple
*/
@@ -507,20 +548,11 @@ HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHe
}
}
/*
* Calculate the distance between values
*/
static inline double
HnswGetDistance(Datum a, Datum b, HnswSupport * support)
{
return DatumGetFloat8(FunctionCall2Coll(support->procinfo, support->collation, a, b));
}
/*
* Load an element and optionally get its distance from q
*/
static void
HnswLoadElementImpl(BlockNumber blkno, OffsetNumber offno, double *distance, HnswQuery * q, Relation index, HnswSupport * support, bool loadVec, double *maxDistance, HnswElement * element)
HnswLoadElementImpl(BlockNumber blkno, OffsetNumber offno, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec, float *maxDistance, HnswElement * element)
{
Buffer buf;
Page page;
@@ -538,10 +570,10 @@ HnswLoadElementImpl(BlockNumber blkno, OffsetNumber offno, double *distance, Hns
/* Calculate distance */
if (distance != NULL)
{
if (DatumGetPointer(q->value) == NULL)
if (DatumGetPointer(*q) == NULL)
*distance = 0;
else
*distance = HnswGetDistance(q->value, PointerGetDatum(&etup->data), support);
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
}
/* Load element */
@@ -560,39 +592,41 @@ HnswLoadElementImpl(BlockNumber blkno, OffsetNumber offno, double *distance, Hns
* Load an element and optionally get its distance from q
*/
void
HnswLoadElement(HnswElement element, double *distance, HnswQuery * q, Relation index, HnswSupport * support, bool loadVec, double *maxDistance)
HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec, float *maxDistance)
{
HnswLoadElementImpl(element->blkno, element->offno, distance, q, index, support, loadVec, maxDistance, &element);
HnswLoadElementImpl(element->blkno, element->offno, distance, q, index, procinfo, collation, loadVec, maxDistance, &element);
}
/*
* Get the distance for an element
*/
static double
GetElementDistance(char *base, HnswElement element, HnswQuery * q, HnswSupport * support)
static float
GetElementDistance(char *base, HnswElement element, Datum q, FmgrInfo *procinfo, Oid collation)
{
Datum value = HnswGetValue(base, element);
return HnswGetDistance(q->value, value, support);
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, q, value));
}
/*
* Create a candidate for the entry point
*/
HnswSearchCandidate *
HnswEntryCandidate(char *base, HnswElement entryPoint, HnswQuery * q, Relation index, HnswSupport * support, bool loadVec)
HnswEntryCandidate(char *base, HnswElement entryPoint, Datum q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec)
{
HnswSearchCandidate *sc = palloc(sizeof(HnswSearchCandidate));
bool inMemory = index == NULL;
HnswSearchCandidate *hc = palloc(sizeof(HnswSearchCandidate));
HnswPtrStore(base, sc->element, entryPoint);
if (inMemory)
sc->distance = GetElementDistance(base, entryPoint, q, support);
HnswPtrStore(base, hc->element, entryPoint);
if (index == NULL)
hc->distance = GetElementDistance(base, entryPoint, q, procinfo, collation);
else
HnswLoadElement(entryPoint, &sc->distance, q, index, support, loadVec, NULL);
return sc;
HnswLoadElement(entryPoint, &hc->distance, &q, index, procinfo, collation, loadVec, NULL);
return hc;
}
#define HnswGetSearchCandidate(membername, ptr) pairingheap_container(HnswSearchCandidate, membername, ptr)
#define HnswGetSearchCandidateConst(membername, ptr) pairingheap_const_container(HnswSearchCandidate, membername, ptr)
/*
* Compare candidate distances
*/
@@ -627,9 +661,9 @@ CompareFurthestCandidates(const pairingheap_node *a, const pairingheap_node *b,
* Init visited
*/
static inline void
InitVisited(char *base, visited_hash * v, bool inMemory, int ef, int m)
InitVisited(char *base, visited_hash * v, Relation index, int ef, int m)
{
if (!inMemory)
if (index != NULL)
v->tids = tidhash_create(CurrentMemoryContext, ef * m * 2, NULL);
else if (base != NULL)
v->offsets = offsethash_create(CurrentMemoryContext, ef * m * 2, NULL);
@@ -641,9 +675,9 @@ InitVisited(char *base, visited_hash * v, bool inMemory, int ef, int m)
* Add to visited
*/
static inline void
AddToVisited(char *base, visited_hash * v, HnswElementPtr elementPtr, bool inMemory, bool *found)
AddToVisited(char *base, visited_hash * v, HnswElementPtr elementPtr, Relation index, bool *found)
{
if (!inMemory)
if (index != NULL)
{
HnswElement element = HnswPtrAccess(base, elementPtr);
ItemPointerData indextid;
@@ -704,57 +738,38 @@ HnswLoadUnvisitedFromMemory(char *base, HnswElement element, HnswUnvisited * unv
HnswCandidate *hc = &localNeighborhood->items[i];
bool found;
AddToVisited(base, v, hc->element, true, &found);
AddToVisited(base, v, hc->element, NULL, &found);
if (!found)
unvisited[(*unvisitedLength)++].element = HnswPtrAccess(base, hc->element);
}
}
/*
* Load neighbor index TIDs
*/
bool
HnswLoadNeighborTids(HnswElement element, ItemPointerData *indextids, Relation index, int m, int lm, int lc)
{
Buffer buf;
Page page;
HnswNeighborTuple ntup;
int start;
buf = ReadBuffer(index, element->neighborPage);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, element->neighborOffno));
/* Ensure expected neighbors */
if (ntup->count != (element->level + 2) * m)
{
UnlockReleaseBuffer(buf);
return false;
}
/* Copy to minimize lock time */
start = (element->level - lc) * m;
memcpy(indextids, ntup->indextids + start, lm * sizeof(ItemPointerData));
UnlockReleaseBuffer(buf);
return true;
}
/*
* Load unvisited neighbors from disk
*/
static void
HnswLoadUnvisitedFromDisk(HnswElement element, HnswUnvisited * unvisited, int *unvisitedLength, visited_hash * v, Relation index, int m, int lm, int lc)
{
Buffer buf;
Page page;
HnswNeighborTuple ntup;
int start;
ItemPointerData indextids[HNSW_MAX_M * 2];
*unvisitedLength = 0;
buf = ReadBuffer(index, element->neighborPage);
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
if (!HnswLoadNeighborTids(element, indextids, index, m, lm, lc))
return;
ntup = (HnswNeighborTuple) PageGetItem(page, PageGetItemId(page, element->neighborOffno));
start = (element->level - lc) * m;
/* Copy to minimize lock time */
memcpy(&indextids, ntup->indextids + start, lm * sizeof(ItemPointerData));
UnlockReleaseBuffer(buf);
*unvisitedLength = 0;
for (int i = 0; i < lm; i++)
{
@@ -775,7 +790,7 @@ HnswLoadUnvisitedFromDisk(HnswElement element, HnswUnvisited * unvisited, int *u
* Algorithm 2 from paper
*/
List *
HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation index, HnswSupport * support, int m, bool inserting, HnswElement skipElement)
HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement)
{
List *w = NIL;
pairingheap *C = pairingheap_allocate(CompareNearestCandidates, NULL);
@@ -788,12 +803,11 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
int lm = HnswGetLayerM(m, lc);
HnswUnvisited *unvisited = palloc(lm * sizeof(HnswUnvisited));
int unvisitedLength;
bool inMemory = index == NULL;
InitVisited(base, &v, inMemory, ef, m);
InitVisited(base, &v, index, ef, m);
/* Create local memory for neighborhood if needed */
if (inMemory)
if (index == NULL)
{
neighborhoodSize = HNSW_NEIGHBOR_ARRAY_SIZE(lm);
localNeighborhood = palloc(neighborhoodSize);
@@ -802,20 +816,20 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
/* Add entry points to v, C, and W */
foreach(lc2, ep)
{
HnswSearchCandidate *sc = (HnswSearchCandidate *) lfirst(lc2);
HnswSearchCandidate *hc = (HnswSearchCandidate *) lfirst(lc2);
bool found;
AddToVisited(base, &v, sc->element, inMemory, &found);
AddToVisited(base, &v, hc->element, index, &found);
pairingheap_add(C, &sc->c_node);
pairingheap_add(W, &sc->w_node);
pairingheap_add(C, &hc->c_node);
pairingheap_add(W, &hc->w_node);
/*
* Do not count elements being deleted towards ef when vacuuming. It
* would be ideal to do this for inserts as well, but this could
* affect insert performance.
*/
if (CountElement(skipElement, HnswPtrAccess(base, sc->element)))
if (CountElement(skipElement, HnswPtrAccess(base, hc->element)))
wlen++;
}
@@ -830,7 +844,7 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
cElement = HnswPtrAccess(base, c->element);
if (inMemory)
if (index == NULL)
HnswLoadUnvisitedFromMemory(base, cElement, unvisited, &unvisitedLength, &v, lc, localNeighborhood, neighborhoodSize);
else
HnswLoadUnvisitedFromDisk(cElement, unvisited, &unvisitedLength, &v, index, m, lm, lc);
@@ -839,15 +853,15 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
{
HnswElement eElement;
HnswSearchCandidate *e;
double eDistance;
float eDistance;
bool alwaysAdd = wlen < ef;
f = HnswGetSearchCandidate(w_node, pairingheap_first(W));
if (inMemory)
if (index == NULL)
{
eElement = unvisited[i].element;
eDistance = GetElementDistance(base, eElement, q, support);
eDistance = GetElementDistance(base, eElement, q, procinfo, collation);
}
else
{
@@ -857,7 +871,7 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
/* Avoid any allocations if not adding */
eElement = NULL;
HnswLoadElementImpl(blkno, offno, &eDistance, q, index, support, inserting, alwaysAdd ? NULL : &f->distance, &eElement);
HnswLoadElementImpl(blkno, offno, &eDistance, &q, index, procinfo, collation, inserting, alwaysAdd ? NULL : &f->distance, &eElement);
if (eElement == NULL)
continue;
@@ -898,9 +912,9 @@ HnswSearchLayer(char *base, HnswQuery * q, List *ep, int ef, int lc, Relation in
/* Add each element of W to w */
while (!pairingheap_is_empty(W))
{
HnswSearchCandidate *sc = HnswGetSearchCandidate(w_node, pairingheap_remove_first(W));
HnswSearchCandidate *hc = HnswGetSearchCandidate(w_node, pairingheap_remove_first(W));
w = lappend(w, sc);
w = lappend(w, hc);
}
return w;
@@ -954,22 +968,32 @@ CompareCandidateDistancesOffset(const ListCell *a, const ListCell *b)
return 0;
}
/*
* Calculate the distance between elements
*/
static float
HnswGetDistance(char *base, HnswElement a, HnswElement b, FmgrInfo *procinfo, Oid collation)
{
Datum aValue = HnswGetValue(base, a);
Datum bValue = HnswGetValue(base, b);
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, aValue, bValue));
}
/*
* Check if an element is closer to q than any element from R
*/
static bool
CheckElementCloser(char *base, HnswCandidate * e, List *r, HnswSupport * support)
CheckElementCloser(char *base, HnswCandidate * e, List *r, FmgrInfo *procinfo, Oid collation)
{
HnswElement eElement = HnswPtrAccess(base, e->element);
Datum eValue = HnswGetValue(base, eElement);
ListCell *lc2;
foreach(lc2, r)
{
HnswCandidate *ri = lfirst(lc2);
HnswElement riElement = HnswPtrAccess(base, ri->element);
Datum riValue = HnswGetValue(base, riElement);
float distance = HnswGetDistance(eValue, riValue, support);
float distance = HnswGetDistance(base, eElement, riElement, procinfo, collation);
if (distance <= e->distance)
return false;
@@ -982,14 +1006,15 @@ CheckElementCloser(char *base, HnswCandidate * e, List *r, HnswSupport * support
* Algorithm 4 from paper
*/
static List *
SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closerSet, HnswCandidate * newCandidate, HnswCandidate * *pruned, bool sortCandidates)
SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid collation, HnswElement e2, HnswCandidate * newCandidate, HnswCandidate * *pruned, bool sortCandidates)
{
List *r = NIL;
List *w = list_copy(c);
HnswCandidate **wd;
int wdlen = 0;
int wdoff = 0;
bool mustCalculate = !(*closerSet);
HnswNeighborArray *neighbors = HnswGetNeighbors(base, e2, lc);
bool mustCalculate = !neighbors->closerSet;
List *added = NIL;
bool removedAny = false;
@@ -1016,7 +1041,7 @@ SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closer
/* Use previous state of r and wd to skip work when possible */
if (mustCalculate)
e->closer = CheckElementCloser(base, e, r, support);
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
else if (list_length(added) > 0)
{
/* Keep Valgrind happy for in-memory, parallel builds */
@@ -1029,7 +1054,7 @@ SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closer
*/
if (e->closer)
{
e->closer = CheckElementCloser(base, e, added, support);
e->closer = CheckElementCloser(base, e, added, procinfo, collation);
if (!e->closer)
removedAny = true;
@@ -1042,7 +1067,7 @@ SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closer
*/
if (removedAny)
{
e->closer = CheckElementCloser(base, e, r, support);
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
if (e->closer)
added = lappend(added, e);
}
@@ -1050,7 +1075,7 @@ SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closer
}
else if (e == newCandidate)
{
e->closer = CheckElementCloser(base, e, r, support);
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
if (e->closer)
added = lappend(added, e);
}
@@ -1066,7 +1091,7 @@ SelectNeighbors(char *base, List *c, int lm, HnswSupport * support, bool *closer
}
/* Cached value can only be used in future if sorted deterministically */
*closerSet = sortCandidates;
neighbors->closerSet = sortCandidates;
/* Keep pruned connections */
while (wdoff < wdlen && list_length(r) < lm)
@@ -1101,16 +1126,18 @@ AddConnections(char *base, HnswElement element, List *neighbors, int lc)
* Update connections
*/
void
HnswUpdateConnection(char *base, HnswNeighborArray * neighbors, HnswElement newElement, float distance, int lm, int *updateIdx, Relation index, HnswSupport * support)
HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation)
{
HnswCandidate newHc;
HnswElement hce = HnswPtrAccess(base, hc->element);
HnswNeighborArray *currentNeighbors = HnswGetNeighbors(base, hce, lc);
HnswCandidate hc2;
HnswPtrStore(base, newHc.element, newElement);
newHc.distance = distance;
HnswPtrStore(base, hc2.element, element);
hc2.distance = hc->distance;
if (neighbors->length < lm)
if (currentNeighbors->length < lm)
{
neighbors->items[neighbors->length++] = newHc;
currentNeighbors->items[currentNeighbors->length++] = hc2;
/* Track update */
if (updateIdx != NULL)
@@ -1119,26 +1146,54 @@ HnswUpdateConnection(char *base, HnswNeighborArray * neighbors, HnswElement newE
else
{
/* Shrink connections */
List *c = NIL;
HnswCandidate *pruned = NULL;
/* Add candidates */
for (int i = 0; i < neighbors->length; i++)
c = lappend(c, &neighbors->items[i]);
c = lappend(c, &newHc);
/* Load elements on insert */
if (index != NULL)
{
Datum q = HnswGetValue(base, hce);
SelectNeighbors(base, c, lm, support, &neighbors->closerSet, &newHc, &pruned, true);
for (int i = 0; i < currentNeighbors->length; i++)
{
HnswCandidate *hc3 = &currentNeighbors->items[i];
HnswElement hc3Element = HnswPtrAccess(base, hc3->element);
if (HnswPtrIsNull(base, hc3Element->value))
HnswLoadElement(hc3Element, &hc3->distance, &q, index, procinfo, collation, true, NULL);
else
hc3->distance = GetElementDistance(base, hc3Element, q, procinfo, collation);
/* Prune element if being deleted */
if (hc3Element->heaptidsLength == 0)
{
pruned = &currentNeighbors->items[i];
break;
}
}
}
/* Should not happen */
if (pruned == NULL)
return;
{
List *c = NIL;
/* Add candidates */
for (int i = 0; i < currentNeighbors->length; i++)
c = lappend(c, &currentNeighbors->items[i]);
c = lappend(c, &hc2);
SelectNeighbors(base, c, lm, lc, procinfo, collation, hce, &hc2, &pruned, true);
/* Should not happen */
if (pruned == NULL)
return;
}
/* Find and replace the pruned element */
for (int i = 0; i < neighbors->length; i++)
for (int i = 0; i < currentNeighbors->length; i++)
{
if (HnswPtrEqual(base, neighbors->items[i].element, pruned->element))
if (HnswPtrEqual(base, currentNeighbors->items[i].element, pruned->element))
{
neighbors->items[i] = newHc;
currentNeighbors->items[i] = hc2;
/* Track update */
if (updateIdx != NULL)
@@ -1198,20 +1253,17 @@ PrecomputeHash(char *base, HnswElement element)
* Algorithm 1 from paper
*/
void
HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, HnswSupport * support, int m, int efConstruction, bool existing)
HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, FmgrInfo *procinfo, Oid collation, int m, int efConstruction, bool existing)
{
List *ep;
List *w;
int level = element->level;
int entryLevel;
HnswQuery q;
Datum q = HnswGetValue(base, element);
HnswElement skipElement = existing ? element : NULL;
bool inMemory = index == NULL;
q.value = HnswGetValue(base, element);
/* Precompute hash */
if (inMemory)
if (index == NULL)
PrecomputeHash(base, element);
/* No neighbors if no entry point */
@@ -1219,13 +1271,13 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
return;
/* Get entry point and level */
ep = list_make1(HnswEntryCandidate(base, entryPoint, &q, index, support, true));
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, procinfo, collation, true));
entryLevel = entryPoint->level;
/* 1st phase: greedy search to insert level */
for (int lc = entryLevel; lc >= level + 1; lc--)
{
w = HnswSearchLayer(base, &q, ep, 1, lc, index, support, m, true, skipElement);
w = HnswSearchLayer(base, q, ep, 1, lc, index, procinfo, collation, m, true, skipElement);
ep = w;
}
@@ -1244,7 +1296,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
List *lw = NIL;
ListCell *lc2;
w = HnswSearchLayer(base, &q, ep, efConstruction, lc, index, support, m, true, skipElement);
w = HnswSearchLayer(base, q, ep, efConstruction, lc, index, procinfo, collation, m, true, skipElement);
/* Convert search candidates to candidates */
foreach(lc2, w)
@@ -1260,7 +1312,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
/* Elements being deleted or skipped can help with search */
/* but should be removed before selecting neighbors */
if (!inMemory)
if (index != NULL)
lw = RemoveElements(base, lw, skipElement);
/*
@@ -1268,7 +1320,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
* sortCandidates to true for in-memory builds to enable closer
* caching, but there does not seem to be a difference in performance.
*/
neighbors = SelectNeighbors(base, lw, lm, support, &HnswGetNeighbors(base, element, lc)->closerSet, NULL, NULL, false);
neighbors = SelectNeighbors(base, lw, lm, lc, procinfo, collation, element, NULL, NULL, false);
AddConnections(base, element, neighbors, lc);
@@ -1278,6 +1330,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
PGDLLEXPORT Datum l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum halfvec_l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum minivec_l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum sparsevec_l2_normalize(PG_FUNCTION_ARGS);
static void
@@ -1326,6 +1379,20 @@ hnsw_halfvec_support(PG_FUNCTION_ARGS)
PG_RETURN_POINTER(&typeInfo);
};
FUNCTION_PREFIX PG_FUNCTION_INFO_V1(hnsw_minivec_support);
Datum
hnsw_minivec_support(PG_FUNCTION_ARGS)
{
static const HnswTypeInfo typeInfo = {
.maxDimensions = HNSW_MAX_DIM * 4,
/* Do not normalize to maximize precision */
.normalize = NULL,
.checkValue = NULL
};
PG_RETURN_POINTER(&typeInfo);
};
FUNCTION_PREFIX PG_FUNCTION_INFO_V1(hnsw_bit_support);
Datum
hnsw_bit_support(PG_FUNCTION_ARGS)

View File

@@ -184,12 +184,13 @@ static void
RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswElement entryPoint)
{
Relation index = vacuumstate->index;
HnswSupport *support = &vacuumstate->support;
Buffer buf;
Page page;
GenericXLogState *state;
int m = vacuumstate->m;
int efConstruction = vacuumstate->efConstruction;
FmgrInfo *procinfo = vacuumstate->procinfo;
Oid collation = vacuumstate->collation;
BufferAccessStrategy bas = vacuumstate->bas;
HnswNeighborTuple ntup = vacuumstate->ntup;
Size ntupSize = HNSW_NEIGHBOR_TUPLE_SIZE(element->level, m);
@@ -204,7 +205,7 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
element->heaptidsLength = 0;
/* Find neighbors for element, skipping itself */
HnswFindElementNeighbors(base, element, entryPoint, index, support, m, efConstruction, true);
HnswFindElementNeighbors(base, element, entryPoint, index, procinfo, collation, m, efConstruction, true);
/* Zero memory for each element */
MemSet(ntup, 0, HNSW_TUPLE_ALLOC_SIZE);
@@ -228,7 +229,7 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
UnlockReleaseBuffer(buf);
/* Update neighbors */
HnswUpdateNeighborsOnDisk(index, support, element, m, true, false);
HnswUpdateNeighborsOnDisk(index, procinfo, collation, element, m, true, false);
}
/*
@@ -238,7 +239,6 @@ static void
RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
{
Relation index = vacuumstate->index;
HnswSupport *support = &vacuumstate->support;
HnswElement highestPoint = &vacuumstate->highestPoint;
HnswElement entryPoint;
MemoryContext oldCtx = MemoryContextSwitchTo(vacuumstate->tmpCtx);
@@ -256,7 +256,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
LockPage(index, HNSW_UPDATE_LOCK, ShareLock);
/* Load element */
HnswLoadElement(highestPoint, NULL, NULL, index, support, true, NULL);
HnswLoadElement(highestPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true, NULL);
/* Repair if needed */
if (NeedsUpdated(vacuumstate, highestPoint))
@@ -294,7 +294,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
* is outdated, this can remove connections at higher levels in
* the graph until they are repaired, but this should be fine.
*/
HnswLoadElement(entryPoint, NULL, NULL, index, support, true, NULL);
HnswLoadElement(entryPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true, NULL);
if (NeedsUpdated(vacuumstate, entryPoint))
{
@@ -573,13 +573,13 @@ InitVacuumState(HnswVacuumState * vacuumstate, IndexVacuumInfo *info, IndexBulkD
vacuumstate->callback_state = callback_state;
vacuumstate->efConstruction = HnswGetEfConstruction(index);
vacuumstate->bas = GetAccessStrategy(BAS_BULKREAD);
vacuumstate->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
vacuumstate->collation = index->rd_indcollation[0];
vacuumstate->ntup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
vacuumstate->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
"Hnsw vacuum temporary context",
ALLOCSET_DEFAULT_SIZES);
HnswInitSupport(&vacuumstate->support, index);
/* Get m from metapage */
HnswGetMetaPageInfo(index, &vacuumstate->m, NULL);

View File

@@ -138,7 +138,7 @@ SampleRows(IvfflatBuildState * buildstate)
* Add tuple to sort
*/
static void
AddTupleToSort(Relation index, ItemPointer tid, Datum *values, bool *isnull, IvfflatBuildState * buildstate)
AddTupleToSort(Relation index, ItemPointer tid, Datum *values, IvfflatBuildState * buildstate)
{
double distance;
double minDistance = DBL_MAX;
@@ -184,11 +184,6 @@ AddTupleToSort(Relation index, ItemPointer tid, Datum *values, bool *isnull, Ivf
slot->tts_isnull[1] = false;
slot->tts_values[2] = value;
slot->tts_isnull[2] = false;
for (int i = 1; i < buildstate->tupdesc->natts; i++)
{
slot->tts_values[2 + i] = values[i];
slot->tts_isnull[2 + i] = isnull[i];
}
ExecStoreVirtualTuple(slot);
/*
@@ -220,7 +215,7 @@ BuildCallback(Relation index, ItemPointer tid, Datum *values,
oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
/* Add tuple to sort */
AddTupleToSort(index, tid, values, isnull, buildstate);
AddTupleToSort(index, tid, values, buildstate);
/* Reset memory context */
MemoryContextSwitchTo(oldCtx);
@@ -231,20 +226,19 @@ BuildCallback(Relation index, ItemPointer tid, Datum *values,
* Get index tuple from sort state
*/
static inline void
GetNextTuple(Tuplesortstate *sortstate, TupleDesc tupdesc, TupleTableSlot *slot, Datum *values, bool *isnull, IndexTuple *itup, int *list)
GetNextTuple(Tuplesortstate *sortstate, TupleDesc tupdesc, TupleTableSlot *slot, IndexTuple *itup, int *list)
{
Datum value;
bool isnull;
if (tuplesort_gettupleslot(sortstate, true, false, slot, NULL))
{
bool unused;
*list = DatumGetInt32(slot_getattr(slot, 1, &unused));
for (int i = 0; i < tupdesc->natts; i++)
values[i] = slot_getattr(slot, 3 + i, &isnull[i]);
*list = DatumGetInt32(slot_getattr(slot, 1, &isnull));
value = slot_getattr(slot, 3, &isnull);
/* Form the index tuple */
*itup = index_form_tuple(tupdesc, values, isnull);
(*itup)->t_tid = *((ItemPointer) DatumGetPointer(slot_getattr(slot, 2, &unused)));
*itup = index_form_tuple(tupdesc, &value, &isnull);
(*itup)->t_tid = *((ItemPointer) DatumGetPointer(slot_getattr(slot, 2, &isnull)));
}
else
*list = -1;
@@ -260,16 +254,14 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
IndexTuple itup = NULL; /* silence compiler warning */
int64 inserted = 0;
TupleTableSlot *slot = MakeSingleTupleTableSlot(buildstate->sortdesc, &TTSOpsMinimalTuple);
TupleDesc tupdesc = buildstate->tupdesc;
Datum *values = palloc(tupdesc->natts * sizeof(Datum));
bool *isnull = palloc(tupdesc->natts * sizeof(bool));
TupleTableSlot *slot = MakeSingleTupleTableSlot(buildstate->tupdesc, &TTSOpsMinimalTuple);
TupleDesc tupdesc = RelationGetDescr(index);
pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE, PROGRESS_IVFFLAT_PHASE_LOAD);
pgstat_progress_update_param(PROGRESS_CREATEIDX_TUPLES_TOTAL, buildstate->indtuples);
GetNextTuple(buildstate->sortstate, tupdesc, slot, values, isnull, &itup, &list);
GetNextTuple(buildstate->sortstate, tupdesc, slot, &itup, &list);
for (int i = 0; i < buildstate->centers->length; i++)
{
@@ -305,7 +297,7 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
pgstat_progress_update_param(PROGRESS_CREATEIDX_TUPLES_DONE, ++inserted);
GetNextTuple(buildstate->sortstate, tupdesc, slot, values, isnull, &itup, &list);
GetNextTuple(buildstate->sortstate, tupdesc, slot, &itup, &list);
}
insertPage = BufferGetBlockNumber(buf);
@@ -315,9 +307,6 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
/* Set the start and insert pages */
IvfflatUpdateList(index, buildstate->listInfo[i], insertPage, InvalidBlockNumber, startPage, forkNum);
}
pfree(values);
pfree(isnull);
}
/*
@@ -330,7 +319,6 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
buildstate->index = index;
buildstate->indexInfo = indexInfo;
buildstate->typeInfo = IvfflatGetTypeInfo(index);
buildstate->tupdesc = RelationGetDescr(index);
buildstate->lists = IvfflatGetLists(index);
buildstate->dimensions = TupleDescAttr(index->rd_att, 0)->atttypmod;
@@ -341,19 +329,6 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("type not supported for ivfflat index")));
/* TODO See if needed */
if (IndexRelationGetNumberOfKeyAttributes(index) > 3)
elog(ERROR, "index cannot have more than three columns");
if (!OidIsValid(index_getprocid(index, 1, IVFFLAT_DISTANCE_PROC)))
elog(ERROR, "first column must be a vector");
for (int i = 1; i < IndexRelationGetNumberOfKeyAttributes(index); i++)
{
if (OidIsValid(index_getprocid(index, i + 1, IVFFLAT_DISTANCE_PROC)))
elog(ERROR, "column %d cannot be a vector", i + 1);
}
/* Require column to have dimensions to be indexed */
if (buildstate->dimensions < 0)
ereport(ERROR,
@@ -381,13 +356,12 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
errmsg("dimensions must be greater than one for this opclass")));
/* Create tuple description for sorting */
buildstate->sortdesc = CreateTemplateTupleDesc(2 + buildstate->tupdesc->natts);
TupleDescInitEntry(buildstate->sortdesc, (AttrNumber) 1, "list", INT4OID, -1, 0);
TupleDescInitEntry(buildstate->sortdesc, (AttrNumber) 2, "tid", TIDOID, -1, 0);
for (int i = 0; i < buildstate->tupdesc->natts; i++)
TupleDescInitEntry(buildstate->sortdesc, (AttrNumber) (3 + i), NULL, buildstate->tupdesc->attrs[i].atttypid, -1, 0);
buildstate->tupdesc = CreateTemplateTupleDesc(3);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 1, "list", INT4OID, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 2, "tid", TIDOID, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 3, "vector", RelationGetDescr(index)->attrs[0].atttypid, -1, 0);
buildstate->slot = MakeSingleTupleTableSlot(buildstate->sortdesc, &TTSOpsVirtual);
buildstate->slot = MakeSingleTupleTableSlot(buildstate->tupdesc, &TTSOpsVirtual);
buildstate->centers = VectorArrayInit(buildstate->lists, buildstate->dimensions, buildstate->typeInfo->itemSize(buildstate->dimensions));
buildstate->listInfo = palloc(sizeof(ListInfo) * buildstate->lists);
@@ -659,7 +633,7 @@ IvfflatParallelScanAndSort(IvfflatSpool * ivfspool, IvfflatShared * ivfshared, S
InitBuildState(&buildstate, ivfspool->heap, ivfspool->index, indexInfo);
memcpy(buildstate.centers->items, ivfcenters, buildstate.centers->itemsize * buildstate.centers->maxlen);
buildstate.centers->length = buildstate.centers->maxlen;
ivfspool->sortstate = InitBuildSortState(buildstate.sortdesc, sortmem, coordinate);
ivfspool->sortstate = InitBuildSortState(buildstate.tupdesc, sortmem, coordinate);
buildstate.sortstate = ivfspool->sortstate;
scan = table_beginscan_parallel(ivfspool->heap,
ParallelTableScanFromIvfflatShared(ivfshared));
@@ -976,7 +950,7 @@ AssignTuples(IvfflatBuildState * buildstate)
}
/* Begin serial/leader tuplesort */
buildstate->sortstate = InitBuildSortState(buildstate->sortdesc, maintenance_work_mem, coordinate);
buildstate->sortstate = InitBuildSortState(buildstate->tupdesc, maintenance_work_mem, coordinate);
/* Add tuples to sort */
if (buildstate->heap != NULL)

View File

@@ -69,8 +69,6 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
GenericCosts costs;
int lists;
double ratio;
double sequentialRatio = 0.5;
double startupPages;
double spc_seq_page_cost;
Relation index;
@@ -87,8 +85,6 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
MemSet(&costs, 0, sizeof(costs));
genericcostestimate(root, path, loop_count, &costs);
index = index_open(path->indexinfo->indexoid, NoLock);
IvfflatGetMetaPageInfo(index, &lists, NULL);
index_close(index, NoLock);
@@ -98,26 +94,41 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
if (ratio > 1.0)
ratio = 1.0;
/*
* This gives us the subset of tuples to visit. This value is passed into
* the generic cost estimator to determine the number of pages to visit
* during the index scan.
*/
costs.numIndexTuples = path->indexinfo->tuples * ratio;
genericcostestimate(root, path, loop_count, &costs);
get_tablespace_page_costs(path->indexinfo->reltablespace, NULL, &spc_seq_page_cost);
/* Change some page cost from random to sequential */
costs.indexTotalCost -= sequentialRatio * costs.numIndexPages * (costs.spc_random_page_cost - spc_seq_page_cost);
/* Startup cost is cost before returning the first row */
costs.indexStartupCost = costs.indexTotalCost * ratio;
/* Adjust cost if needed since TOAST not included in seq scan cost */
startupPages = costs.numIndexPages * ratio;
if (startupPages > path->indexinfo->rel->pages && ratio < 0.5)
if (costs.numIndexPages > path->indexinfo->rel->pages && ratio < 0.5)
{
/* Change rest of page cost from random to sequential */
costs.indexStartupCost -= (1 - sequentialRatio) * startupPages * (costs.spc_random_page_cost - spc_seq_page_cost);
/* Change all page cost from random to sequential */
costs.indexTotalCost -= costs.numIndexPages * (costs.spc_random_page_cost - spc_seq_page_cost);
/* Remove cost of extra pages */
costs.indexStartupCost -= (startupPages - path->indexinfo->rel->pages) * spc_seq_page_cost;
costs.indexTotalCost -= (costs.numIndexPages - path->indexinfo->rel->pages) * spc_seq_page_cost;
}
else
{
/* Change some page cost from random to sequential */
costs.indexTotalCost -= 0.5 * costs.numIndexPages * (costs.spc_random_page_cost - spc_seq_page_cost);
}
*indexStartupCost = costs.indexStartupCost;
/*
* If the list selectivity is lower than what is returned from the generic
* cost estimator, use that.
*/
if (ratio < costs.indexSelectivity)
costs.indexSelectivity = ratio;
/* Use total cost since most work happens before first tuple is returned */
*indexStartupCost = costs.indexTotalCost;
*indexTotalCost = costs.indexTotalCost;
*indexSelectivity = costs.indexSelectivity;
*indexCorrelation = costs.indexCorrelation;
@@ -167,7 +178,7 @@ ivfflathandler(PG_FUNCTION_ARGS)
amroutine->amcanorderbyop = true;
amroutine->amcanbackward = false; /* can change direction mid-scan */
amroutine->amcanunique = false;
amroutine->amcanmulticol = true;
amroutine->amcanmulticol = false;
amroutine->amoptionalkey = true;
amroutine->amsearcharray = false;
amroutine->amsearchnulls = false;

View File

@@ -165,7 +165,6 @@ typedef struct IvfflatBuildState
Relation index;
IndexInfo *indexInfo;
const IvfflatTypeInfo *typeInfo;
TupleDesc tupdesc;
/* Settings */
int dimensions;
@@ -199,7 +198,7 @@ typedef struct IvfflatBuildState
/* Sorting */
Tuplesortstate *sortstate;
TupleDesc sortdesc;
TupleDesc tupdesc;
TupleTableSlot *slot;
/* Memory */

View File

@@ -78,8 +78,6 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
BlockNumber insertPage = InvalidBlockNumber;
ListInfo listInfo;
BlockNumber originalInsertPage;
TupleDesc tupdesc = RelationGetDescr(index);
Datum *newValues = palloc(tupdesc->natts * sizeof(Datum));
/* Detoast once for all calls */
value = PointerGetDatum(PG_DETOAST_DATUM(values[0]));
@@ -100,16 +98,12 @@ InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, R
IvfflatGetMetaPageInfo(index, NULL, NULL);
/* Find the insert page - sets the page and list info */
FindInsertPage(index, &value, &insertPage, &listInfo);
FindInsertPage(index, values, &insertPage, &listInfo);
Assert(BlockNumberIsValid(insertPage));
originalInsertPage = insertPage;
newValues[0] = value;
for (int i = 1; i < tupdesc->natts; i++)
newValues[i] = values[i];
/* Form tuple */
itup = index_form_tuple(tupdesc, newValues, isnull);
itup = index_form_tuple(RelationGetDescr(index), &value, isnull);
itup->t_tid = *heap_tid;
/* Get tuple size */

View File

@@ -104,34 +104,6 @@ GetScanLists(IndexScanDesc scan, Datum value)
}
}
/*
* Check if matches scan keys
*/
static bool
MatchesScanKeys(IndexScanDesc scan, IndexTuple itup, TupleDesc tupdesc)
{
for (int i = 0; i < scan->numberOfKeys; i++)
{
ScanKey key = &scan->keyData[i];
bool attnull = key->sk_flags & SK_ISNULL;
bool isnull;
Datum value = index_getattr(itup, key->sk_attno, tupdesc, &isnull);
if (isnull || attnull)
{
if (isnull != attnull)
return false;
}
else
{
if (!DatumGetBool(FunctionCall2Coll(&key->sk_func, key->sk_collation, value, key->sk_argument)))
return false;
}
}
return true;
}
/*
* Get items
*/
@@ -168,10 +140,6 @@ GetScanItems(IndexScanDesc scan, Datum value)
ItemId itemid = PageGetItemId(page, offno);
itup = (IndexTuple) PageGetItem(page, itemid);
if (!MatchesScanKeys(scan, itup, tupdesc))
continue;
datum = index_getattr(itup, 1, tupdesc, &isnull);
/*

View File

@@ -7,6 +7,7 @@
#include "halfutils.h"
#include "halfvec.h"
#include "ivfflat.h"
#include "minivec.h"
#include "storage/bufmgr.h"
/*
@@ -70,6 +71,9 @@ IvfflatOptionalProcInfo(Relation index, uint16 procnum)
Datum
IvfflatNormValue(const IvfflatTypeInfo * typeInfo, Oid collation, Datum value)
{
if (!typeInfo->normalize)
return value;
return DirectFunctionCall1Coll(typeInfo->normalize, collation, value);
}
@@ -231,6 +235,7 @@ IvfflatUpdateList(Relation index, ListInfo listInfo,
PGDLLEXPORT Datum l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum halfvec_l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum minivec_l2_normalize(PG_FUNCTION_ARGS);
PGDLLEXPORT Datum sparsevec_l2_normalize(PG_FUNCTION_ARGS);
static Size
@@ -245,6 +250,12 @@ HalfvecItemSize(int dimensions)
return HALFVEC_SIZE(dimensions);
}
static Size
MinivecItemSize(int dimensions)
{
return MINIVEC_SIZE(dimensions);
}
static Size
BitItemSize(int dimensions)
{
@@ -275,6 +286,18 @@ HalfvecUpdateCenter(Pointer v, int dimensions, float *x)
vec->x[k] = Float4ToHalfUnchecked(x[k]);
}
static void
MinivecUpdateCenter(Pointer v, int dimensions, float *x)
{
MiniVector *vec = (MiniVector *) v;
SET_VARSIZE(vec, MINIVEC_SIZE(dimensions));
vec->dim = dimensions;
for (int k = 0; k < dimensions; k++)
vec->x[k] = Float4ToFp8Unchecked(x[k]);
}
static void
BitUpdateCenter(Pointer v, int dimensions, float *x)
{
@@ -309,6 +332,15 @@ HalfvecSumCenter(Pointer v, float *x)
x[k] += HalfToFloat4(vec->x[k]);
}
static void
MinivecSumCenter(Pointer v, float *x)
{
MiniVector *vec = (MiniVector *) v;
for (int k = 0; k < vec->dim; k++)
x[k] += Fp8ToFloat4(vec->x[k]);
}
static void
BitSumCenter(Pointer v, float *x)
{
@@ -357,6 +389,22 @@ ivfflat_halfvec_support(PG_FUNCTION_ARGS)
PG_RETURN_POINTER(&typeInfo);
};
FUNCTION_PREFIX PG_FUNCTION_INFO_V1(ivfflat_minivec_support);
Datum
ivfflat_minivec_support(PG_FUNCTION_ARGS)
{
static const IvfflatTypeInfo typeInfo = {
.maxDimensions = IVFFLAT_MAX_DIM * 4,
/* Do not normalize to maximize precision */
.normalize = NULL,
.itemSize = MinivecItemSize,
.updateCenter = MinivecUpdateCenter,
.sumCenter = MinivecSumCenter
};
PG_RETURN_POINTER(&typeInfo);
};
FUNCTION_PREFIX PG_FUNCTION_INFO_V1(ivfflat_bit_support);
Datum
ivfflat_bit_support(PG_FUNCTION_ARGS)

1051
src/minivec.c Normal file

File diff suppressed because it is too large Load Diff

163
src/minivec.h Normal file
View File

@@ -0,0 +1,163 @@
#ifndef MINIVEC_H
#define MINIVEC_H
#define MINIVEC_MAX_DIM 16000
#define fp8 uint8
#define MINIVEC_SIZE(_dim) (offsetof(MiniVector, x) + sizeof(fp8)*(_dim))
#define DatumGetMiniVector(x) ((MiniVector *) PG_DETOAST_DATUM(x))
#define PG_GETARG_MINIVEC_P(x) DatumGetMiniVector(PG_GETARG_DATUM(x))
#define PG_RETURN_MINIVEC_P(x) PG_RETURN_POINTER(x)
typedef struct MiniVector
{
int32 vl_len_; /* varlena header (do not touch directly!) */
int16 dim; /* number of dimensions */
int16 unused; /* reserved for future use, always zero */
fp8 x[FLEXIBLE_ARRAY_MEMBER];
} MiniVector;
MiniVector *InitMiniVector(int dim);
/*
* Check if fp8 is NaN
*/
static inline bool
Fp8IsNan(fp8 num)
{
return (num & 0x7F) == 0x7F;
}
/*
* Check if fp8 is zero
*/
static inline bool
Fp8IsZero(fp8 num)
{
return num == 0;
}
/*
* Convert a fp8 to a float4
*/
static inline float
Fp8ToFloat4(fp8 num)
{
/* Lookup table for non-sign bits */
/* Uses uint32 for correctness */
uint32 lookup[128] = {0, 989855744, 998244352, 1002438656, 1006632960, 1008730112, 1010827264, 1012924416, 1015021568, 1016070144, 1017118720, 1018167296, 1019215872, 1020264448, 1021313024, 1022361600, 1023410176, 1024458752, 1025507328, 1026555904, 1027604480, 1028653056, 1029701632, 1030750208, 1031798784, 1032847360, 1033895936, 1034944512, 1035993088, 1037041664, 1038090240, 1039138816, 1040187392, 1041235968, 1042284544, 1043333120, 1044381696, 1045430272, 1046478848, 1047527424, 1048576000, 1049624576, 1050673152, 1051721728, 1052770304, 1053818880, 1054867456, 1055916032, 1056964608, 1058013184, 1059061760, 1060110336, 1061158912, 1062207488, 1063256064, 1064304640, 1065353216, 1066401792, 1067450368, 1068498944, 1069547520, 1070596096, 1071644672, 1072693248, 1073741824, 1074790400, 1075838976, 1076887552, 1077936128, 1078984704, 1080033280, 1081081856, 1082130432, 1083179008, 1084227584, 1085276160, 1086324736, 1087373312, 1088421888, 1089470464, 1090519040, 1091567616, 1092616192, 1093664768, 1094713344, 1095761920, 1096810496, 1097859072, 1098907648, 1099956224, 1101004800, 1102053376, 1103101952, 1104150528, 1105199104, 1106247680, 1107296256, 1108344832, 1109393408, 1110441984, 1111490560, 1112539136, 1113587712, 1114636288, 1115684864, 1116733440, 1117782016, 1118830592, 1119879168, 1120927744, 1121976320, 1123024896, 1124073472, 1125122048, 1126170624, 1127219200, 1128267776, 1129316352, 1130364928, 1131413504, 1132462080, 1133510656, 1134559232, 1135607808, 1136656384, 1137704960, 1138753536, 2146435072};
union
{
float f;
uint32 i;
} swap;
swap.i = lookup[num & 0x7F];
return (num & 0x80) == 0x80 ? -swap.f : swap.f;
}
/*
* Convert a float4 to a fp8
*/
static inline fp8
Float4ToFp8Unchecked(float num)
{
union
{
float f;
uint32 i;
} swap;
uint32 bin;
int exponent;
int mantissa;
uint8 result;
swap.f = num;
bin = swap.i;
exponent = (bin & 0x7F800000) >> 23;
mantissa = bin & 0x007FFFFF;
/* Sign */
result = (bin & 0x80000000) >> 24;
if (isinf(num) || isnan(num))
{
/* NaN */
result |= 0x7F;
}
else if (exponent > 114)
{
int m;
int gr;
int s;
exponent -= 127;
s = mantissa & 0x0007FFFF;
/* Subnormal */
if (exponent < -6)
{
int diff = -exponent - 6;
mantissa >>= diff;
mantissa += 1 << (23 - diff);
s |= mantissa & 0x0007FFFF;
}
m = mantissa >> 20;
/* Round */
gr = (mantissa >> 19) % 4;
if (gr == 3 || (gr == 1 && s != 0))
m += 1;
if (m == 8)
{
m = 0;
exponent += 1;
}
if (exponent > 8)
{
/* Infinite, which is NaN */
result |= 0x7F;
}
else
{
if (exponent >= -6)
result |= (exponent + 7) << 3;
result |= m;
}
}
return result;
}
/*
* Convert a float4 to a fp8
*/
static inline fp8
Float4ToFp8(float num)
{
fp8 result = Float4ToFp8Unchecked(num);
if (unlikely(Fp8IsNan(result)) && !isinf(num))
{
char *buf = palloc(FLOAT_SHORTEST_DECIMAL_LEN);
float_to_shortest_decimal_buf(num, buf);
ereport(ERROR,
(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
errmsg("\"%s\" is out of range for type minivec", buf)));
}
return result;
}
#endif

View File

@@ -13,6 +13,7 @@
#include "ivfflat.h"
#include "lib/stringinfo.h"
#include "libpq/pqformat.h"
#include "minivec.h"
#include "port.h" /* for strtof() */
#include "sparsevec.h"
#include "utils/array.h"
@@ -542,6 +543,28 @@ halfvec_to_vector(PG_FUNCTION_ARGS)
PG_RETURN_POINTER(result);
}
/*
* Convert fp8 vector to vector
*/
FUNCTION_PREFIX PG_FUNCTION_INFO_V1(minivec_to_vector);
Datum
minivec_to_vector(PG_FUNCTION_ARGS)
{
MiniVector *vec = PG_GETARG_MINIVEC_P(0);
int32 typmod = PG_GETARG_INT32(1);
Vector *result;
CheckDim(vec->dim);
CheckExpectedDim(typmod, vec->dim);
result = InitVector(vec->dim);
for (int i = 0; i < vec->dim; i++)
result->x[i] = Fp8ToFloat4(vec->x[i]);
PG_RETURN_POINTER(result);
}
VECTOR_TARGET_CLONES static float
VectorL2SquaredDistance(int dim, float *ax, float *bx)
{

View File

@@ -38,6 +38,26 @@ SELECT * FROM t ORDER BY val;
(4 rows)
DROP TABLE t;
-- minivec
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t (val);
SELECT * FROM t WHERE val = '[1,2,3]';
val
---------
[1,2,3]
(1 row)
SELECT * FROM t ORDER BY val;
val
---------
[0,0,0]
[1,1,1]
[1,2,3]
(4 rows)
DROP TABLE t;
-- sparsevec
CREATE TABLE t (val sparsevec(3));

View File

@@ -140,6 +140,64 @@ SELECT '{1e-8,-1e-8}'::real[]::halfvec;
[0,-0]
(1 row)
SELECT '[1,2,3]'::vector::minivec;
minivec
---------
[1,2,3]
(1 row)
SELECT '[1,2,3]'::vector::minivec(3);
minivec
---------
[1,2,3]
(1 row)
SELECT '[1,2,3]'::vector::minivec(2);
ERROR: expected 2 dimensions, not 3
SELECT '[465]'::vector::minivec;
ERROR: "465" is out of range for type minivec
SELECT '[1e-8]'::vector::minivec;
minivec
---------
[0]
(1 row)
SELECT '[1,2,3]'::minivec::vector;
vector
---------
[1,2,3]
(1 row)
SELECT '[1,2,3]'::minivec::vector(3);
vector
---------
[1,2,3]
(1 row)
SELECT '[1,2,3]'::minivec::vector(2);
ERROR: expected 2 dimensions, not 3
SELECT '{1,2,3}'::real[]::minivec;
minivec
---------
[1,2,3]
(1 row)
SELECT '{1,2,3}'::real[]::minivec(3);
minivec
---------
[1,2,3]
(1 row)
SELECT '{1,2,3}'::real[]::minivec(2);
ERROR: expected 2 dimensions, not 3
SELECT '{465,-465}'::real[]::minivec;
ERROR: "465" is out of range for type minivec
SELECT '{1e-8,-1e-8}'::real[]::minivec;
minivec
---------
[0,-0]
(1 row)
SELECT '[0,1.5,0,3.5,0]'::vector::sparsevec;
sparsevec
-----------------

View File

@@ -30,6 +30,23 @@ SELECT * FROM t2 ORDER BY val;
(4 rows)
DROP TABLE t;
DROP TABLE t2;
-- minivec
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE TABLE t2 (val minivec(3));
\copy t TO 'results/minivec.bin' WITH (FORMAT binary)
\copy t2 FROM 'results/minivec.bin' WITH (FORMAT binary)
SELECT * FROM t2 ORDER BY val;
val
---------
[0,0,0]
[1,1,1]
[1,2,3]
(4 rows)
DROP TABLE t;
DROP TABLE t2;
-- sparsevec

View File

@@ -0,0 +1,84 @@
SET enable_seqscan = off;
-- L2
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_l2_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <-> '[3,3,3]';
val
---------
[1,2,3]
[1,2,4]
[1,1,1]
[0,0,0]
(4 rows)
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::minivec)) t2;
count
-------
4
(1 row)
SELECT COUNT(*) FROM t;
count
-------
5
(1 row)
TRUNCATE t;
NOTICE: ivfflat index created with little data
DETAIL: This will cause low recall.
HINT: Drop the index until the table has more data.
SELECT * FROM t ORDER BY val <-> '[3,3,3]';
val
-----
(0 rows)
DROP TABLE t;
-- inner product
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_ip_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <#> '[3,3,3]';
val
---------
[1,2,4]
[1,2,3]
[1,1,1]
[0,0,0]
(4 rows)
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <#> (SELECT NULL::minivec)) t2;
count
-------
4
(1 row)
DROP TABLE t;
-- cosine
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_cosine_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <=> '[3,3,3]';
val
---------
[1,1,1]
[1,2,3]
[1,2,4]
(3 rows)
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <=> '[0,0,0]') t2;
count
-------
3
(1 row)
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <=> (SELECT NULL::minivec)) t2;
count
-------
3
(1 row)
DROP TABLE t;

588
test/expected/minivec.out Normal file
View File

@@ -0,0 +1,588 @@
SELECT '[1,2,3]'::minivec;
minivec
---------
[1,2,3]
(1 row)
SELECT '[-1,-2,-3]'::minivec;
minivec
------------
[-1,-2,-3]
(1 row)
SELECT '[1.,2.,3.]'::minivec;
minivec
---------
[1,2,3]
(1 row)
SELECT ' [ 1, 2 , 3 ] '::minivec;
minivec
---------
[1,2,3]
(1 row)
SELECT '[1.23456]'::minivec;
minivec
---------
[1.25]
(1 row)
SELECT '[hello,1]'::minivec;
ERROR: invalid input syntax for type minivec: "[hello,1]"
LINE 1: SELECT '[hello,1]'::minivec;
^
SELECT '[NaN,1]'::minivec;
ERROR: NaN not allowed in minivec
LINE 1: SELECT '[NaN,1]'::minivec;
^
SELECT '[Infinity,1]'::minivec;
ERROR: "Infinity" is out of range for type minivec
LINE 1: SELECT '[Infinity,1]'::minivec;
^
SELECT '[-Infinity,1]'::minivec;
ERROR: "-Infinity" is out of range for type minivec
LINE 1: SELECT '[-Infinity,1]'::minivec;
^
SELECT '[65519,-65519]'::minivec;
ERROR: "65519" is out of range for type minivec
LINE 1: SELECT '[65519,-65519]'::minivec;
^
SELECT '[65520,-65520]'::minivec;
ERROR: "65520" is out of range for type minivec
LINE 1: SELECT '[65520,-65520]'::minivec;
^
SELECT '[1e-8,-1e-8]'::minivec;
minivec
---------
[0,-0]
(1 row)
SELECT '[4e38,1]'::minivec;
ERROR: "4e38" is out of range for type minivec
LINE 1: SELECT '[4e38,1]'::minivec;
^
SELECT '[1e-46,1]'::minivec;
minivec
---------
[0,1]
(1 row)
SELECT '[1,2,3'::minivec;
ERROR: invalid input syntax for type minivec: "[1,2,3"
LINE 1: SELECT '[1,2,3'::minivec;
^
SELECT '[1,2,3]9'::minivec;
ERROR: invalid input syntax for type minivec: "[1,2,3]9"
LINE 1: SELECT '[1,2,3]9'::minivec;
^
DETAIL: Junk after closing right brace.
SELECT '1,2,3'::minivec;
ERROR: invalid input syntax for type minivec: "1,2,3"
LINE 1: SELECT '1,2,3'::minivec;
^
DETAIL: Vector contents must start with "[".
SELECT ''::minivec;
ERROR: invalid input syntax for type minivec: ""
LINE 1: SELECT ''::minivec;
^
DETAIL: Vector contents must start with "[".
SELECT '['::minivec;
ERROR: invalid input syntax for type minivec: "["
LINE 1: SELECT '['::minivec;
^
SELECT '[ '::minivec;
ERROR: invalid input syntax for type minivec: "[ "
LINE 1: SELECT '[ '::minivec;
^
SELECT '[,'::minivec;
ERROR: invalid input syntax for type minivec: "[,"
LINE 1: SELECT '[,'::minivec;
^
SELECT '[]'::minivec;
ERROR: minivec must have at least 1 dimension
LINE 1: SELECT '[]'::minivec;
^
SELECT '[ ]'::minivec;
ERROR: minivec must have at least 1 dimension
LINE 1: SELECT '[ ]'::minivec;
^
SELECT '[,]'::minivec;
ERROR: invalid input syntax for type minivec: "[,]"
LINE 1: SELECT '[,]'::minivec;
^
SELECT '[1,]'::minivec;
ERROR: invalid input syntax for type minivec: "[1,]"
LINE 1: SELECT '[1,]'::minivec;
^
SELECT '[1a]'::minivec;
ERROR: invalid input syntax for type minivec: "[1a]"
LINE 1: SELECT '[1a]'::minivec;
^
SELECT '[1,,3]'::minivec;
ERROR: invalid input syntax for type minivec: "[1,,3]"
LINE 1: SELECT '[1,,3]'::minivec;
^
SELECT '[1, ,3]'::minivec;
ERROR: invalid input syntax for type minivec: "[1, ,3]"
LINE 1: SELECT '[1, ,3]'::minivec;
^
SELECT '[1,2,3]'::minivec(3);
minivec
---------
[1,2,3]
(1 row)
SELECT '[1,2,3]'::minivec(2);
ERROR: expected 2 dimensions, not 3
SELECT '[1,2,3]'::minivec(3, 2);
ERROR: invalid type modifier
LINE 1: SELECT '[1,2,3]'::minivec(3, 2);
^
SELECT '[1,2,3]'::minivec('a');
ERROR: invalid input syntax for type integer: "a"
LINE 1: SELECT '[1,2,3]'::minivec('a');
^
SELECT '[1,2,3]'::minivec(0);
ERROR: dimensions for type minivec must be at least 1
LINE 1: SELECT '[1,2,3]'::minivec(0);
^
SELECT '[1,2,3]'::minivec(16001);
ERROR: dimensions for type minivec cannot exceed 16000
LINE 1: SELECT '[1,2,3]'::minivec(16001);
^
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::minivec[]);
unnest
---------
[1,2,3]
[4,5,6]
(2 rows)
SELECT '{"[1,2,3]"}'::minivec(2)[];
ERROR: expected 2 dimensions, not 3
SELECT '[1,2,3]'::minivec + '[4,5,6]';
?column?
----------
[5,7,9]
(1 row)
SELECT '[448]'::minivec + '[448]';
ERROR: value out of range: overflow
SELECT '[1,2]'::minivec + '[3]';
ERROR: different minivec dimensions 2 and 1
SELECT '[1,2,3]'::minivec - '[4,5,6]';
?column?
------------
[-3,-3,-3]
(1 row)
SELECT '[-448]'::minivec - '[448]';
ERROR: value out of range: overflow
SELECT '[1,2]'::minivec - '[3]';
ERROR: different minivec dimensions 2 and 1
SELECT '[1,2,3]'::minivec * '[4,5,6]';
?column?
-----------
[4,10,18]
(1 row)
SELECT '[448]'::minivec * '[448]';
ERROR: value out of range: overflow
SELECT '[1e-7]'::minivec * '[1e-7]';
?column?
----------
[0]
(1 row)
SELECT '[1,2]'::minivec * '[3]';
ERROR: different minivec dimensions 2 and 1
SELECT '[1,2,3]'::minivec || '[4,5]';
?column?
-------------
[1,2,3,4,5]
(1 row)
SELECT array_fill(0, ARRAY[16000])::minivec || '[1]';
ERROR: minivec cannot have more than 16000 dimensions
SELECT '[1,2,3]'::minivec < '[1,2,3]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec < '[1,2]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec <= '[1,2,3]';
?column?
----------
t
(1 row)
SELECT '[1,2,3]'::minivec <= '[1,2]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec = '[1,2,3]';
?column?
----------
t
(1 row)
SELECT '[1,2,3]'::minivec = '[1,2]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec != '[1,2,3]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec != '[1,2]';
?column?
----------
t
(1 row)
SELECT '[1,2,3]'::minivec >= '[1,2,3]';
?column?
----------
t
(1 row)
SELECT '[1,2,3]'::minivec >= '[1,2]';
?column?
----------
t
(1 row)
SELECT '[1,2,3]'::minivec > '[1,2,3]';
?column?
----------
f
(1 row)
SELECT '[1,2,3]'::minivec > '[1,2]';
?column?
----------
t
(1 row)
SELECT minivec_cmp('[1,2,3]', '[1,2,3]');
minivec_cmp
-------------
0
(1 row)
SELECT minivec_cmp('[1,2,3]', '[0,0,0]');
minivec_cmp
-------------
1
(1 row)
SELECT minivec_cmp('[0,0,0]', '[1,2,3]');
minivec_cmp
-------------
-1
(1 row)
SELECT minivec_cmp('[1,2]', '[1,2,3]');
minivec_cmp
-------------
-1
(1 row)
SELECT minivec_cmp('[1,2,3]', '[1,2]');
minivec_cmp
-------------
1
(1 row)
SELECT minivec_cmp('[1,2]', '[2,3,4]');
minivec_cmp
-------------
-1
(1 row)
SELECT minivec_cmp('[2,3]', '[1,2,3]');
minivec_cmp
-------------
1
(1 row)
SELECT vector_dims('[1,2,3]'::minivec);
vector_dims
-------------
3
(1 row)
SELECT round(l2_norm('[1,1]'::minivec)::numeric, 5);
round
---------
1.41421
(1 row)
SELECT l2_norm('[3,4]'::minivec);
l2_norm
---------
5
(1 row)
SELECT l2_norm('[0,1]'::minivec);
l2_norm
---------
1
(1 row)
SELECT l2_norm('[0,0]'::minivec);
l2_norm
---------
0
(1 row)
SELECT l2_norm('[2]'::minivec);
l2_norm
---------
2
(1 row)
SELECT l2_distance('[0,0]'::minivec, '[3,4]');
l2_distance
-------------
5
(1 row)
SELECT l2_distance('[0,0]'::minivec, '[0,1]');
l2_distance
-------------
1
(1 row)
SELECT l2_distance('[1,2]'::minivec, '[3]');
ERROR: different minivec dimensions 2 and 1
SELECT l2_distance('[1,1,1,1,1,1,1,1,1]'::minivec, '[1,1,1,1,1,1,1,4,5]');
l2_distance
-------------
5
(1 row)
SELECT '[0,0]'::minivec <-> '[3,4]';
?column?
----------
5
(1 row)
SELECT inner_product('[1,2]'::minivec, '[3,4]');
inner_product
---------------
11
(1 row)
SELECT inner_product('[1,2]'::minivec, '[3]');
ERROR: different minivec dimensions 2 and 1
SELECT inner_product('[448]'::minivec, '[448]');
inner_product
---------------
200704
(1 row)
SELECT inner_product('[1,1,1,1,1,1,1,1,1]'::minivec, '[1,2,3,4,5,6,7,8,9]');
inner_product
---------------
45
(1 row)
SELECT '[1,2]'::minivec <#> '[3,4]';
?column?
----------
-11
(1 row)
SELECT cosine_distance('[1,2]'::minivec, '[2,4]');
cosine_distance
-----------------
0
(1 row)
SELECT cosine_distance('[1,2]'::minivec, '[0,0]');
cosine_distance
-----------------
NaN
(1 row)
SELECT cosine_distance('[1,1]'::minivec, '[1,1]');
cosine_distance
-----------------
0
(1 row)
SELECT cosine_distance('[1,0]'::minivec, '[0,2]');
cosine_distance
-----------------
1
(1 row)
SELECT cosine_distance('[1,1]'::minivec, '[-1,-1]');
cosine_distance
-----------------
2
(1 row)
SELECT cosine_distance('[1,2]'::minivec, '[3]');
ERROR: different minivec dimensions 2 and 1
SELECT cosine_distance('[1,1]'::minivec, '[1.1,1.1]');
cosine_distance
-----------------
0
(1 row)
SELECT cosine_distance('[1,1]'::minivec, '[-1.1,-1.1]');
cosine_distance
-----------------
2
(1 row)
SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[1,2,3,4,5,6,7,8,9]');
cosine_distance
-----------------
0
(1 row)
SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[-1,-2,-3,-4,-5,-6,-7,-8,-9]');
cosine_distance
-----------------
2
(1 row)
SELECT '[1,2]'::minivec <=> '[2,4]';
?column?
----------
0
(1 row)
SELECT l1_distance('[0,0]'::minivec, '[3,4]');
l1_distance
-------------
7
(1 row)
SELECT l1_distance('[0,0]'::minivec, '[0,1]');
l1_distance
-------------
1
(1 row)
SELECT l1_distance('[1,2]'::minivec, '[3]');
ERROR: different minivec dimensions 2 and 1
SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[1,2,3,4,5,6,7,8,9]');
l1_distance
-------------
0
(1 row)
SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[0,3,2,5,4,7,6,9,8]');
l1_distance
-------------
9
(1 row)
SELECT '[0,0]'::minivec <+> '[3,4]';
?column?
----------
7
(1 row)
SELECT l2_normalize('[3,4]'::minivec);
l2_normalize
----------------
[0.625,0.8125]
(1 row)
SELECT l2_normalize('[3,0]'::minivec);
l2_normalize
--------------
[1,0]
(1 row)
SELECT l2_normalize('[0,0.1]'::minivec);
l2_normalize
--------------
[0,1]
(1 row)
SELECT l2_normalize('[0,0]'::minivec);
l2_normalize
--------------
[0,0]
(1 row)
SELECT l2_normalize('[448]'::minivec);
l2_normalize
--------------
[1]
(1 row)
SELECT binary_quantize('[1,0,-1]'::minivec);
binary_quantize
-----------------
100
(1 row)
SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::minivec);
binary_quantize
-----------------
01001110101
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, 1, 3);
subvector
-----------
[1,2,3]
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 2);
subvector
-----------
[3,4]
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, -1, 3);
subvector
-----------
[1]
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 9);
subvector
-----------
[3,4,5]
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, 1, 0);
ERROR: minivec must have at least 1 dimension
SELECT subvector('[1,2,3,4,5]'::minivec, 3, -1);
ERROR: minivec must have at least 1 dimension
SELECT subvector('[1,2,3,4,5]'::minivec, -1, 2);
ERROR: minivec must have at least 1 dimension
SELECT subvector('[1,2,3,4,5]'::minivec, 2147483647, 10);
ERROR: minivec must have at least 1 dimension
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 2147483647);
subvector
-----------
[3,4,5]
(1 row)
SELECT subvector('[1,2,3,4,5]'::minivec, -2147483644, 2147483647);
subvector
-----------
[1,2]
(1 row)

View File

@@ -22,6 +22,17 @@ SELECT * FROM t ORDER BY val;
DROP TABLE t;
-- minivec
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t (val);
SELECT * FROM t WHERE val = '[1,2,3]';
SELECT * FROM t ORDER BY val;
DROP TABLE t;
-- sparsevec
CREATE TABLE t (val sparsevec(3));

View File

@@ -38,6 +38,22 @@ SELECT '{1,2,3}'::real[]::halfvec(2);
SELECT '{65520,-65520}'::real[]::halfvec;
SELECT '{1e-8,-1e-8}'::real[]::halfvec;
SELECT '[1,2,3]'::vector::minivec;
SELECT '[1,2,3]'::vector::minivec(3);
SELECT '[1,2,3]'::vector::minivec(2);
SELECT '[465]'::vector::minivec;
SELECT '[1e-8]'::vector::minivec;
SELECT '[1,2,3]'::minivec::vector;
SELECT '[1,2,3]'::minivec::vector(3);
SELECT '[1,2,3]'::minivec::vector(2);
SELECT '{1,2,3}'::real[]::minivec;
SELECT '{1,2,3}'::real[]::minivec(3);
SELECT '{1,2,3}'::real[]::minivec(2);
SELECT '{465,-465}'::real[]::minivec;
SELECT '{1e-8,-1e-8}'::real[]::minivec;
SELECT '[0,1.5,0,3.5,0]'::vector::sparsevec;
SELECT '[0,1.5,0,3.5,0]'::vector::sparsevec(5);
SELECT '[0,1.5,0,3.5,0]'::vector::sparsevec(4);

View File

@@ -28,6 +28,21 @@ SELECT * FROM t2 ORDER BY val;
DROP TABLE t;
DROP TABLE t2;
-- minivec
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE TABLE t2 (val minivec(3));
\copy t TO 'results/minivec.bin' WITH (FORMAT binary)
\copy t2 FROM 'results/minivec.bin' WITH (FORMAT binary)
SELECT * FROM t2 ORDER BY val;
DROP TABLE t;
DROP TABLE t2;
-- sparsevec
CREATE TABLE t (val sparsevec(3));

View File

@@ -0,0 +1,45 @@
SET enable_seqscan = off;
-- L2
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_l2_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <-> '[3,3,3]';
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <-> (SELECT NULL::minivec)) t2;
SELECT COUNT(*) FROM t;
TRUNCATE t;
SELECT * FROM t ORDER BY val <-> '[3,3,3]';
DROP TABLE t;
-- inner product
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_ip_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <#> '[3,3,3]';
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <#> (SELECT NULL::minivec)) t2;
DROP TABLE t;
-- cosine
CREATE TABLE t (val minivec(3));
INSERT INTO t (val) VALUES ('[0,0,0]'), ('[1,2,3]'), ('[1,1,1]'), (NULL);
CREATE INDEX ON t USING ivfflat (val minivec_cosine_ops) WITH (lists = 1);
INSERT INTO t (val) VALUES ('[1,2,4]');
SELECT * FROM t ORDER BY val <=> '[3,3,3]';
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <=> '[0,0,0]') t2;
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <=> (SELECT NULL::minivec)) t2;
DROP TABLE t;

134
test/sql/minivec.sql Normal file
View File

@@ -0,0 +1,134 @@
SELECT '[1,2,3]'::minivec;
SELECT '[-1,-2,-3]'::minivec;
SELECT '[1.,2.,3.]'::minivec;
SELECT ' [ 1, 2 , 3 ] '::minivec;
SELECT '[1.23456]'::minivec;
SELECT '[hello,1]'::minivec;
SELECT '[NaN,1]'::minivec;
SELECT '[Infinity,1]'::minivec;
SELECT '[-Infinity,1]'::minivec;
SELECT '[65519,-65519]'::minivec;
SELECT '[65520,-65520]'::minivec;
SELECT '[1e-8,-1e-8]'::minivec;
SELECT '[4e38,1]'::minivec;
SELECT '[1e-46,1]'::minivec;
SELECT '[1,2,3'::minivec;
SELECT '[1,2,3]9'::minivec;
SELECT '1,2,3'::minivec;
SELECT ''::minivec;
SELECT '['::minivec;
SELECT '[ '::minivec;
SELECT '[,'::minivec;
SELECT '[]'::minivec;
SELECT '[ ]'::minivec;
SELECT '[,]'::minivec;
SELECT '[1,]'::minivec;
SELECT '[1a]'::minivec;
SELECT '[1,,3]'::minivec;
SELECT '[1, ,3]'::minivec;
SELECT '[1,2,3]'::minivec(3);
SELECT '[1,2,3]'::minivec(2);
SELECT '[1,2,3]'::minivec(3, 2);
SELECT '[1,2,3]'::minivec('a');
SELECT '[1,2,3]'::minivec(0);
SELECT '[1,2,3]'::minivec(16001);
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::minivec[]);
SELECT '{"[1,2,3]"}'::minivec(2)[];
SELECT '[1,2,3]'::minivec + '[4,5,6]';
SELECT '[448]'::minivec + '[448]';
SELECT '[1,2]'::minivec + '[3]';
SELECT '[1,2,3]'::minivec - '[4,5,6]';
SELECT '[-448]'::minivec - '[448]';
SELECT '[1,2]'::minivec - '[3]';
SELECT '[1,2,3]'::minivec * '[4,5,6]';
SELECT '[448]'::minivec * '[448]';
SELECT '[1e-7]'::minivec * '[1e-7]';
SELECT '[1,2]'::minivec * '[3]';
SELECT '[1,2,3]'::minivec || '[4,5]';
SELECT array_fill(0, ARRAY[16000])::minivec || '[1]';
SELECT '[1,2,3]'::minivec < '[1,2,3]';
SELECT '[1,2,3]'::minivec < '[1,2]';
SELECT '[1,2,3]'::minivec <= '[1,2,3]';
SELECT '[1,2,3]'::minivec <= '[1,2]';
SELECT '[1,2,3]'::minivec = '[1,2,3]';
SELECT '[1,2,3]'::minivec = '[1,2]';
SELECT '[1,2,3]'::minivec != '[1,2,3]';
SELECT '[1,2,3]'::minivec != '[1,2]';
SELECT '[1,2,3]'::minivec >= '[1,2,3]';
SELECT '[1,2,3]'::minivec >= '[1,2]';
SELECT '[1,2,3]'::minivec > '[1,2,3]';
SELECT '[1,2,3]'::minivec > '[1,2]';
SELECT minivec_cmp('[1,2,3]', '[1,2,3]');
SELECT minivec_cmp('[1,2,3]', '[0,0,0]');
SELECT minivec_cmp('[0,0,0]', '[1,2,3]');
SELECT minivec_cmp('[1,2]', '[1,2,3]');
SELECT minivec_cmp('[1,2,3]', '[1,2]');
SELECT minivec_cmp('[1,2]', '[2,3,4]');
SELECT minivec_cmp('[2,3]', '[1,2,3]');
SELECT vector_dims('[1,2,3]'::minivec);
SELECT round(l2_norm('[1,1]'::minivec)::numeric, 5);
SELECT l2_norm('[3,4]'::minivec);
SELECT l2_norm('[0,1]'::minivec);
SELECT l2_norm('[0,0]'::minivec);
SELECT l2_norm('[2]'::minivec);
SELECT l2_distance('[0,0]'::minivec, '[3,4]');
SELECT l2_distance('[0,0]'::minivec, '[0,1]');
SELECT l2_distance('[1,2]'::minivec, '[3]');
SELECT l2_distance('[1,1,1,1,1,1,1,1,1]'::minivec, '[1,1,1,1,1,1,1,4,5]');
SELECT '[0,0]'::minivec <-> '[3,4]';
SELECT inner_product('[1,2]'::minivec, '[3,4]');
SELECT inner_product('[1,2]'::minivec, '[3]');
SELECT inner_product('[448]'::minivec, '[448]');
SELECT inner_product('[1,1,1,1,1,1,1,1,1]'::minivec, '[1,2,3,4,5,6,7,8,9]');
SELECT '[1,2]'::minivec <#> '[3,4]';
SELECT cosine_distance('[1,2]'::minivec, '[2,4]');
SELECT cosine_distance('[1,2]'::minivec, '[0,0]');
SELECT cosine_distance('[1,1]'::minivec, '[1,1]');
SELECT cosine_distance('[1,0]'::minivec, '[0,2]');
SELECT cosine_distance('[1,1]'::minivec, '[-1,-1]');
SELECT cosine_distance('[1,2]'::minivec, '[3]');
SELECT cosine_distance('[1,1]'::minivec, '[1.1,1.1]');
SELECT cosine_distance('[1,1]'::minivec, '[-1.1,-1.1]');
SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[1,2,3,4,5,6,7,8,9]');
SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[-1,-2,-3,-4,-5,-6,-7,-8,-9]');
SELECT '[1,2]'::minivec <=> '[2,4]';
SELECT l1_distance('[0,0]'::minivec, '[3,4]');
SELECT l1_distance('[0,0]'::minivec, '[0,1]');
SELECT l1_distance('[1,2]'::minivec, '[3]');
SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[1,2,3,4,5,6,7,8,9]');
SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::minivec, '[0,3,2,5,4,7,6,9,8]');
SELECT '[0,0]'::minivec <+> '[3,4]';
SELECT l2_normalize('[3,4]'::minivec);
SELECT l2_normalize('[3,0]'::minivec);
SELECT l2_normalize('[0,0.1]'::minivec);
SELECT l2_normalize('[0,0]'::minivec);
SELECT l2_normalize('[448]'::minivec);
SELECT binary_quantize('[1,0,-1]'::minivec);
SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::minivec);
SELECT subvector('[1,2,3,4,5]'::minivec, 1, 3);
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 2);
SELECT subvector('[1,2,3,4,5]'::minivec, -1, 3);
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 9);
SELECT subvector('[1,2,3,4,5]'::minivec, 1, 0);
SELECT subvector('[1,2,3,4,5]'::minivec, 3, -1);
SELECT subvector('[1,2,3,4,5]'::minivec, -1, 2);
SELECT subvector('[1,2,3,4,5]'::minivec, 2147483647, 10);
SELECT subvector('[1,2,3,4,5]'::minivec, 3, 2147483647);
SELECT subvector('[1,2,3,4,5]'::minivec, -2147483644, 2147483647);

View File

@@ -6,7 +6,13 @@ use Test::More;
my $dim = 3;
my $array_sql = join(",", ('random()') x $dim);
my @r = ();
for (1 .. $dim)
{
my $v = int(rand(1000)) + 1;
push(@r, "i % $v");
}
my $array_sql = join(", ", @r);
# Initialize node
my $node = PostgreSQL::Test::Cluster->new('node');
@@ -17,20 +23,19 @@ $node->start;
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(1, 100000) i;"
"INSERT INTO tst SELECT i % 10, ARRAY[$array_sql] FROM generate_series(1, 100000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v vector_l2_ops);");
# Get size
my $size = $node->safe_psql("postgres", "SELECT pg_total_relation_size('tst_v_idx');");
# Store values
$node->safe_psql("postgres", "CREATE TABLE tmp AS SELECT * FROM tst;");
# Delete all, vacuum, and insert same data
$node->safe_psql("postgres", "DELETE FROM tst;");
$node->safe_psql("postgres", "VACUUM tst;");
$node->safe_psql("postgres", "INSERT INTO tst SELECT * FROM tmp;");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i % 10, ARRAY[$array_sql] FROM generate_series(1, 100000) i;"
);
# Check size
my $new_size = $node->safe_psql("postgres", "SELECT pg_total_relation_size('tst_v_idx');");

View File

@@ -94,7 +94,8 @@ like($explain, qr/Seq Scan/);
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE v <-> '$query' < 1 ORDER BY v <-> '$query';
));
like($explain, qr/Seq Scan/);
# TODO Do not use index
like($explain, qr/Index Scan using idx/);
# Test attribute index
$node->safe_psql("postgres", "CREATE INDEX attribute_idx ON tst (c);");
@@ -109,6 +110,7 @@ $node->safe_psql("postgres", "CREATE INDEX partial_idx ON tst USING ivfflat (v v
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE c = $c ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using partial_idx/);
# TODO Use partial index
like($explain, qr/Index Scan using idx/);
done_testing();

View File

@@ -18,13 +18,9 @@ $node->start;
# Create table and index
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim), c int4, t text);");
$node->safe_psql("postgres", "CREATE TABLE cat (i int4 PRIMARY KEY, t text, b boolean);");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql], i % $nc, 'test ' || i FROM generate_series(1, 10000) i;"
);
$node->safe_psql("postgres",
"INSERT INTO cat SELECT i, 'cat ' || i, i % 5 = 0 FROM generate_series(1, $nc) i;"
);
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING hnsw (v vector_l2_ops);");
$node->safe_psql("postgres", "ANALYZE tst;");
@@ -41,7 +37,8 @@ my $c = int(rand() * $nc);
my $explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE c = $c ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Seq Scan/);
# TODO Do not use index
like($explain, qr/Index Scan using idx/);
# Test attribute filtering with few rows removed
$explain = $node->safe_psql("postgres", qq(
@@ -59,7 +56,8 @@ like($explain, qr/Index Scan using idx/);
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE c < 1 ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Seq Scan/);
# TODO Do not use index
like($explain, qr/Index Scan using idx/);
# Test attribute filtering with few rows removed like
$explain = $node->safe_psql("postgres", qq(
@@ -98,25 +96,13 @@ $explain = $node->safe_psql("postgres", qq(
));
like($explain, qr/Seq Scan/);
# Test join
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT cat.t FROM cat INNER JOIN tst ON cat.i = tst.c ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
# Test join with attribute filtering
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT cat.t FROM cat INNER JOIN tst ON cat.i = tst.c WHERE cat.b = 't' ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
# Test attribute index
$node->safe_psql("postgres", "CREATE INDEX attribute_idx ON tst (c);");
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE c = $c ORDER BY v <-> '$query' LIMIT $limit;
));
# Use attribute index
like($explain, qr/Bitmap Index Scan on attribute_idx/);
# TODO Use attribute index
like($explain, qr/Index Scan using idx/);
# Test partial index
$node->safe_psql("postgres", "CREATE INDEX partial_idx ON tst USING hnsw (v vector_l2_ops) WHERE (c = $c);");

View File

@@ -40,6 +40,10 @@ for (1 .. 50)
$actual = $node->safe_psql("postgres", "SELECT halfvec_cmp(v::halfvec, '$query'::real[]::halfvec) FROM tst");
is($expected, $actual);
# Test minivec
$actual = $node->safe_psql("postgres", "SELECT minivec_cmp(v::minivec, '$query'::real[]::minivec) FROM tst");
is($expected, $actual);
# Test sparsevec
$actual = $node->safe_psql("postgres", "SELECT sparsevec_cmp(v::vector::sparsevec, '$query'::real[]::vector::sparsevec) FROM tst");
is($expected, $actual);

View File

@@ -45,6 +45,10 @@ for my $function (@functions)
my $actual = $node->safe_psql("postgres", "SELECT $function(v::halfvec, '$query'::vector::halfvec) FROM tst");
is($expected, $actual, "halfvec $function");
# Test minivec
$actual = $node->safe_psql("postgres", "SELECT $function(v::minivec, '$query'::vector::minivec) FROM tst");
is($expected, $actual, "minivec $function");
# Test sparsevec
$actual = $node->safe_psql("postgres", "SELECT $function(v::sparsevec, '$query'::vector::sparsevec) FROM tst");
is($expected, $actual, "sparsevec $function");

View File

@@ -12,8 +12,8 @@ $node->start;
# Create extension
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
my @types = ("vector", "halfvec", "sparsevec");
my @inputs = ("[1.23,4.56,7.89]", "[1.23,4.56,7.89]", "{1:1.23,2:4.56,3:7.89}/3");
my @types = ("vector", "halfvec", "minivec", "sparsevec");
my @inputs = ("[1.23,4.56,7.89]", "[1.23,4.56,7.89]", "[1.23,4.56,7.89]", "{1:1.23,2:4.56,3:7.89}/3");
my @subs = (" ", " ", ",", ":", "-", "1", "9", "\0", "2147483648", "-2147483649");
for my $i (0 .. $#types)

View File

@@ -1,60 +0,0 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my @dims = (384, 1536);
my $limit = 10;
# Initialize node
my $node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
for my $dim (@dims)
{
my $array_sql = join(",", ('random()') x $dim);
# Create table and index
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(1, 2000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING hnsw (v vector_l2_ops);");
$node->safe_psql("postgres", "ANALYZE tst;");
# Generate query
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
my $query = "[" . join(",", @r) . "]";
my $explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
# 3x the rows are needed for distance filters
# since the planner uses DEFAULT_INEQ_SEL for the selectivity (should be 1)
# Recreate index for performance
$node->safe_psql("postgres", "DROP INDEX idx;");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(2001, 6000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING hnsw (v vector_l2_ops);");
$node->safe_psql("postgres", "ANALYZE tst;");
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE v <-> '$query' < 1 ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
$node->safe_psql("postgres", "DROP TABLE tst;");
}
done_testing();

View File

@@ -0,0 +1,132 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my $node;
my @queries = ();
my @expected;
my $limit = 20;
my $dim = 10;
my $array_sql = join(",", ('2 * random() * random()') x $dim);
sub test_recall
{
my ($min, $operator) = @_;
my $correct = 0;
my $total = 0;
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v $operator '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan/);
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SELECT i FROM tst ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
my @expected_ids = split("\n", $expected[$i]);
foreach (@expected_ids)
{
if (exists($actual_set{$_}))
{
$correct++;
}
$total++;
}
}
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
$node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v minivec($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(1, 10000) i;"
);
# Generate queries
for (1 .. 20)
{
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
push(@queries, "[" . join(",", @r) . "]");
}
# Check each index type
my @operators = ("<->", "<#>", "<=>", "<+>");
my @opclasses = ("minivec_l2_ops", "minivec_ip_ops", "minivec_cosine_ops", "minivec_l1_ops");
for my $i (0 .. $#operators)
{
my $operator = $operators[$i];
my $opclass = $opclasses[$i];
# Get exact results
@expected = ();
foreach (@queries)
{
my $res = $node->safe_psql("postgres", "SELECT i FROM tst ORDER BY v $operator '$_' LIMIT $limit;");
push(@expected, $res);
}
# Build index serially
$node->safe_psql("postgres", qq(
SET max_parallel_maintenance_workers = 0;
CREATE INDEX idx ON tst USING hnsw (v $opclass);
));
# Test approximate results
my $min = 0.98;
test_recall($min, $operator);
$node->safe_psql("postgres", "DROP INDEX idx;");
# Build index in parallel in memory
my ($ret, $stdout, $stderr) = $node->psql("postgres", qq(
SET client_min_messages = DEBUG;
SET min_parallel_table_scan_size = 1;
CREATE INDEX idx ON tst USING hnsw (v $opclass);
));
is($ret, 0, $stderr);
like($stderr, qr/using \d+ parallel workers/);
# Test approximate results
test_recall($min, $operator);
$node->safe_psql("postgres", "DROP INDEX idx;");
# Build index in parallel on disk
# Set parallel_workers on table to use workers with low maintenance_work_mem
($ret, $stdout, $stderr) = $node->psql("postgres", qq(
ALTER TABLE tst SET (parallel_workers = 2);
SET client_min_messages = DEBUG;
SET maintenance_work_mem = '4MB';
CREATE INDEX idx ON tst USING hnsw (v $opclass);
ALTER TABLE tst RESET (parallel_workers);
));
is($ret, 0, $stderr);
like($stderr, qr/using \d+ parallel workers/);
like($stderr, qr/hnsw graph no longer fits into maintenance_work_mem/);
$node->safe_psql("postgres", "DROP INDEX idx;");
}
done_testing();

View File

@@ -0,0 +1,113 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my $node;
my @queries = ();
my @expected;
my $limit = 20;
my $dim = 10;
my $array_sql = join(",", ('2 * random() * random()') x $dim);
sub test_recall
{
my ($min, $operator) = @_;
my $correct = 0;
my $total = 0;
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v $operator '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan/);
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SELECT i FROM tst ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
my @expected_ids = split("\n", $expected[$i]);
foreach (@expected_ids)
{
if (exists($actual_set{$_}))
{
$correct++;
}
$total++;
}
}
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
$node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i serial, v minivec($dim));");
# Generate queries
for (1 .. 20)
{
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
push(@queries, "[" . join(",", @r) . "]");
}
# Check each index type
my @operators = ("<->", "<#>", "<=>", "<+>");
my @opclasses = ("minivec_l2_ops", "minivec_ip_ops", "minivec_cosine_ops", "minivec_l1_ops");
for my $i (0 .. $#operators)
{
my $operator = $operators[$i];
my $opclass = $opclasses[$i];
# Add index
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING hnsw (v $opclass);");
# Use concurrent inserts
$node->pgbench(
"--no-vacuum --client=10 --transactions=1000",
0,
[qr{actually processed}],
[qr{^$}],
"concurrent INSERTs",
{
"040_hnsw_minivec_insert_recall_$opclass" => "INSERT INTO tst (v) VALUES (ARRAY[$array_sql]);"
}
);
# Get exact results
@expected = ();
foreach (@queries)
{
my $res = $node->safe_psql("postgres", qq(
SET enable_indexscan = off;
SELECT i FROM tst ORDER BY v $operator '$_' LIMIT $limit;
));
push(@expected, $res);
}
# Test approximate results
my $min = 0.98;
test_recall($min, $operator);
$node->safe_psql("postgres", "DROP INDEX idx;");
$node->safe_psql("postgres", "TRUNCATE tst;");
}
done_testing();

View File

@@ -1,50 +0,0 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my @dims = (384, 1536);
my $limit = 10;
# Initialize node
my $node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
for my $dim (@dims)
{
my $array_sql = join(",", ('random()') x $dim);
# Create table and index
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(1, 5000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING ivfflat (v vector_l2_ops) WITH (lists = 5);");
$node->safe_psql("postgres", "ANALYZE tst;");
# Generate query
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
my $query = "[" . join(",", @r) . "]";
my $explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
$explain = $node->safe_psql("postgres", qq(
EXPLAIN ANALYZE SELECT i FROM tst WHERE v <-> '$query' < 1 ORDER BY v <-> '$query' LIMIT $limit;
));
like($explain, qr/Index Scan using idx/);
$node->safe_psql("postgres", "DROP TABLE tst;");
}
done_testing();

View File

@@ -0,0 +1,97 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my $node;
my @queries = ();
my @expected;
my $limit = 20;
sub test_recall
{
my ($min, $ef_search, $test_name) = @_;
my $correct = 0;
my $total = 0;
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET hnsw.ef_search = $ef_search;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan/);
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET hnsw.ef_search = $ef_search;
SELECT i FROM tst ORDER BY v <-> '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
my @expected_ids = split("\n", $expected[$i]);
foreach (@expected_ids)
{
if (exists($actual_set{$_}))
{
$correct++;
}
$total++;
}
}
cmp_ok($correct / $total, ">=", $min, $test_name);
}
# Initialize node
$node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v minivec(3));");
$node->safe_psql("postgres", "ALTER TABLE tst SET (autovacuum_enabled = false);");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[random(), random(), random()] FROM generate_series(1, 10000) i;"
);
# Add index
$node->safe_psql("postgres", "CREATE INDEX ON tst USING hnsw (v minivec_l2_ops) WITH (m = 4, ef_construction = 8);");
# Delete data
$node->safe_psql("postgres", "DELETE FROM tst WHERE i > 2500;");
# Generate queries
for (1 .. 20)
{
my $r1 = rand();
my $r2 = rand();
my $r3 = rand();
push(@queries, "[$r1,$r2,$r3]");
}
# Get exact results
@expected = ();
foreach (@queries)
{
my $res = $node->safe_psql("postgres", qq(
SET enable_indexscan = off;
SELECT i FROM tst ORDER BY v <-> '$_' LIMIT $limit;
));
push(@expected, $res);
}
test_recall(0.18, $limit, "before vacuum");
test_recall(0.93, 100, "before vacuum");
# TODO Test concurrent inserts with vacuum
$node->safe_psql("postgres", "VACUUM tst;");
test_recall(0.95, $limit, "after vacuum");
done_testing();

View File

@@ -1,197 +0,0 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my $node;
my @queries = ();
my @where = ();
my @expected;
my $limit = 20;
my $dim = 3;
my $array_sql = join(",", ('random()') x $dim);
my $nc = 100;
my $nc2 = 10;
sub test_recall
{
my ($probes, $min, $operator) = @_;
my $correct = 0;
my $total = 0;
for my $j (0 .. 2)
{
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
EXPLAIN ANALYZE SELECT i FROM tst WHERE $where[$j] ORDER BY v $operator '$queries[$j]' LIMIT $limit;
));
like($explain, qr/Index Cond/);
}
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
SELECT i FROM tst WHERE $where[$i] ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
is(scalar(@actual_ids), $limit);
my @expected_ids = split("\n", $expected[$i]);
foreach (@expected_ids)
{
if (exists($actual_set{$_}))
{
$correct++;
}
$total++;
}
}
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
$node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim), c int4, c2 int4);");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql], i % $nc, i % $nc2 FROM generate_series(1, 50000) i;"
);
# Generate queries
for my $i (1 .. 100)
{
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
push(@queries, "[" . join(",", @r) . "]");
if ($i % 3 == 0)
{
my $c = int(rand() * $nc);
push(@where, "c = $c");
}
elsif ($i % 3 == 1)
{
my $c2 = int(rand() * $nc2);
push(@where, "c2 = $c2");
}
else
{
# use c2 to ensure results
my $c2 = int(rand() * $nc2);
push(@where, "c = $c2 AND c2 = $c2");
}
}
# Add index
$node->safe_psql("postgres", qq(
CREATE INDEX ON tst USING ivfflat (v vector_l2_ops, c, c2) WITH (lists = 100);
));
# Insert more rows
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql], i % $nc, i % $nc2 FROM generate_series(1, 50000) i;"
);
# Get exact results
@expected = ();
for my $i (0 .. $#queries)
{
my $res = $node->safe_psql("postgres", qq(
SET enable_indexscan = off;
SELECT i FROM tst WHERE $where[$i] ORDER BY v <-> '$queries[$i]' LIMIT $limit;
));
push(@expected, $res);
}
# Test recall
test_recall(10, 0.99, '<->');
# Test vacuum
$node->safe_psql("postgres", "DELETE FROM tst WHERE c > 5;");
$node->safe_psql("postgres", "VACUUM tst;");
# Test less than
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c < 10 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(c < 10\)/);
# Test less than or equal
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c <= 10 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(c <= 10\)/);
# Test greater than or equal
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c >= 90 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(c >= 90\)/);
# Test greater than
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c > 90 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(c > 90\)/);
# Test multiple attribute columns
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c = 1 AND c2 = 1 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(\(c = 1\) AND \(c2 = 1\)\)/);
# Test only last attribute column
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c2 = 1 ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Cond: \(c2 = 1\)/);
# Test only vector column
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v <-> '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan/);
# Test only attribute columns
$explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
EXPLAIN ANALYZE SELECT i FROM tst WHERE c = 1;
));
like($explain, qr/Seq Scan/);
# Test columns
my ($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING ivfflat (c);");
like($stderr, qr/first column must be a vector/);
($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING ivfflat (c, v vector_cosine_ops);");
like($stderr, qr/first column must be a vector/);
($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING ivfflat (v vector_cosine_ops, c, c, c);");
like($stderr, qr/index cannot have more than three columns/);
($ret, $stdout, $stderr) = $node->psql("postgres", "CREATE INDEX ON tst USING ivfflat (v vector_cosine_ops, v vector_cosine_ops);");
like($stderr, qr/column 2 cannot be a vector/);
done_testing();

View File

@@ -0,0 +1,58 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
# Initialize node
my $node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (v minivec(3));");
sub insert_vectors
{
for my $i (1 .. 20)
{
$node->safe_psql("postgres", "INSERT INTO tst VALUES ('[1,1,1]');");
}
}
sub test_duplicates
{
my $res = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET hnsw.ef_search = 1;
SELECT COUNT(*) FROM (SELECT * FROM tst ORDER BY v <-> '[1,1,1]') t;
));
is($res, 10);
}
# Test duplicates with build
insert_vectors();
$node->safe_psql("postgres", "CREATE INDEX idx ON tst USING hnsw (v minivec_l2_ops);");
test_duplicates();
# Reset
$node->safe_psql("postgres", "TRUNCATE tst;");
# Test duplicates with inserts
insert_vectors();
test_duplicates();
# Test fallback path for inserts
$node->pgbench(
"--no-vacuum --client=5 --transactions=100",
0,
[qr{actually processed}],
[qr{^$}],
"concurrent INSERTs",
{
"042_hnsw_minivec_duplicates" => "INSERT INTO tst VALUES ('[1,1,1]');"
}
);
done_testing();

View File

@@ -0,0 +1,154 @@
use strict;
use warnings FATAL => 'all';
use PostgreSQL::Test::Cluster;
use PostgreSQL::Test::Utils;
use Test::More;
my $node;
my @queries = ();
my @expected;
my $limit = 20;
my $dim = 10;
my $array_sql = join(",", ('random()') x $dim);
sub test_recall
{
my ($probes, $min, $operator) = @_;
my $correct = 0;
my $total = 0;
my $explain = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
EXPLAIN ANALYZE SELECT i FROM tst ORDER BY v $operator '$queries[0]' LIMIT $limit;
));
like($explain, qr/Index Scan using idx on tst/);
for my $i (0 .. $#queries)
{
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
SELECT i FROM tst ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my @expected_ids = split("\n", $expected[$i]);
my %expected_set = map { $_ => 1 } @expected_ids;
foreach (@actual_ids)
{
if (exists($expected_set{$_}))
{
$correct++;
}
}
$total += $limit;
}
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
$node = PostgreSQL::Test::Cluster->new('node');
$node->init;
$node->start;
# Create table
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v minivec($dim));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[$array_sql] FROM generate_series(1, 100000) i;"
);
# Generate queries
for (1 .. 20)
{
my @r = ();
for (1 .. $dim)
{
push(@r, rand());
}
push(@queries, "[" . join(",", @r) . "]");
}
# Check each index type
my @operators = ("<->", "<#>", "<=>");
my @opclasses = ("minivec_l2_ops", "minivec_ip_ops", "minivec_cosine_ops");
for my $i (0 .. $#operators)
{
my $operator = $operators[$i];
my $opclass = $opclasses[$i];
# Get exact results
@expected = ();
foreach (@queries)
{
my $res = $node->safe_psql("postgres", qq(
WITH top AS (
SELECT v $operator '$_' AS distance FROM tst ORDER BY distance LIMIT $limit
)
SELECT i FROM tst WHERE (v $operator '$_') <= (SELECT MAX(distance) FROM top)
));
push(@expected, $res);
}
# Build index serially
$node->safe_psql("postgres", qq(
SET max_parallel_maintenance_workers = 0;
CREATE INDEX idx ON tst USING ivfflat (v $opclass);
));
# Test approximate results
if ($operator ne "<#>")
{
# TODO Fix test (uniform random vectors all have similar inner product)
test_recall(1, 0.33, $operator);
test_recall(10, 0.93, $operator);
}
# Test probes equals lists
if ($operator eq "<=>")
{
test_recall(100, 0.98, $operator);
}
else
{
test_recall(100, 1.00, $operator);
}
$node->safe_psql("postgres", "DROP INDEX idx;");
# Build index in parallel
my ($ret, $stdout, $stderr) = $node->psql("postgres", qq(
SET client_min_messages = DEBUG;
SET min_parallel_table_scan_size = 1;
CREATE INDEX idx ON tst USING ivfflat (v $opclass);
));
is($ret, 0, $stderr);
like($stderr, qr/using \d+ parallel workers/);
# Test approximate results
if ($operator ne "<#>")
{
# TODO Fix test (uniform random vectors all have similar inner product)
test_recall(1, 0.33, $operator);
test_recall(10, 0.93, $operator);
}
# Test probes equals lists
if ($operator eq "<=>")
{
test_recall(100, 0.98, $operator);
}
else
{
test_recall(100, 1.00, $operator);
}
$node->safe_psql("postgres", "DROP INDEX idx;");
}
done_testing();