mirror of
https://github.com/pgvector/pgvector.git
synced 2026-07-22 12:07:34 +08:00
Compare commits
1 Commits
bitvector
...
hnsw-suppo
| Author | SHA1 | Date | |
|---|---|---|---|
|
eb9fa5cf8c |
22
.github/workflows/build.yml
vendored
22
.github/workflows/build.yml
vendored
@@ -28,7 +28,7 @@ jobs:
|
||||
dev-files: true
|
||||
- run: make
|
||||
env:
|
||||
PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
|
||||
PG_CFLAGS: -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
|
||||
- run: |
|
||||
export PG_CONFIG=`which pg_config`
|
||||
sudo --preserve-env=PG_CONFIG make install
|
||||
@@ -49,7 +49,7 @@ jobs:
|
||||
postgres-version: 14
|
||||
- run: make
|
||||
env:
|
||||
PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter
|
||||
PG_CFLAGS: -Wall -Wextra -Werror -Wno-unused-parameter
|
||||
- run: make install
|
||||
- run: make installcheck
|
||||
- if: ${{ failure() }}
|
||||
@@ -60,9 +60,7 @@ jobs:
|
||||
wget -q https://github.com/postgres/postgres/archive/refs/tags/REL_14_10.tar.gz
|
||||
tar xf REL_14_10.tar.gz
|
||||
- run: make prove_installcheck PROVE_FLAGS="-I ./postgres-REL_14_10/src/test/perl" PERL5LIB="/Users/runner/perl5/lib/perl5"
|
||||
- run: make clean && /usr/local/opt/llvm@15/bin/scan-build --status-bugs make
|
||||
env:
|
||||
PG_CFLAGS: -DUSE_ASSERT_CHECKING
|
||||
- run: make clean && /usr/local/opt/llvm@15/bin/scan-build --status-bugs make PG_CFLAGS="-DUSE_ASSERT_CHECKING"
|
||||
windows:
|
||||
runs-on: windows-latest
|
||||
if: ${{ !startsWith(github.ref_name, 'mac') }}
|
||||
@@ -73,7 +71,6 @@ jobs:
|
||||
postgres-version: 14
|
||||
- run: |
|
||||
call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat" && ^
|
||||
cd %TEMP% && ^
|
||||
nmake /NOLOGO /F Makefile.win && ^
|
||||
nmake /NOLOGO /F Makefile.win install && ^
|
||||
nmake /NOLOGO /F Makefile.win installcheck && ^
|
||||
@@ -100,15 +97,4 @@ jobs:
|
||||
sudo -u postgres make installcheck
|
||||
sudo -u postgres make prove_installcheck
|
||||
env:
|
||||
PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
|
||||
valgrind:
|
||||
if: ${{ !startsWith(github.ref_name, 'mac') && !startsWith(github.ref_name, 'windows') }}
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- uses: actions/checkout@v4
|
||||
- uses: ankane/setup-postgres-valgrind@v1
|
||||
with:
|
||||
postgres-version: 16
|
||||
- run: make
|
||||
- run: sudo --preserve-env=PG_CONFIG make install
|
||||
- run: make installcheck
|
||||
PG_CFLAGS: -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
|
||||
|
||||
15
CHANGELOG.md
15
CHANGELOG.md
@@ -1,19 +1,6 @@
|
||||
## 0.7.0 (unreleased)
|
||||
|
||||
- Added support for binary vectors to HNSW
|
||||
- Added `hamming_distance` function
|
||||
- Added `jaccard_distance` function
|
||||
- Added `quantize_binary` function
|
||||
|
||||
## 0.6.2 (2024-03-18)
|
||||
|
||||
- Reduced lock contention with parallel HNSW index builds
|
||||
|
||||
## 0.6.1 (2024-03-04)
|
||||
## 0.6.1 (unreleased)
|
||||
|
||||
- Fixed error with `ANALYZE` and vectors with different dimensions
|
||||
- Fixed segmentation fault with `shared_preload_libraries`
|
||||
- Fixed vector subtraction being marked as commutative
|
||||
|
||||
## 0.6.0 (2024-01-29)
|
||||
|
||||
|
||||
2
LICENSE
2
LICENSE
@@ -1,4 +1,4 @@
|
||||
Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
|
||||
Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
|
||||
|
||||
Portions Copyright (c) 1994, The Regents of the University of California
|
||||
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
"name": "vector",
|
||||
"abstract": "Open-source vector similarity search for Postgres",
|
||||
"description": "Supports L2 distance, inner product, and cosine distance",
|
||||
"version": "0.6.2",
|
||||
"version": "0.6.0",
|
||||
"maintainer": [
|
||||
"Andrew Kane <andrew@ankane.org>"
|
||||
],
|
||||
@@ -20,7 +20,7 @@
|
||||
"vector": {
|
||||
"file": "sql/vector.sql",
|
||||
"docfile": "README.md",
|
||||
"version": "0.6.2",
|
||||
"version": "0.6.0",
|
||||
"abstract": "Open-source vector similarity search for Postgres"
|
||||
}
|
||||
},
|
||||
|
||||
10
Makefile
10
Makefile
@@ -1,9 +1,9 @@
|
||||
EXTENSION = vector
|
||||
EXTVERSION = 0.6.2
|
||||
EXTVERSION = 0.6.0
|
||||
|
||||
MODULE_big = vector
|
||||
DATA = $(wildcard sql/*--*.sql)
|
||||
OBJS = src/bitvector.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/vector.o
|
||||
OBJS = 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/vector.o
|
||||
HEADERS = src/vector.h
|
||||
|
||||
TESTS = $(wildcard test/sql/*.sql)
|
||||
@@ -71,9 +71,11 @@ PG_MAJOR ?= 16
|
||||
.PHONY: docker
|
||||
|
||||
docker:
|
||||
docker build --pull --no-cache --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .
|
||||
docker build --pull --no-cache --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) .
|
||||
docker build --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .
|
||||
|
||||
.PHONY: docker-release
|
||||
|
||||
docker-release:
|
||||
docker buildx build --push --pull --no-cache --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .
|
||||
docker buildx build --push --pull --no-cache --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR) .
|
||||
docker buildx build --push --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) .
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
EXTENSION = vector
|
||||
EXTVERSION = 0.6.2
|
||||
EXTVERSION = 0.6.0
|
||||
|
||||
OBJS = src\bitvector.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\vector.obj
|
||||
OBJS = 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\vector.obj
|
||||
HEADERS = src\vector.h
|
||||
|
||||
REGRESS = btree cast copy functions input ivfflat_cosine ivfflat_ip ivfflat_l2 ivfflat_options ivfflat_unlogged
|
||||
|
||||
198
README.md
198
README.md
@@ -5,7 +5,7 @@ Open-source vector similarity search for Postgres
|
||||
Store your vectors with the rest of your data. Supports:
|
||||
|
||||
- exact and approximate nearest neighbor search
|
||||
- L2 distance, inner product, cosine distance, and more
|
||||
- L2 distance, inner product, and cosine distance
|
||||
- any [language](#languages) with a Postgres client
|
||||
|
||||
Plus [ACID](https://en.wikipedia.org/wiki/ACID) compliance, point-in-time recovery, JOINs, and all of the other [great features](https://www.postgresql.org/about/) of Postgres
|
||||
@@ -20,15 +20,15 @@ Compile and install the extension (supports Postgres 12+)
|
||||
|
||||
```sh
|
||||
cd /tmp
|
||||
git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
|
||||
git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
|
||||
cd pgvector
|
||||
make
|
||||
make install # may need sudo
|
||||
```
|
||||
|
||||
See the [installation notes](#installation-notes---linux-and-mac) if you run into issues
|
||||
See the [installation notes](#installation-notes) if you run into issues
|
||||
|
||||
You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), [pkg](#pkg), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres). There are also instructions for [GitHub Actions](https://github.com/pgvector/setup-pgvector).
|
||||
You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#pgxn), [APT](#apt), [Yum](#yum), or [conda-forge](#conda-forge), and it comes preinstalled with [Postgres.app](#postgresapp) and many [hosted providers](#hosted-postgres). There are also instructions for [GitHub Actions](https://github.com/pgvector/setup-pgvector).
|
||||
|
||||
### Windows
|
||||
|
||||
@@ -44,15 +44,12 @@ Then use `nmake` to build:
|
||||
|
||||
```cmd
|
||||
set "PGROOT=C:\Program Files\PostgreSQL\16"
|
||||
cd %TEMP%
|
||||
git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
|
||||
git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
|
||||
cd pgvector
|
||||
nmake /F Makefile.win
|
||||
nmake /F Makefile.win install
|
||||
```
|
||||
|
||||
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).
|
||||
|
||||
## Getting Started
|
||||
@@ -105,12 +102,6 @@ Insert vectors
|
||||
INSERT INTO items (embedding) VALUES ('[1,2,3]'), ('[4,5,6]');
|
||||
```
|
||||
|
||||
Or load vectors in bulk using `COPY` ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/bulk_loading.py))
|
||||
|
||||
```sql
|
||||
COPY items (embedding) FROM STDIN WITH (FORMAT BINARY);
|
||||
```
|
||||
|
||||
Upsert vectors
|
||||
|
||||
```sql
|
||||
@@ -221,19 +212,7 @@ Cosine distance
|
||||
CREATE INDEX ON items USING hnsw (embedding vector_cosine_ops);
|
||||
```
|
||||
|
||||
Hamming distance - added in 0.7.0
|
||||
|
||||
```sql
|
||||
CREATE INDEX ON items USING hnsw (embedding bit_hamming_ops);
|
||||
```
|
||||
|
||||
Jaccard distance - added in 0.7.0
|
||||
|
||||
```sql
|
||||
CREATE INDEX ON items USING hnsw (embedding bit_jaccard_ops);
|
||||
```
|
||||
|
||||
Vectors with up to 2,000 dimensions can be indexed, or bit vectors with up to 64,000 dimensions.
|
||||
Vectors with up to 2,000 dimensions can be indexed.
|
||||
|
||||
### Index Options
|
||||
|
||||
@@ -285,8 +264,6 @@ HINT: Increase maintenance_work_mem to speed up builds.
|
||||
|
||||
Note: Do not set `maintenance_work_mem` so high that it exhausts the memory on the server
|
||||
|
||||
Like other index types, it’s faster to create an index after loading your initial data
|
||||
|
||||
Starting with 0.6.0, you can also speed up index creation by increasing the number of parallel workers (2 by default)
|
||||
|
||||
```sql
|
||||
@@ -425,39 +402,13 @@ You can use [Reciprocal Rank Fusion](https://github.com/pgvector/pgvector-python
|
||||
|
||||
## Performance
|
||||
|
||||
### Tuning
|
||||
|
||||
Use a tool like [PgTune](https://pgtune.leopard.in.ua/) to set initial values for Postgres server parameters.
|
||||
|
||||
### Loading
|
||||
|
||||
Use `COPY` for bulk loading data ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/bulk_loading.py)).
|
||||
|
||||
```sql
|
||||
COPY items (embedding) FROM STDIN WITH (FORMAT BINARY);
|
||||
```
|
||||
|
||||
Add any indexes *after* loading the initial data for best performance.
|
||||
|
||||
### Indexing
|
||||
|
||||
See index build time for [HNSW](#index-build-time) and [IVFFlat](#index-build-time-1).
|
||||
|
||||
In production environments, create indexes concurrently to avoid blocking writes.
|
||||
|
||||
```sql
|
||||
CREATE INDEX CONCURRENTLY ...
|
||||
```
|
||||
|
||||
### Querying
|
||||
|
||||
Use `EXPLAIN ANALYZE` to debug performance.
|
||||
|
||||
```sql
|
||||
EXPLAIN ANALYZE SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 5;
|
||||
```
|
||||
|
||||
#### Exact Search
|
||||
### Exact Search
|
||||
|
||||
To speed up queries without an index, increase `max_parallel_workers_per_gather`.
|
||||
|
||||
@@ -471,7 +422,7 @@ If vectors are normalized to length 1 (like [OpenAI embeddings](https://platform
|
||||
SELECT * FROM items ORDER BY embedding <#> '[3,1,2]' LIMIT 5;
|
||||
```
|
||||
|
||||
#### Approximate Search
|
||||
### Approximate Search
|
||||
|
||||
To speed up queries with an IVFFlat index, increase the number of inverted lists (at the expense of recall).
|
||||
|
||||
@@ -479,7 +430,7 @@ To speed up queries with an IVFFlat index, increase the number of inverted lists
|
||||
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 1000);
|
||||
```
|
||||
|
||||
### Vacuuming
|
||||
## Vacuuming
|
||||
|
||||
Vacuuming can take a while for HNSW indexes. Speed it up by reindexing first.
|
||||
|
||||
@@ -488,41 +439,6 @@ REINDEX INDEX CONCURRENTLY index_name;
|
||||
VACUUM table_name;
|
||||
```
|
||||
|
||||
## Monitoring
|
||||
|
||||
Monitor performance with [pg_stat_statements](https://www.postgresql.org/docs/current/pgstatstatements.html) (be sure to add it to `shared_preload_libraries`).
|
||||
|
||||
```sql
|
||||
CREATE EXTENSION pg_stat_statements;
|
||||
```
|
||||
|
||||
Get the most time-consuming queries with:
|
||||
|
||||
```sql
|
||||
SELECT query, calls, ROUND((total_plan_time + total_exec_time) / calls) AS avg_time_ms,
|
||||
ROUND((total_plan_time + total_exec_time) / 60000) AS total_time_min
|
||||
FROM pg_stat_statements ORDER BY total_plan_time + total_exec_time DESC LIMIT 20;
|
||||
```
|
||||
|
||||
Note: Replace `total_plan_time + total_exec_time` with `total_time` for Postgres < 13
|
||||
|
||||
Monitor recall by comparing results from approximate search with exact search.
|
||||
|
||||
```sql
|
||||
BEGIN;
|
||||
SET LOCAL enable_indexscan = off; -- use exact search
|
||||
SELECT ...
|
||||
COMMIT;
|
||||
```
|
||||
|
||||
## Scaling
|
||||
|
||||
Scale pgvector the same way you scale Postgres.
|
||||
|
||||
Scale vertically by increasing memory, CPU, and storage on a single instance. Use existing tools to [tune parameters](#tuning) and [monitor performance](#monitoring).
|
||||
|
||||
Scale horizontally with [replicas](https://www.postgresql.org/docs/current/hot-standby.html), or use [Citus](https://github.com/citusdata/citus) or another approach for sharding ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/citus.py)).
|
||||
|
||||
## Languages
|
||||
|
||||
Use pgvector from any language with a Postgres client. You can even generate and store vectors in one language and query them in another.
|
||||
@@ -616,18 +532,6 @@ and query with:
|
||||
SELECT * FROM items ORDER BY embedding::vector(3) <-> '[3,1,2]' LIMIT 5;
|
||||
```
|
||||
|
||||
#### Are binary vectors supported?
|
||||
|
||||
You can store binary vectors and perform exact nearest neighbor search by Hamming distance in Postgres without an extension ([example](https://github.com/pgvector/pgvector-python/blob/master/examples/hash_image_search.py)).
|
||||
|
||||
```tsql
|
||||
CREATE TABLE items (id bigserial PRIMARY KEY, embedding bit(3));
|
||||
INSERT INTO items (embedding) VALUES (B'000'), (B'111');
|
||||
SELECT * FROM items ORDER BY bit_count(embedding # B'101') LIMIT 5;
|
||||
```
|
||||
|
||||
Indexing is not currently supported.
|
||||
|
||||
#### Do indexes need to fit into memory?
|
||||
|
||||
No, but like other index types, you’ll likely see better performance if they do. You can get the size of an index with:
|
||||
@@ -640,17 +544,7 @@ SELECT pg_size_pretty(pg_relation_size('index_name'));
|
||||
|
||||
#### Why isn’t a query using an index?
|
||||
|
||||
The query needs to have an `ORDER BY` and `LIMIT`, and the `ORDER BY` must be the result of a distance operator, not an expression.
|
||||
|
||||
```sql
|
||||
-- index
|
||||
ORDER BY embedding <=> '[3,1,2]' LIMIT 5;
|
||||
|
||||
-- no index
|
||||
ORDER BY 1 - (embedding <=> '[3,1,2]') DESC LIMIT 5;
|
||||
```
|
||||
|
||||
You can encourage the planner to use an index for a query with:
|
||||
The cost estimation in pgvector < 0.4.3 does not always work well with the planner. You can encourage the planner to use an index for a query with:
|
||||
|
||||
```sql
|
||||
BEGIN;
|
||||
@@ -679,10 +573,6 @@ or choose to store vectors inline:
|
||||
ALTER TABLE items ALTER COLUMN embedding SET STORAGE PLAIN;
|
||||
```
|
||||
|
||||
#### Why are there less results for a query after adding an HNSW index?
|
||||
|
||||
Results are limited by the size of the dynamic candidate list (`hnsw.ef_search`). There may be even less results due to dead tuples or filtering conditions in the query. We recommend setting `hnsw.ef_search` to at least twice the `LIMIT` of the query. If you need more than 500 results, use an IVFFlat index instead.
|
||||
|
||||
#### Why are there less results for a query after adding an IVFFlat index?
|
||||
|
||||
The index was likely created with too little data for the number of lists. Drop the index until the table has more data.
|
||||
@@ -691,8 +581,6 @@ The index was likely created with too little data for the number of lists. Drop
|
||||
DROP INDEX index_name;
|
||||
```
|
||||
|
||||
Results can also be limited by the number of probes (`ivfflat.probes`).
|
||||
|
||||
## Reference
|
||||
|
||||
### Vector Type
|
||||
@@ -718,7 +606,6 @@ cosine_distance(vector, vector) → double precision | cosine distance |
|
||||
inner_product(vector, vector) → double precision | inner product |
|
||||
l2_distance(vector, vector) → double precision | Euclidean distance |
|
||||
l1_distance(vector, vector) → double precision | taxicab distance | 0.5.0
|
||||
quantize_binary(vector) → bit | quantize | 0.7.0
|
||||
vector_dims(vector) → integer | number of dimensions |
|
||||
vector_norm(vector) → double precision | Euclidean norm |
|
||||
|
||||
@@ -729,21 +616,7 @@ Function | Description | Added
|
||||
avg(vector) → vector | average |
|
||||
sum(vector) → vector | sum | 0.5.0
|
||||
|
||||
### Bit Operators
|
||||
|
||||
Operator | Description | Added
|
||||
--- | --- | ---
|
||||
<~> | Hamming distance | 0.7.0
|
||||
<%> | Jaccard distance | 0.7.0
|
||||
|
||||
### Bit Functions
|
||||
|
||||
Function | Description | Added
|
||||
--- | --- | ---
|
||||
hamming_distance(bit, bit) → double precision | Hamming distance | 0.7.0
|
||||
jaccard_distance(bit, bit) → double precision | Jaccard distance | 0.7.0
|
||||
|
||||
## Installation Notes - Linux and Mac
|
||||
## Installation Notes
|
||||
|
||||
### Postgres Location
|
||||
|
||||
@@ -783,26 +656,6 @@ Note: Replace `16` with your Postgres server version
|
||||
|
||||
If compilation fails and the output includes `warning: no such sysroot directory` on Mac, reinstall Xcode Command Line Tools.
|
||||
|
||||
### Portability
|
||||
|
||||
By default, pgvector compiles with `-march=native` on some platforms for best performance. However, this can lead to `Illegal instruction` errors if trying to run the compiled extension on a different machine.
|
||||
|
||||
To compile for portability, use:
|
||||
|
||||
```sh
|
||||
make OPTFLAGS=""
|
||||
```
|
||||
|
||||
## Installation Notes - Windows
|
||||
|
||||
### Missing Header
|
||||
|
||||
If compilation fails with `Cannot open include file: 'postgres.h': No such file or directory`, make sure `PGROOT` is correct.
|
||||
|
||||
### Permissions
|
||||
|
||||
If installation fails with `Access is denied`, re-run the installation instructions as an administrator.
|
||||
|
||||
## Additional Installation Methods
|
||||
|
||||
### Docker
|
||||
@@ -818,7 +671,7 @@ This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (r
|
||||
You can also build the image manually:
|
||||
|
||||
```sh
|
||||
git clone --branch v0.6.2 https://github.com/pgvector/pgvector.git
|
||||
git clone --branch v0.6.0 https://github.com/pgvector/pgvector.git
|
||||
cd pgvector
|
||||
docker build --build-arg PG_MAJOR=16 -t myuser/pgvector .
|
||||
```
|
||||
@@ -863,21 +716,6 @@ sudo dnf install pgvector_16
|
||||
|
||||
Note: Replace `16` with your Postgres server version
|
||||
|
||||
### pkg
|
||||
|
||||
Install the FreeBSD package with:
|
||||
|
||||
```sh
|
||||
pkg install postgresql15-pg_vector
|
||||
```
|
||||
|
||||
or the port with:
|
||||
|
||||
```sh
|
||||
cd /usr/ports/databases/pgvector
|
||||
make install
|
||||
```
|
||||
|
||||
### conda-forge
|
||||
|
||||
With Conda Postgres, install from [conda-forge](https://anaconda.org/conda-forge/pgvector) with:
|
||||
@@ -987,12 +825,6 @@ make installcheck REGRESS=functions # regression test
|
||||
make prove_installcheck PROVE_TESTS=test/t/001_ivfflat_wal.pl # TAP test
|
||||
```
|
||||
|
||||
To enable assertions:
|
||||
|
||||
```sh
|
||||
make clean && PG_CFLAGS="-DUSE_ASSERT_CHECKING" make && make install
|
||||
```
|
||||
|
||||
To enable benchmarking:
|
||||
|
||||
```sh
|
||||
@@ -1005,6 +837,12 @@ To show memory usage:
|
||||
make clean && PG_CFLAGS="-DHNSW_MEMORY -DIVFFLAT_MEMORY" make && make install
|
||||
```
|
||||
|
||||
To enable assertions:
|
||||
|
||||
```sh
|
||||
make clean && PG_CFLAGS="-DUSE_ASSERT_CHECKING" make && make install
|
||||
```
|
||||
|
||||
To get k-means metrics:
|
||||
|
||||
```sh
|
||||
|
||||
@@ -1,16 +0,0 @@
|
||||
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
|
||||
\echo Use "ALTER EXTENSION vector UPDATE TO '0.6.1'" to load this file. \quit
|
||||
|
||||
DROP OPERATOR - (vector, vector);
|
||||
|
||||
CREATE OPERATOR - (
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = vector_sub
|
||||
);
|
||||
|
||||
ALTER OPERATOR <= (vector, vector) SET (
|
||||
RESTRICT = scalarlesel, JOIN = scalarlejoinsel
|
||||
);
|
||||
|
||||
ALTER OPERATOR >= (vector, vector) SET (
|
||||
RESTRICT = scalargesel, JOIN = scalargejoinsel
|
||||
);
|
||||
@@ -1,2 +0,0 @@
|
||||
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
|
||||
\echo Use "ALTER EXTENSION vector UPDATE TO '0.6.2'" to load this file. \quit
|
||||
@@ -1,31 +0,0 @@
|
||||
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
|
||||
\echo Use "ALTER EXTENSION vector UPDATE TO '0.7.0'" to load this file. \quit
|
||||
|
||||
CREATE FUNCTION quantize_binary(vector) RETURNS bit
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE FUNCTION hamming_distance(bit, bit) RETURNS float8
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE FUNCTION jaccard_distance(bit, bit) RETURNS float8
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE OPERATOR <~> (
|
||||
LEFTARG = bit, RIGHTARG = bit, PROCEDURE = hamming_distance,
|
||||
COMMUTATOR = '<~>'
|
||||
);
|
||||
|
||||
CREATE OPERATOR <%> (
|
||||
LEFTARG = bit, RIGHTARG = bit, PROCEDURE = jaccard_distance,
|
||||
COMMUTATOR = '<%>'
|
||||
);
|
||||
|
||||
CREATE OPERATOR CLASS bit_hamming_ops
|
||||
FOR TYPE bit USING hnsw AS
|
||||
OPERATOR 1 <~> (bit, bit) FOR ORDER BY float_ops,
|
||||
FUNCTION 1 hamming_distance(bit, bit);
|
||||
|
||||
CREATE OPERATOR CLASS bit_jaccard_ops
|
||||
FOR TYPE bit USING hnsw AS
|
||||
OPERATOR 1 <%> (bit, bit) FOR ORDER BY float_ops,
|
||||
FUNCTION 1 jaccard_distance(bit, bit);
|
||||
@@ -58,9 +58,6 @@ CREATE FUNCTION vector_sub(vector, vector) RETURNS vector
|
||||
CREATE FUNCTION vector_mul(vector, vector) RETURNS vector
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE FUNCTION quantize_binary(vector) RETURNS bit
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
-- private functions
|
||||
|
||||
CREATE FUNCTION vector_lt(vector, vector) RETURNS bool
|
||||
@@ -183,7 +180,8 @@ CREATE OPERATOR + (
|
||||
);
|
||||
|
||||
CREATE OPERATOR - (
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = vector_sub
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = vector_sub,
|
||||
COMMUTATOR = -
|
||||
);
|
||||
|
||||
CREATE OPERATOR * (
|
||||
@@ -197,10 +195,11 @@ CREATE OPERATOR < (
|
||||
RESTRICT = scalarltsel, JOIN = scalarltjoinsel
|
||||
);
|
||||
|
||||
-- should use scalarlesel and scalarlejoinsel, but not supported in Postgres < 11
|
||||
CREATE OPERATOR <= (
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = vector_le,
|
||||
COMMUTATOR = >= , NEGATOR = > ,
|
||||
RESTRICT = scalarlesel, JOIN = scalarlejoinsel
|
||||
RESTRICT = scalarltsel, JOIN = scalarltjoinsel
|
||||
);
|
||||
|
||||
CREATE OPERATOR = (
|
||||
@@ -215,10 +214,11 @@ CREATE OPERATOR <> (
|
||||
RESTRICT = eqsel, JOIN = eqjoinsel
|
||||
);
|
||||
|
||||
-- should use scalargesel and scalargejoinsel, but not supported in Postgres < 11
|
||||
CREATE OPERATOR >= (
|
||||
LEFTARG = vector, RIGHTARG = vector, PROCEDURE = vector_ge,
|
||||
COMMUTATOR = <= , NEGATOR = < ,
|
||||
RESTRICT = scalargesel, JOIN = scalargejoinsel
|
||||
RESTRICT = scalargtsel, JOIN = scalargtjoinsel
|
||||
);
|
||||
|
||||
CREATE OPERATOR > (
|
||||
@@ -290,31 +290,3 @@ CREATE OPERATOR CLASS vector_cosine_ops
|
||||
OPERATOR 1 <=> (vector, vector) FOR ORDER BY float_ops,
|
||||
FUNCTION 1 vector_negative_inner_product(vector, vector),
|
||||
FUNCTION 2 vector_norm(vector);
|
||||
|
||||
-- bit functions
|
||||
|
||||
CREATE FUNCTION hamming_distance(bit, bit) RETURNS float8
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE FUNCTION jaccard_distance(bit, bit) RETURNS float8
|
||||
AS 'MODULE_PATHNAME' LANGUAGE C IMMUTABLE STRICT PARALLEL SAFE;
|
||||
|
||||
CREATE OPERATOR <~> (
|
||||
LEFTARG = bit, RIGHTARG = bit, PROCEDURE = hamming_distance,
|
||||
COMMUTATOR = '<~>'
|
||||
);
|
||||
|
||||
CREATE OPERATOR <%> (
|
||||
LEFTARG = bit, RIGHTARG = bit, PROCEDURE = jaccard_distance,
|
||||
COMMUTATOR = '<%>'
|
||||
);
|
||||
|
||||
CREATE OPERATOR CLASS bit_hamming_ops
|
||||
FOR TYPE bit USING hnsw AS
|
||||
OPERATOR 1 <~> (bit, bit) FOR ORDER BY float_ops,
|
||||
FUNCTION 1 hamming_distance(bit, bit);
|
||||
|
||||
CREATE OPERATOR CLASS bit_jaccard_ops
|
||||
FOR TYPE bit USING hnsw AS
|
||||
OPERATOR 1 <%> (bit, bit) FOR ORDER BY float_ops,
|
||||
FUNCTION 1 jaccard_distance(bit, bit);
|
||||
|
||||
@@ -1,90 +0,0 @@
|
||||
#include "postgres.h"
|
||||
|
||||
#include "bitvector.h"
|
||||
#include "port/pg_bitutils.h"
|
||||
#include "utils/varbit.h"
|
||||
|
||||
#if PG_VERSION_NUM >= 160000
|
||||
#include "varatt.h"
|
||||
#endif
|
||||
|
||||
/*
|
||||
* Allocate and initialize a new bit vector
|
||||
*/
|
||||
VarBit *
|
||||
InitBitVector(int dim)
|
||||
{
|
||||
VarBit *result;
|
||||
int size;
|
||||
|
||||
size = VARBITTOTALLEN(dim);
|
||||
result = (VarBit *) palloc0(size);
|
||||
SET_VARSIZE(result, size);
|
||||
VARBITLEN(result) = dim;
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
/*
|
||||
* Ensure same number of bits
|
||||
*/
|
||||
static inline void
|
||||
CheckBitLengths(uint32 aLen, uint32 bLen)
|
||||
{
|
||||
if (aLen != bLen)
|
||||
ereport(ERROR,
|
||||
(errcode(ERRCODE_DATA_EXCEPTION),
|
||||
errmsg("different bit lengths %u and %u", aLen, bLen)));
|
||||
}
|
||||
|
||||
/*
|
||||
* Get the Hamming distance between two bit strings
|
||||
*/
|
||||
PGDLLEXPORT PG_FUNCTION_INFO_V1(hamming_distance);
|
||||
Datum
|
||||
hamming_distance(PG_FUNCTION_ARGS)
|
||||
{
|
||||
VarBit *a = PG_GETARG_VARBIT_P(0);
|
||||
VarBit *b = PG_GETARG_VARBIT_P(1);
|
||||
unsigned char *ax = VARBITS(a);
|
||||
unsigned char *bx = VARBITS(b);
|
||||
uint64 distance = 0;
|
||||
|
||||
CheckBitLengths(VARBITLEN(a), VARBITLEN(b));
|
||||
|
||||
/* TODO Improve performance */
|
||||
for (uint32 i = 0; i < VARBITBYTES(a); i++)
|
||||
distance += pg_number_of_ones[ax[i] ^ bx[i]];
|
||||
|
||||
PG_RETURN_FLOAT8((double) distance);
|
||||
}
|
||||
|
||||
/*
|
||||
* Get the Jaccard distance between two bit strings
|
||||
*/
|
||||
PGDLLEXPORT PG_FUNCTION_INFO_V1(jaccard_distance);
|
||||
Datum
|
||||
jaccard_distance(PG_FUNCTION_ARGS)
|
||||
{
|
||||
VarBit *a = PG_GETARG_VARBIT_P(0);
|
||||
VarBit *b = PG_GETARG_VARBIT_P(1);
|
||||
unsigned char *ax = VARBITS(a);
|
||||
unsigned char *bx = VARBITS(b);
|
||||
uint64 ab = 0;
|
||||
uint64 aa;
|
||||
uint64 bb;
|
||||
|
||||
CheckBitLengths(VARBITLEN(a), VARBITLEN(b));
|
||||
|
||||
/* TODO Improve performance */
|
||||
for (uint32 i = 0; i < VARBITBYTES(a); i++)
|
||||
ab += pg_number_of_ones[ax[i] & bx[i]];
|
||||
|
||||
if (ab == 0)
|
||||
PG_RETURN_FLOAT8(1);
|
||||
|
||||
aa = pg_popcount((char *) ax, VARBITBYTES(a));
|
||||
bb = pg_popcount((char *) bx, VARBITBYTES(b));
|
||||
|
||||
PG_RETURN_FLOAT8(1 - (ab / ((double) (aa + bb - ab))));
|
||||
}
|
||||
@@ -1,8 +0,0 @@
|
||||
#ifndef BITVECTOR_H
|
||||
#define BITVECTOR_H
|
||||
|
||||
#include "utils/varbit.h"
|
||||
|
||||
VarBit *InitBitVector(int dim);
|
||||
|
||||
#endif
|
||||
@@ -8,7 +8,6 @@
|
||||
#include "commands/progress.h"
|
||||
#include "commands/vacuum.h"
|
||||
#include "hnsw.h"
|
||||
#include "miscadmin.h"
|
||||
#include "utils/guc.h"
|
||||
#include "utils/selfuncs.h"
|
||||
|
||||
@@ -29,7 +28,7 @@ static relopt_kind hnsw_relopt_kind;
|
||||
* this grows bigger, we should use a shmem_request_hook and
|
||||
* RequestAddinShmemSpace() to pre-reserve space for this.
|
||||
*/
|
||||
void
|
||||
static void
|
||||
HnswInitLockTranche(void)
|
||||
{
|
||||
int *tranche_ids;
|
||||
@@ -54,8 +53,7 @@ HnswInitLockTranche(void)
|
||||
void
|
||||
HnswInit(void)
|
||||
{
|
||||
if (!process_shared_preload_libraries_in_progress)
|
||||
HnswInitLockTranche();
|
||||
HnswInitLockTranche();
|
||||
|
||||
hnsw_relopt_kind = add_reloption_kind();
|
||||
add_int_reloption(hnsw_relopt_kind, "m", "Max number of connections",
|
||||
|
||||
38
src/hnsw.h
38
src/hnsw.h
@@ -80,7 +80,7 @@
|
||||
|
||||
#if PG_VERSION_NUM < 130000
|
||||
#define list_delete_last(list) list_truncate(list, list_length(list) - 1)
|
||||
#define list_sort(list, cmp) ((list) = list_qsort(list, cmp))
|
||||
#define list_sort(list, cmp) list_qsort(list, cmp)
|
||||
#endif
|
||||
|
||||
#define HnswIsElementTuple(tup) ((tup)->type == HNSW_ELEMENT_TUPLE_TYPE)
|
||||
@@ -129,7 +129,7 @@ HnswPtrDeclare(HnswNeighborArray, HnswNeighborArrayRelptr, HnswNeighborArrayPtr)
|
||||
HnswPtrDeclare(HnswNeighborArrayPtr, HnswNeighborsRelptr, HnswNeighborsPtr);
|
||||
HnswPtrDeclare(char, DatumRelptr, DatumPtr);
|
||||
|
||||
struct HnswElementData
|
||||
typedef struct HnswElementData
|
||||
{
|
||||
HnswElementPtr next;
|
||||
ItemPointerData heaptids[HNSW_HEAPTIDS];
|
||||
@@ -144,7 +144,7 @@ struct HnswElementData
|
||||
BlockNumber neighborPage;
|
||||
DatumPtr value;
|
||||
LWLock lock;
|
||||
};
|
||||
} HnswElementData;
|
||||
|
||||
typedef HnswElementData * HnswElement;
|
||||
|
||||
@@ -155,12 +155,12 @@ typedef struct HnswCandidate
|
||||
bool closer;
|
||||
} HnswCandidate;
|
||||
|
||||
struct HnswNeighborArray
|
||||
typedef struct HnswNeighborArray
|
||||
{
|
||||
int length;
|
||||
bool closerSet;
|
||||
HnswCandidate items[FLEXIBLE_ARRAY_MEMBER];
|
||||
};
|
||||
} HnswNeighborArray;
|
||||
|
||||
typedef struct HnswPairingHeapNode
|
||||
{
|
||||
@@ -185,7 +185,6 @@ typedef struct HnswGraph
|
||||
|
||||
/* Entry state */
|
||||
LWLock entryLock;
|
||||
LWLock entryWaitLock;
|
||||
HnswElementPtr entryPoint;
|
||||
|
||||
/* Allocations state */
|
||||
@@ -235,6 +234,12 @@ typedef struct HnswAllocator
|
||||
void *state;
|
||||
} HnswAllocator;
|
||||
|
||||
typedef struct HnswSupport
|
||||
{
|
||||
FmgrInfo *procinfo;
|
||||
Oid collation;
|
||||
} HnswSupport;
|
||||
|
||||
typedef struct HnswBuildState
|
||||
{
|
||||
/* Info */
|
||||
@@ -253,7 +258,7 @@ typedef struct HnswBuildState
|
||||
double reltuples;
|
||||
|
||||
/* Support functions */
|
||||
FmgrInfo *procinfo;
|
||||
HnswSupport support;
|
||||
FmgrInfo *normprocinfo;
|
||||
Oid collation;
|
||||
|
||||
@@ -330,7 +335,7 @@ typedef struct HnswScanOpaqueData
|
||||
MemoryContext tmpCtx;
|
||||
|
||||
/* Support functions */
|
||||
FmgrInfo *procinfo;
|
||||
HnswSupport support;
|
||||
FmgrInfo *normprocinfo;
|
||||
Oid collation;
|
||||
} HnswScanOpaqueData;
|
||||
@@ -350,8 +355,7 @@ typedef struct HnswVacuumState
|
||||
int efConstruction;
|
||||
|
||||
/* Support functions */
|
||||
FmgrInfo *procinfo;
|
||||
Oid collation;
|
||||
HnswSupport support;
|
||||
|
||||
/* Variables */
|
||||
struct tidhash_hash *deleted;
|
||||
@@ -367,30 +371,30 @@ typedef struct HnswVacuumState
|
||||
int HnswGetM(Relation index);
|
||||
int HnswGetEfConstruction(Relation index);
|
||||
FmgrInfo *HnswOptionalProcInfo(Relation index, uint16 procnum);
|
||||
void HnswInitSupport(HnswSupport * support, Relation index);
|
||||
bool HnswNormValue(FmgrInfo *procinfo, Oid collation, Datum *value, Vector * result);
|
||||
Buffer HnswNewBuffer(Relation index, ForkNumber forkNum);
|
||||
void HnswInitPage(Buffer buf, Page page);
|
||||
void HnswInit(void);
|
||||
List *HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement);
|
||||
List *HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, HnswSupport * support, 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, FmgrInfo *procinfo, Oid collation, int m, int efConstruction, bool existing);
|
||||
HnswCandidate *HnswEntryCandidate(char *base, HnswElement em, Datum q, Relation rel, FmgrInfo *procinfo, Oid collation, bool loadVec);
|
||||
void HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, HnswSupport * support, int m, int efConstruction, bool existing);
|
||||
HnswCandidate *HnswEntryCandidate(char *base, HnswElement em, Datum q, Relation rel, HnswSupport * support, 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);
|
||||
void HnswInitNeighbors(char *base, HnswElement element, int m, HnswAllocator * alloc);
|
||||
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 HnswUpdateNeighborsOnDisk(Relation index, HnswSupport * support, HnswElement e, int m, bool checkExisting, bool building);
|
||||
void HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHeaptids, bool loadVec);
|
||||
void HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec);
|
||||
void HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, HnswSupport * support, bool loadVec);
|
||||
void HnswSetElementTuple(char *base, HnswElementTuple etup, HnswElement element);
|
||||
void HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation);
|
||||
void HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, HnswSupport * support);
|
||||
void HnswLoadNeighbors(HnswElement element, Relation index, int m);
|
||||
void HnswInitLockTranche(void);
|
||||
PGDLLEXPORT void HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc);
|
||||
|
||||
/* Index access methods */
|
||||
|
||||
@@ -44,7 +44,6 @@
|
||||
#include "access/xact.h"
|
||||
#include "access/xloginsert.h"
|
||||
#include "catalog/index.h"
|
||||
#include "catalog/pg_type_d.h"
|
||||
#include "commands/progress.h"
|
||||
#include "hnsw.h"
|
||||
#include "miscadmin.h"
|
||||
@@ -177,7 +176,7 @@ CreateGraphPages(HnswBuildState * buildstate)
|
||||
Size etupSize;
|
||||
Size ntupSize;
|
||||
Size combinedSize;
|
||||
Pointer valuePtr = HnswPtrAccess(base, element->value);
|
||||
void *valuePtr = HnswPtrAccess(base, element->value);
|
||||
|
||||
/* Update iterator */
|
||||
iter = element->next;
|
||||
@@ -374,7 +373,7 @@ AddElementInMemory(char *base, HnswGraph * graph, HnswElement element)
|
||||
* Update neighbors
|
||||
*/
|
||||
static void
|
||||
UpdateNeighborsInMemory(char *base, FmgrInfo *procinfo, Oid collation, HnswElement e, int m)
|
||||
UpdateNeighborsInMemory(char *base, HnswSupport * support, HnswElement e, int m)
|
||||
{
|
||||
for (int lc = e->level; lc >= 0; lc--)
|
||||
{
|
||||
@@ -391,7 +390,7 @@ UpdateNeighborsInMemory(char *base, FmgrInfo *procinfo, Oid collation, HnswEleme
|
||||
|
||||
/* Use element for lock instead of hc since hc can be replaced */
|
||||
LWLockAcquire(&neighborElement->lock, LW_EXCLUSIVE);
|
||||
HnswUpdateConnection(base, e, hc, lm, lc, NULL, NULL, procinfo, collation);
|
||||
HnswUpdateConnection(base, e, hc, lm, lc, NULL, NULL, support);
|
||||
LWLockRelease(&neighborElement->lock);
|
||||
}
|
||||
}
|
||||
@@ -401,7 +400,7 @@ UpdateNeighborsInMemory(char *base, FmgrInfo *procinfo, Oid collation, HnswEleme
|
||||
* Update graph in memory
|
||||
*/
|
||||
static void
|
||||
UpdateGraphInMemory(FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement entryPoint, HnswBuildState * buildstate)
|
||||
UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, HnswBuildState * buildstate)
|
||||
{
|
||||
HnswGraph *graph = buildstate->graph;
|
||||
char *base = buildstate->hnswarea;
|
||||
@@ -414,7 +413,7 @@ UpdateGraphInMemory(FmgrInfo *procinfo, Oid collation, HnswElement element, int
|
||||
AddElementInMemory(base, graph, element);
|
||||
|
||||
/* Update neighbors */
|
||||
UpdateNeighborsInMemory(base, procinfo, collation, element, m);
|
||||
UpdateNeighborsInMemory(base, support, element, m);
|
||||
|
||||
/* Update entry point if needed (already have lock) */
|
||||
if (entryPoint == NULL || element->level > entryPoint->level)
|
||||
@@ -427,20 +426,13 @@ UpdateGraphInMemory(FmgrInfo *procinfo, Oid collation, HnswElement element, int
|
||||
static void
|
||||
InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
|
||||
{
|
||||
FmgrInfo *procinfo = buildstate->procinfo;
|
||||
Oid collation = buildstate->collation;
|
||||
HnswGraph *graph = buildstate->graph;
|
||||
HnswElement entryPoint;
|
||||
LWLock *entryLock = &graph->entryLock;
|
||||
LWLock *entryWaitLock = &graph->entryWaitLock;
|
||||
int efConstruction = buildstate->efConstruction;
|
||||
int m = buildstate->m;
|
||||
char *base = buildstate->hnswarea;
|
||||
|
||||
/* Wait if another process needs exclusive lock on entry lock */
|
||||
LWLockAcquire(entryWaitLock, LW_EXCLUSIVE);
|
||||
LWLockRelease(entryWaitLock);
|
||||
|
||||
/* Get entry point */
|
||||
LWLockAcquire(entryLock, LW_SHARED);
|
||||
entryPoint = HnswPtrAccess(base, graph->entryPoint);
|
||||
@@ -451,20 +443,18 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
|
||||
/* Release shared lock */
|
||||
LWLockRelease(entryLock);
|
||||
|
||||
/* Tell other processes to wait and get exclusive lock */
|
||||
LWLockAcquire(entryWaitLock, LW_EXCLUSIVE);
|
||||
/* Get exclusive lock */
|
||||
LWLockAcquire(entryLock, LW_EXCLUSIVE);
|
||||
LWLockRelease(entryWaitLock);
|
||||
|
||||
/* Get latest entry point after lock is acquired */
|
||||
entryPoint = HnswPtrAccess(base, graph->entryPoint);
|
||||
}
|
||||
|
||||
/* Find neighbors for element */
|
||||
HnswFindElementNeighbors(base, element, entryPoint, NULL, procinfo, collation, m, efConstruction, false);
|
||||
HnswFindElementNeighbors(base, element, entryPoint, NULL, &buildstate->support, m, efConstruction, false);
|
||||
|
||||
/* Update graph in memory */
|
||||
UpdateGraphInMemory(procinfo, collation, element, m, efConstruction, entryPoint, buildstate);
|
||||
UpdateGraphInMemory(&buildstate->support, element, m, efConstruction, entryPoint, buildstate);
|
||||
|
||||
/* Release entry lock */
|
||||
LWLockRelease(entryLock);
|
||||
@@ -609,9 +599,6 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
|
||||
static void
|
||||
InitGraph(HnswGraph * graph, char *base, long memoryTotal)
|
||||
{
|
||||
/* Initialize the lock tranche if needed */
|
||||
HnswInitLockTranche();
|
||||
|
||||
HnswPtrStore(base, graph->head, (HnswElement) NULL);
|
||||
HnswPtrStore(base, graph->entryPoint, (HnswElement) NULL);
|
||||
graph->memoryUsed = 0;
|
||||
@@ -620,7 +607,6 @@ InitGraph(HnswGraph * graph, char *base, long memoryTotal)
|
||||
graph->indtuples = 0;
|
||||
SpinLockInit(&graph->lock);
|
||||
LWLockInitialize(&graph->entryLock, hnsw_lock_tranche_id);
|
||||
LWLockInitialize(&graph->entryWaitLock, hnsw_lock_tranche_id);
|
||||
LWLockInitialize(&graph->allocatorLock, hnsw_lock_tranche_id);
|
||||
LWLockInitialize(&graph->flushLock, hnsw_lock_tranche_id);
|
||||
}
|
||||
@@ -672,12 +658,6 @@ HnswSharedMemoryAlloc(Size size, void *state)
|
||||
static void
|
||||
InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, IndexInfo *indexInfo, ForkNumber forkNum)
|
||||
{
|
||||
int maxDimensions = HNSW_MAX_DIM;
|
||||
Oid typid = TupleDescAttr(index->rd_att, 0)->atttypid;
|
||||
|
||||
if (typid == BITOID || typid == VARBITOID)
|
||||
maxDimensions *= 32;
|
||||
|
||||
buildstate->heap = heap;
|
||||
buildstate->index = index;
|
||||
buildstate->indexInfo = indexInfo;
|
||||
@@ -691,8 +671,8 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
|
||||
if (buildstate->dimensions < 0)
|
||||
elog(ERROR, "column does not have dimensions");
|
||||
|
||||
if (buildstate->dimensions > maxDimensions)
|
||||
elog(ERROR, "column cannot have more than %d dimensions for hnsw index", maxDimensions);
|
||||
if (buildstate->dimensions > HNSW_MAX_DIM)
|
||||
elog(ERROR, "column cannot have more than %d dimensions for hnsw index", HNSW_MAX_DIM);
|
||||
|
||||
if (buildstate->efConstruction < 2 * buildstate->m)
|
||||
elog(ERROR, "ef_construction must be greater than or equal to 2 * m");
|
||||
@@ -701,7 +681,7 @@ InitBuildState(HnswBuildState * buildstate, Relation heap, Relation index, Index
|
||||
buildstate->indtuples = 0;
|
||||
|
||||
/* Get support functions */
|
||||
buildstate->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
|
||||
HnswInitSupport(&buildstate->support, index);
|
||||
buildstate->normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
|
||||
buildstate->collation = index->rd_indcollation[0];
|
||||
|
||||
@@ -992,14 +972,6 @@ HnswBeginParallel(HnswBuildState * buildstate, bool isconcurrent, int request)
|
||||
/* Report less than allocated so never fails */
|
||||
InitGraph(&hnswshared->graphData, hnswarea, esthnswarea - 1024 * 1024);
|
||||
|
||||
/*
|
||||
* Avoid base address for relptr for Postgres < 14.5
|
||||
* https://github.com/postgres/postgres/commit/7201cd18627afc64850537806da7f22150d1a83b
|
||||
*/
|
||||
#if PG_VERSION_NUM < 140005
|
||||
hnswshared->graphData.memoryUsed += MAXALIGN(1);
|
||||
#endif
|
||||
|
||||
shm_toc_insert(pcxt->toc, PARALLEL_KEY_HNSW_SHARED, hnswshared);
|
||||
shm_toc_insert(pcxt->toc, PARALLEL_KEY_HNSW_AREA, hnswarea);
|
||||
|
||||
|
||||
@@ -340,7 +340,7 @@ ConnectionExists(HnswElement e, HnswNeighborTuple ntup, int startIdx, int lm)
|
||||
* Update neighbors
|
||||
*/
|
||||
void
|
||||
HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement e, int m, bool checkExisting, bool building)
|
||||
HnswUpdateNeighborsOnDisk(Relation index, HnswSupport * support, HnswElement e, int m, bool checkExisting, bool building)
|
||||
{
|
||||
char *base = NULL;
|
||||
|
||||
@@ -373,7 +373,7 @@ HnswUpdateNeighborsOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, Hns
|
||||
*/
|
||||
|
||||
/* Select neighbors */
|
||||
HnswUpdateConnection(NULL, e, hc, lm, lc, &idx, index, procinfo, collation);
|
||||
HnswUpdateConnection(NULL, e, hc, lm, lc, &idx, index, support);
|
||||
|
||||
/* New element was not selected as a neighbor */
|
||||
if (idx == -1)
|
||||
@@ -527,7 +527,7 @@ FindDuplicateOnDisk(Relation index, HnswElement element, bool building)
|
||||
* Update graph on disk
|
||||
*/
|
||||
static void
|
||||
UpdateGraphOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement element, int m, int efConstruction, HnswElement entryPoint, bool building)
|
||||
UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, bool building)
|
||||
{
|
||||
BlockNumber newInsertPage = InvalidBlockNumber;
|
||||
|
||||
@@ -543,7 +543,7 @@ UpdateGraphOnDisk(Relation index, FmgrInfo *procinfo, Oid collation, HnswElement
|
||||
HnswUpdateMetaPage(index, 0, NULL, newInsertPage, MAIN_FORKNUM, building);
|
||||
|
||||
/* Update neighbors */
|
||||
HnswUpdateNeighborsOnDisk(index, procinfo, collation, element, m, false, building);
|
||||
HnswUpdateNeighborsOnDisk(index, support, element, m, false, building);
|
||||
|
||||
/* Update entry point if needed */
|
||||
if (entryPoint == NULL || element->level > entryPoint->level)
|
||||
@@ -560,10 +560,11 @@ HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull,
|
||||
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;
|
||||
HnswSupport support;
|
||||
|
||||
HnswInitSupport(&support, index);
|
||||
|
||||
/*
|
||||
* Get a shared lock. This allows vacuum to ensure no in-flight inserts
|
||||
@@ -594,10 +595,10 @@ HnswInsertTupleOnDisk(Relation index, Datum value, Datum *values, bool *isnull,
|
||||
}
|
||||
|
||||
/* Find neighbors for element */
|
||||
HnswFindElementNeighbors(base, element, entryPoint, index, procinfo, collation, m, efConstruction, false);
|
||||
HnswFindElementNeighbors(base, element, entryPoint, index, &support, m, efConstruction, false);
|
||||
|
||||
/* Update graph on disk */
|
||||
UpdateGraphOnDisk(index, procinfo, collation, element, m, efConstruction, entryPoint, building);
|
||||
UpdateGraphOnDisk(index, &support, element, m, efConstruction, entryPoint, building);
|
||||
|
||||
/* Release lock */
|
||||
UnlockPage(index, HNSW_UPDATE_LOCK, lockmode);
|
||||
|
||||
@@ -1,8 +1,6 @@
|
||||
#include "postgres.h"
|
||||
|
||||
#include "access/relscan.h"
|
||||
#include "bitvector.h"
|
||||
#include "catalog/pg_type_d.h"
|
||||
#include "hnsw.h"
|
||||
#include "pgstat.h"
|
||||
#include "storage/bufmgr.h"
|
||||
@@ -17,8 +15,6 @@ GetScanItems(IndexScanDesc scan, Datum q)
|
||||
{
|
||||
HnswScanOpaque so = (HnswScanOpaque) scan->opaque;
|
||||
Relation index = scan->indexRelation;
|
||||
FmgrInfo *procinfo = so->procinfo;
|
||||
Oid collation = so->collation;
|
||||
List *ep;
|
||||
List *w;
|
||||
int m;
|
||||
@@ -31,15 +27,15 @@ GetScanItems(IndexScanDesc scan, Datum q)
|
||||
if (entryPoint == NULL)
|
||||
return NIL;
|
||||
|
||||
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, procinfo, collation, false));
|
||||
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, &so->support, false));
|
||||
|
||||
for (int lc = entryPoint->level; lc >= 1; lc--)
|
||||
{
|
||||
w = HnswSearchLayer(base, q, ep, 1, lc, index, procinfo, collation, m, false, NULL);
|
||||
w = HnswSearchLayer(base, q, ep, 1, lc, index, &so->support, m, false, NULL);
|
||||
ep = w;
|
||||
}
|
||||
|
||||
return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, procinfo, collation, m, false, NULL);
|
||||
return HnswSearchLayer(base, q, ep, hnsw_ef_search, 0, index, &so->support, m, false, NULL);
|
||||
}
|
||||
|
||||
/*
|
||||
@@ -75,15 +71,7 @@ GetScanValue(IndexScanDesc scan)
|
||||
Datum value;
|
||||
|
||||
if (scan->orderByData->sk_flags & SK_ISNULL)
|
||||
{
|
||||
Oid typid = TupleDescAttr(scan->indexRelation->rd_att, 0)->atttypid;
|
||||
int dimensions = GetDimensions(scan->indexRelation);
|
||||
|
||||
if (typid == BITOID || typid == VARBITOID)
|
||||
value = PointerGetDatum(InitBitVector(dimensions));
|
||||
else
|
||||
value = PointerGetDatum(InitVector(dimensions));
|
||||
}
|
||||
value = PointerGetDatum(InitVector(GetDimensions(scan->indexRelation)));
|
||||
else
|
||||
{
|
||||
value = scan->orderByData->sk_argument;
|
||||
@@ -118,7 +106,7 @@ hnswbeginscan(Relation index, int nkeys, int norderbys)
|
||||
ALLOCSET_DEFAULT_SIZES);
|
||||
|
||||
/* Set support functions */
|
||||
so->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
|
||||
HnswInitSupport(&so->support, index);
|
||||
so->normprocinfo = HnswOptionalProcInfo(index, HNSW_NORM_PROC);
|
||||
so->collation = index->rd_indcollation[0];
|
||||
|
||||
|
||||
100
src/hnswutils.c
100
src/hnswutils.c
@@ -149,6 +149,16 @@ HnswOptionalProcInfo(Relation index, uint16 procnum)
|
||||
return index_getprocinfo(index, 1, procnum);
|
||||
}
|
||||
|
||||
/*
|
||||
* Init support function
|
||||
*/
|
||||
void
|
||||
HnswInitSupport(HnswSupport * support, Relation index)
|
||||
{
|
||||
support->procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
|
||||
support->collation = index->rd_indcollation[0];
|
||||
}
|
||||
|
||||
/*
|
||||
* Divide by the norm
|
||||
*
|
||||
@@ -555,7 +565,7 @@ HnswLoadElementFromTuple(HnswElement element, HnswElementTuple etup, bool loadHe
|
||||
* Load an element and optionally get its distance from q
|
||||
*/
|
||||
void
|
||||
HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec)
|
||||
HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index, HnswSupport * support, bool loadVec)
|
||||
{
|
||||
Buffer buf;
|
||||
Page page;
|
||||
@@ -575,7 +585,7 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
|
||||
|
||||
/* Calculate distance */
|
||||
if (distance != NULL)
|
||||
*distance = (float) DatumGetFloat8(FunctionCall2Coll(procinfo, collation, *q, PointerGetDatum(&etup->data)));
|
||||
*distance = (float) DatumGetFloat8(FunctionCall2Coll(support->procinfo, support->collation, *q, PointerGetDatum(&etup->data)));
|
||||
|
||||
UnlockReleaseBuffer(buf);
|
||||
}
|
||||
@@ -584,27 +594,27 @@ HnswLoadElement(HnswElement element, float *distance, Datum *q, Relation index,
|
||||
* Get the distance for a candidate
|
||||
*/
|
||||
static float
|
||||
GetCandidateDistance(char *base, HnswCandidate * hc, Datum q, FmgrInfo *procinfo, Oid collation)
|
||||
GetCandidateDistance(char *base, HnswCandidate * hc, Datum q, HnswSupport * support)
|
||||
{
|
||||
HnswElement hce = HnswPtrAccess(base, hc->element);
|
||||
Datum value = HnswGetValue(base, hce);
|
||||
|
||||
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, q, value));
|
||||
return DatumGetFloat8(FunctionCall2Coll(support->procinfo, support->collation, q, value));
|
||||
}
|
||||
|
||||
/*
|
||||
* Create a candidate for the entry point
|
||||
*/
|
||||
HnswCandidate *
|
||||
HnswEntryCandidate(char *base, HnswElement entryPoint, Datum q, Relation index, FmgrInfo *procinfo, Oid collation, bool loadVec)
|
||||
HnswEntryCandidate(char *base, HnswElement entryPoint, Datum q, Relation index, HnswSupport * support, bool loadVec)
|
||||
{
|
||||
HnswCandidate *hc = palloc(sizeof(HnswCandidate));
|
||||
|
||||
HnswPtrStore(base, hc->element, entryPoint);
|
||||
if (index == NULL)
|
||||
hc->distance = GetCandidateDistance(base, hc, q, procinfo, collation);
|
||||
hc->distance = GetCandidateDistance(base, hc, q, support);
|
||||
else
|
||||
HnswLoadElement(entryPoint, &hc->distance, &q, index, procinfo, collation, loadVec);
|
||||
HnswLoadElement(entryPoint, &hc->distance, &q, index, support, loadVec);
|
||||
return hc;
|
||||
}
|
||||
|
||||
@@ -722,7 +732,7 @@ CountElement(char *base, HnswElement skipElement, HnswCandidate * hc)
|
||||
* Algorithm 2 from paper
|
||||
*/
|
||||
List *
|
||||
HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, FmgrInfo *procinfo, Oid collation, int m, bool inserting, HnswElement skipElement)
|
||||
HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, HnswSupport * support, int m, bool inserting, HnswElement skipElement)
|
||||
{
|
||||
List *w = NIL;
|
||||
pairingheap *C = pairingheap_allocate(CompareNearestCandidates, NULL);
|
||||
@@ -804,9 +814,9 @@ HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, F
|
||||
f = ((HnswPairingHeapNode *) pairingheap_first(W))->inner;
|
||||
|
||||
if (index == NULL)
|
||||
eDistance = GetCandidateDistance(base, e, q, procinfo, collation);
|
||||
eDistance = GetCandidateDistance(base, e, q, support);
|
||||
else
|
||||
HnswLoadElement(eElement, &eDistance, &q, index, procinfo, collation, inserting);
|
||||
HnswLoadElement(eElement, &eDistance, &q, index, support, inserting);
|
||||
|
||||
Assert(!eElement->deleted);
|
||||
|
||||
@@ -860,15 +870,12 @@ HnswSearchLayer(char *base, Datum q, List *ep, int ef, int lc, Relation index, F
|
||||
static int
|
||||
#if PG_VERSION_NUM >= 130000
|
||||
CompareCandidateDistances(const ListCell *a, const ListCell *b)
|
||||
{
|
||||
HnswCandidate *hca = lfirst(a);
|
||||
HnswCandidate *hcb = lfirst(b);
|
||||
#else
|
||||
CompareCandidateDistances(const void *a, const void *b)
|
||||
{
|
||||
HnswCandidate *hca = lfirst(*(ListCell **) a);
|
||||
HnswCandidate *hcb = lfirst(*(ListCell **) b);
|
||||
#endif
|
||||
{
|
||||
HnswCandidate *hca = lfirst((ListCell *) a);
|
||||
HnswCandidate *hcb = lfirst((ListCell *) b);
|
||||
|
||||
if (hca->distance < hcb->distance)
|
||||
return 1;
|
||||
@@ -891,15 +898,12 @@ CompareCandidateDistances(const void *a, const void *b)
|
||||
static int
|
||||
#if PG_VERSION_NUM >= 130000
|
||||
CompareCandidateDistancesOffset(const ListCell *a, const ListCell *b)
|
||||
{
|
||||
HnswCandidate *hca = lfirst(a);
|
||||
HnswCandidate *hcb = lfirst(b);
|
||||
#else
|
||||
CompareCandidateDistancesOffset(const void *a, const void *b)
|
||||
{
|
||||
HnswCandidate *hca = lfirst(*(ListCell **) a);
|
||||
HnswCandidate *hcb = lfirst(*(ListCell **) b);
|
||||
#endif
|
||||
{
|
||||
HnswCandidate *hca = lfirst((ListCell *) a);
|
||||
HnswCandidate *hcb = lfirst((ListCell *) b);
|
||||
|
||||
if (hca->distance < hcb->distance)
|
||||
return 1;
|
||||
@@ -920,19 +924,19 @@ CompareCandidateDistancesOffset(const void *a, const void *b)
|
||||
* Calculate the distance between elements
|
||||
*/
|
||||
static float
|
||||
HnswGetDistance(char *base, HnswElement a, HnswElement b, FmgrInfo *procinfo, Oid collation)
|
||||
HnswGetDistance(char *base, HnswElement a, HnswElement b, HnswSupport * support)
|
||||
{
|
||||
Datum aValue = HnswGetValue(base, a);
|
||||
Datum bValue = HnswGetValue(base, b);
|
||||
|
||||
return DatumGetFloat8(FunctionCall2Coll(procinfo, collation, aValue, bValue));
|
||||
return DatumGetFloat8(FunctionCall2Coll(support->procinfo, support->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, FmgrInfo *procinfo, Oid collation)
|
||||
CheckElementCloser(char *base, HnswCandidate * e, List *r, HnswSupport * support)
|
||||
{
|
||||
HnswElement eElement = HnswPtrAccess(base, e->element);
|
||||
ListCell *lc2;
|
||||
@@ -941,7 +945,7 @@ CheckElementCloser(char *base, HnswCandidate * e, List *r, FmgrInfo *procinfo, O
|
||||
{
|
||||
HnswCandidate *ri = lfirst(lc2);
|
||||
HnswElement riElement = HnswPtrAccess(base, ri->element);
|
||||
float distance = HnswGetDistance(base, eElement, riElement, procinfo, collation);
|
||||
float distance = HnswGetDistance(base, eElement, riElement, support);
|
||||
|
||||
if (distance <= e->distance)
|
||||
return false;
|
||||
@@ -954,13 +958,11 @@ CheckElementCloser(char *base, HnswCandidate * e, List *r, FmgrInfo *procinfo, O
|
||||
* Algorithm 4 from paper
|
||||
*/
|
||||
static List *
|
||||
SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid collation, HnswElement e2, HnswCandidate * newCandidate, HnswCandidate * *pruned, bool sortCandidates)
|
||||
SelectNeighbors(char *base, List *c, int lm, int lc, HnswSupport * support, HnswElement e2, HnswCandidate * newCandidate, HnswCandidate * *pruned, bool sortCandidates)
|
||||
{
|
||||
List *r = NIL;
|
||||
List *w = list_copy(c);
|
||||
HnswCandidate **wd;
|
||||
int wdlen = 0;
|
||||
int wdoff = 0;
|
||||
pairingheap *wd;
|
||||
HnswNeighborArray *neighbors = HnswGetNeighbors(base, e2, lc);
|
||||
bool mustCalculate = !neighbors->closerSet;
|
||||
List *added = NIL;
|
||||
@@ -969,7 +971,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
if (list_length(w) <= lm)
|
||||
return w;
|
||||
|
||||
wd = palloc(sizeof(HnswCandidate *) * list_length(w));
|
||||
wd = pairingheap_allocate(CompareNearestCandidates, NULL);
|
||||
|
||||
/* Ensure order of candidates is deterministic for closer caching */
|
||||
if (sortCandidates)
|
||||
@@ -989,7 +991,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
|
||||
/* Use previous state of r and wd to skip work when possible */
|
||||
if (mustCalculate)
|
||||
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
|
||||
e->closer = CheckElementCloser(base, e, r, support);
|
||||
else if (list_length(added) > 0)
|
||||
{
|
||||
/* Keep Valgrind happy for in-memory, parallel builds */
|
||||
@@ -1002,7 +1004,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
*/
|
||||
if (e->closer)
|
||||
{
|
||||
e->closer = CheckElementCloser(base, e, added, procinfo, collation);
|
||||
e->closer = CheckElementCloser(base, e, added, support);
|
||||
|
||||
if (!e->closer)
|
||||
removedAny = true;
|
||||
@@ -1015,7 +1017,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
*/
|
||||
if (removedAny)
|
||||
{
|
||||
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
|
||||
e->closer = CheckElementCloser(base, e, r, support);
|
||||
if (e->closer)
|
||||
added = lappend(added, e);
|
||||
}
|
||||
@@ -1023,7 +1025,7 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
}
|
||||
else if (e == newCandidate)
|
||||
{
|
||||
e->closer = CheckElementCloser(base, e, r, procinfo, collation);
|
||||
e->closer = CheckElementCloser(base, e, r, support);
|
||||
if (e->closer)
|
||||
added = lappend(added, e);
|
||||
}
|
||||
@@ -1035,21 +1037,21 @@ SelectNeighbors(char *base, List *c, int lm, int lc, FmgrInfo *procinfo, Oid col
|
||||
if (e->closer)
|
||||
r = lappend(r, e);
|
||||
else
|
||||
wd[wdlen++] = e;
|
||||
pairingheap_add(wd, &(CreatePairingHeapNode(e)->ph_node));
|
||||
}
|
||||
|
||||
/* Cached value can only be used in future if sorted deterministically */
|
||||
neighbors->closerSet = sortCandidates;
|
||||
|
||||
/* Keep pruned connections */
|
||||
while (wdoff < wdlen && list_length(r) < lm)
|
||||
r = lappend(r, wd[wdoff++]);
|
||||
while (!pairingheap_is_empty(wd) && list_length(r) < lm)
|
||||
r = lappend(r, ((HnswPairingHeapNode *) pairingheap_remove_first(wd))->inner);
|
||||
|
||||
/* Return pruned for update connections */
|
||||
if (pruned != NULL)
|
||||
{
|
||||
if (wdoff < wdlen)
|
||||
*pruned = wd[wdoff];
|
||||
if (!pairingheap_is_empty(wd))
|
||||
*pruned = ((HnswPairingHeapNode *) pairingheap_first(wd))->inner;
|
||||
else
|
||||
*pruned = linitial(w);
|
||||
}
|
||||
@@ -1074,7 +1076,7 @@ AddConnections(char *base, HnswElement element, List *neighbors, int lc)
|
||||
* Update connections
|
||||
*/
|
||||
void
|
||||
HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, FmgrInfo *procinfo, Oid collation)
|
||||
HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm, int lc, int *updateIdx, Relation index, HnswSupport * support)
|
||||
{
|
||||
HnswElement hce = HnswPtrAccess(base, hc->element);
|
||||
HnswNeighborArray *currentNeighbors = HnswGetNeighbors(base, hce, lc);
|
||||
@@ -1107,9 +1109,9 @@ HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm
|
||||
HnswElement hc3Element = HnswPtrAccess(base, hc3->element);
|
||||
|
||||
if (HnswPtrIsNull(base, hc3Element->value))
|
||||
HnswLoadElement(hc3Element, &hc3->distance, &q, index, procinfo, collation, true);
|
||||
HnswLoadElement(hc3Element, &hc3->distance, &q, index, support, true);
|
||||
else
|
||||
hc3->distance = GetCandidateDistance(base, hc3, q, procinfo, collation);
|
||||
hc3->distance = GetCandidateDistance(base, hc3, q, support);
|
||||
|
||||
/* Prune element if being deleted */
|
||||
if (hc3Element->heaptidsLength == 0)
|
||||
@@ -1129,7 +1131,7 @@ HnswUpdateConnection(char *base, HnswElement element, HnswCandidate * hc, int lm
|
||||
c = lappend(c, ¤tNeighbors->items[i]);
|
||||
c = lappend(c, &hc2);
|
||||
|
||||
SelectNeighbors(base, c, lm, lc, procinfo, collation, hce, &hc2, &pruned, true);
|
||||
SelectNeighbors(base, c, lm, lc, support, hce, &hc2, &pruned, true);
|
||||
|
||||
/* Should not happen */
|
||||
if (pruned == NULL)
|
||||
@@ -1203,7 +1205,7 @@ PrecomputeHash(char *base, HnswElement element)
|
||||
* Algorithm 1 from paper
|
||||
*/
|
||||
void
|
||||
HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, FmgrInfo *procinfo, Oid collation, int m, int efConstruction, bool existing)
|
||||
HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint, Relation index, HnswSupport * support, int m, int efConstruction, bool existing)
|
||||
{
|
||||
List *ep;
|
||||
List *w;
|
||||
@@ -1223,13 +1225,13 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
|
||||
return;
|
||||
|
||||
/* Get entry point and level */
|
||||
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, procinfo, collation, true));
|
||||
ep = list_make1(HnswEntryCandidate(base, entryPoint, q, index, support, 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, procinfo, collation, m, true, skipElement);
|
||||
w = HnswSearchLayer(base, q, ep, 1, lc, index, support, m, true, skipElement);
|
||||
ep = w;
|
||||
}
|
||||
|
||||
@@ -1247,7 +1249,7 @@ HnswFindElementNeighbors(char *base, HnswElement element, HnswElement entryPoint
|
||||
List *neighbors;
|
||||
List *lw;
|
||||
|
||||
w = HnswSearchLayer(base, q, ep, efConstruction, lc, index, procinfo, collation, m, true, skipElement);
|
||||
w = HnswSearchLayer(base, q, ep, efConstruction, lc, index, support, m, true, skipElement);
|
||||
|
||||
/* Elements being deleted or skipped can help with search */
|
||||
/* but should be removed before selecting neighbors */
|
||||
@@ -1261,7 +1263,7 @@ 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, lc, procinfo, collation, element, NULL, NULL, false);
|
||||
neighbors = SelectNeighbors(base, lw, lm, lc, support, element, NULL, NULL, false);
|
||||
|
||||
AddConnections(base, element, neighbors, lc);
|
||||
|
||||
|
||||
@@ -189,8 +189,6 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
|
||||
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);
|
||||
@@ -205,7 +203,7 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
|
||||
element->heaptidsLength = 0;
|
||||
|
||||
/* Find neighbors for element, skipping itself */
|
||||
HnswFindElementNeighbors(base, element, entryPoint, index, procinfo, collation, m, efConstruction, true);
|
||||
HnswFindElementNeighbors(base, element, entryPoint, index, &vacuumstate->support, m, efConstruction, true);
|
||||
|
||||
/* Zero memory for each element */
|
||||
MemSet(ntup, 0, HNSW_TUPLE_ALLOC_SIZE);
|
||||
@@ -229,7 +227,7 @@ RepairGraphElement(HnswVacuumState * vacuumstate, HnswElement element, HnswEleme
|
||||
UnlockReleaseBuffer(buf);
|
||||
|
||||
/* Update neighbors */
|
||||
HnswUpdateNeighborsOnDisk(index, procinfo, collation, element, m, true, false);
|
||||
HnswUpdateNeighborsOnDisk(index, &vacuumstate->support, element, m, true, false);
|
||||
}
|
||||
|
||||
/*
|
||||
@@ -256,7 +254,7 @@ RepairGraphEntryPoint(HnswVacuumState * vacuumstate)
|
||||
LockPage(index, HNSW_UPDATE_LOCK, ShareLock);
|
||||
|
||||
/* Load element */
|
||||
HnswLoadElement(highestPoint, NULL, NULL, index, vacuumstate->procinfo, vacuumstate->collation, true);
|
||||
HnswLoadElement(highestPoint, NULL, NULL, index, &vacuumstate->support, true);
|
||||
|
||||
/* Repair if needed */
|
||||
if (NeedsUpdated(vacuumstate, highestPoint))
|
||||
@@ -294,7 +292,7 @@ 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, vacuumstate->procinfo, vacuumstate->collation, true);
|
||||
HnswLoadElement(entryPoint, NULL, NULL, index, &vacuumstate->support, true);
|
||||
|
||||
if (NeedsUpdated(vacuumstate, entryPoint))
|
||||
{
|
||||
@@ -573,8 +571,8 @@ 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->support.procinfo = index_getprocinfo(index, 1, HNSW_DISTANCE_PROC);
|
||||
vacuumstate->support.collation = index->rd_indcollation[0];
|
||||
vacuumstate->ntup = palloc0(HNSW_TUPLE_ALLOC_SIZE);
|
||||
vacuumstate->tmpCtx = AllocSetContextCreate(CurrentMemoryContext,
|
||||
"Hnsw vacuum temporary context",
|
||||
|
||||
21
src/vector.c
21
src/vector.c
@@ -2,7 +2,6 @@
|
||||
|
||||
#include <math.h>
|
||||
|
||||
#include "bitvector.h"
|
||||
#include "catalog/pg_type.h"
|
||||
#include "common/shortest_dec.h"
|
||||
#include "fmgr.h"
|
||||
@@ -861,26 +860,6 @@ vector_mul(PG_FUNCTION_ARGS)
|
||||
PG_RETURN_POINTER(result);
|
||||
}
|
||||
|
||||
/*
|
||||
* Quantize a vector
|
||||
*/
|
||||
PGDLLEXPORT PG_FUNCTION_INFO_V1(quantize_binary);
|
||||
Datum
|
||||
quantize_binary(PG_FUNCTION_ARGS)
|
||||
{
|
||||
Vector *a = PG_GETARG_VECTOR_P(0);
|
||||
float *ax = a->x;
|
||||
VarBit *result = InitBitVector(a->dim);
|
||||
unsigned char *rx = VARBITS(result);
|
||||
|
||||
/* TODO Improve */
|
||||
for (int i = 0; i < a->dim; i++)
|
||||
rx[i / 8] |= (ax[i] > 0) << (7 - (i % 8));
|
||||
|
||||
PG_RETURN_VARBIT_P(result);
|
||||
}
|
||||
|
||||
|
||||
/*
|
||||
* Internal helper to compare vectors
|
||||
*/
|
||||
|
||||
@@ -1,64 +0,0 @@
|
||||
SELECT hamming_distance(B'111', B'111');
|
||||
hamming_distance
|
||||
------------------
|
||||
0
|
||||
(1 row)
|
||||
|
||||
SELECT hamming_distance(B'111', B'110');
|
||||
hamming_distance
|
||||
------------------
|
||||
1
|
||||
(1 row)
|
||||
|
||||
SELECT hamming_distance(B'111', B'100');
|
||||
hamming_distance
|
||||
------------------
|
||||
2
|
||||
(1 row)
|
||||
|
||||
SELECT hamming_distance(B'111', B'000');
|
||||
hamming_distance
|
||||
------------------
|
||||
3
|
||||
(1 row)
|
||||
|
||||
SELECT hamming_distance(B'111', B'00');
|
||||
ERROR: different bit lengths 3 and 2
|
||||
SELECT jaccard_distance(B'1111', B'1111');
|
||||
jaccard_distance
|
||||
------------------
|
||||
0
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'1110');
|
||||
jaccard_distance
|
||||
------------------
|
||||
0.25
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'1100');
|
||||
jaccard_distance
|
||||
------------------
|
||||
0.5
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'1000');
|
||||
jaccard_distance
|
||||
------------------
|
||||
0.75
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'0000');
|
||||
jaccard_distance
|
||||
------------------
|
||||
1
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1100', B'1000');
|
||||
jaccard_distance
|
||||
------------------
|
||||
0.5
|
||||
(1 row)
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'000');
|
||||
ERROR: different bit lengths 4 and 3
|
||||
@@ -208,18 +208,6 @@ SELECT l1_distance('[3e38]', '[-3e38]');
|
||||
Infinity
|
||||
(1 row)
|
||||
|
||||
SELECT quantize_binary('[1,0,-1]');
|
||||
quantize_binary
|
||||
-----------------
|
||||
100
|
||||
(1 row)
|
||||
|
||||
SELECT quantize_binary('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]');
|
||||
quantize_binary
|
||||
-----------------
|
||||
01001110101
|
||||
(1 row)
|
||||
|
||||
SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]']) v;
|
||||
avg
|
||||
-----------
|
||||
|
||||
@@ -1,21 +0,0 @@
|
||||
SET enable_seqscan = off;
|
||||
CREATE TABLE t (val bit(3));
|
||||
INSERT INTO t (val) VALUES (B'000'), (B'100'), (B'111'), (NULL);
|
||||
CREATE INDEX ON t USING hnsw (val bit_hamming_ops);
|
||||
INSERT INTO t (val) VALUES (B'110');
|
||||
SELECT * FROM t ORDER BY val <~> B'111';
|
||||
val
|
||||
-----
|
||||
111
|
||||
110
|
||||
100
|
||||
000
|
||||
(4 rows)
|
||||
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <~> (SELECT NULL::bit)) t2;
|
||||
count
|
||||
-------
|
||||
4
|
||||
(1 row)
|
||||
|
||||
DROP TABLE t;
|
||||
@@ -1,21 +0,0 @@
|
||||
SET enable_seqscan = off;
|
||||
CREATE TABLE t (val bit(4));
|
||||
INSERT INTO t (val) VALUES (B'0000'), (B'1100'), (B'1111'), (NULL);
|
||||
CREATE INDEX ON t USING hnsw (val bit_jaccard_ops);
|
||||
INSERT INTO t (val) VALUES (B'1110');
|
||||
SELECT * FROM t ORDER BY val <%> B'1111';
|
||||
val
|
||||
------
|
||||
1111
|
||||
1110
|
||||
1100
|
||||
0000
|
||||
(4 rows)
|
||||
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <%> (SELECT NULL::bit)) t2;
|
||||
count
|
||||
-------
|
||||
4
|
||||
(1 row)
|
||||
|
||||
DROP TABLE t;
|
||||
@@ -116,30 +116,8 @@ SELECT '[1, ,3]'::vector;
|
||||
ERROR: invalid input syntax for type vector: "[1, ,3]"
|
||||
LINE 1: SELECT '[1, ,3]'::vector;
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(3);
|
||||
vector
|
||||
---------
|
||||
[1,2,3]
|
||||
(1 row)
|
||||
|
||||
SELECT '[1,2,3]'::vector(2);
|
||||
ERROR: expected 2 dimensions, not 3
|
||||
SELECT '[1,2,3]'::vector(3, 2);
|
||||
ERROR: invalid type modifier
|
||||
LINE 1: SELECT '[1,2,3]'::vector(3, 2);
|
||||
^
|
||||
SELECT '[1,2,3]'::vector('a');
|
||||
ERROR: invalid input syntax for type integer: "a"
|
||||
LINE 1: SELECT '[1,2,3]'::vector('a');
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(0);
|
||||
ERROR: dimensions for type vector must be at least 1
|
||||
LINE 1: SELECT '[1,2,3]'::vector(0);
|
||||
^
|
||||
SELECT '[1,2,3]'::vector(16001);
|
||||
ERROR: dimensions for type vector cannot exceed 16000
|
||||
LINE 1: SELECT '[1,2,3]'::vector(16001);
|
||||
^
|
||||
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
|
||||
unnest
|
||||
---------
|
||||
|
||||
@@ -1,13 +0,0 @@
|
||||
SELECT hamming_distance(B'111', B'111');
|
||||
SELECT hamming_distance(B'111', B'110');
|
||||
SELECT hamming_distance(B'111', B'100');
|
||||
SELECT hamming_distance(B'111', B'000');
|
||||
SELECT hamming_distance(B'111', B'00');
|
||||
|
||||
SELECT jaccard_distance(B'1111', B'1111');
|
||||
SELECT jaccard_distance(B'1111', B'1110');
|
||||
SELECT jaccard_distance(B'1111', B'1100');
|
||||
SELECT jaccard_distance(B'1111', B'1000');
|
||||
SELECT jaccard_distance(B'1111', B'0000');
|
||||
SELECT jaccard_distance(B'1100', B'1000');
|
||||
SELECT jaccard_distance(B'1111', B'000');
|
||||
@@ -48,9 +48,6 @@ SELECT l1_distance('[0,0]', '[0,1]');
|
||||
SELECT l1_distance('[1,2]', '[3]');
|
||||
SELECT l1_distance('[3e38]', '[-3e38]');
|
||||
|
||||
SELECT quantize_binary('[1,0,-1]');
|
||||
SELECT quantize_binary('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]');
|
||||
|
||||
SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]']) v;
|
||||
SELECT avg(v) FROM unnest(ARRAY['[1,2,3]'::vector, '[3,5,7]', NULL]) v;
|
||||
SELECT avg(v) FROM unnest(ARRAY[]::vector[]) v;
|
||||
|
||||
@@ -1,12 +0,0 @@
|
||||
SET enable_seqscan = off;
|
||||
|
||||
CREATE TABLE t (val bit(3));
|
||||
INSERT INTO t (val) VALUES (B'000'), (B'100'), (B'111'), (NULL);
|
||||
CREATE INDEX ON t USING hnsw (val bit_hamming_ops);
|
||||
|
||||
INSERT INTO t (val) VALUES (B'110');
|
||||
|
||||
SELECT * FROM t ORDER BY val <~> B'111';
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <~> (SELECT NULL::bit)) t2;
|
||||
|
||||
DROP TABLE t;
|
||||
@@ -1,12 +0,0 @@
|
||||
SET enable_seqscan = off;
|
||||
|
||||
CREATE TABLE t (val bit(4));
|
||||
INSERT INTO t (val) VALUES (B'0000'), (B'1100'), (B'1111'), (NULL);
|
||||
CREATE INDEX ON t USING hnsw (val bit_jaccard_ops);
|
||||
|
||||
INSERT INTO t (val) VALUES (B'1110');
|
||||
|
||||
SELECT * FROM t ORDER BY val <%> B'1111';
|
||||
SELECT COUNT(*) FROM (SELECT * FROM t ORDER BY val <%> (SELECT NULL::bit)) t2;
|
||||
|
||||
DROP TABLE t;
|
||||
@@ -22,13 +22,7 @@ SELECT '[1,]'::vector;
|
||||
SELECT '[1a]'::vector;
|
||||
SELECT '[1,,3]'::vector;
|
||||
SELECT '[1, ,3]'::vector;
|
||||
|
||||
SELECT '[1,2,3]'::vector(3);
|
||||
SELECT '[1,2,3]'::vector(2);
|
||||
SELECT '[1,2,3]'::vector(3, 2);
|
||||
SELECT '[1,2,3]'::vector('a');
|
||||
SELECT '[1,2,3]'::vector(0);
|
||||
SELECT '[1,2,3]'::vector(16001);
|
||||
|
||||
SELECT unnest('{"[1,2,3]", "[4,5,6]"}'::vector[]);
|
||||
SELECT '{"[1,2,3]"}'::vector(2)[];
|
||||
|
||||
@@ -1,137 +0,0 @@
|
||||
use strict;
|
||||
use warnings;
|
||||
use PostgresNode;
|
||||
use TestLib;
|
||||
use Test::More;
|
||||
|
||||
my $node;
|
||||
my @queries = ();
|
||||
my @expected;
|
||||
my $limit = 20;
|
||||
my $dim = 52;
|
||||
my $max = 2**$dim;
|
||||
|
||||
sub test_recall
|
||||
{
|
||||
my ($min, $operator) = @_;
|
||||
my $correct = 0;
|
||||
my $total = 0;
|
||||
|
||||
my $explain = $node->safe_psql("postgres", qq(
|
||||
SET enable_seqscan = off;
|
||||
SET hnsw.ef_search = 100;
|
||||
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;
|
||||
SET hnsw.ef_search = 100;
|
||||
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 = get_new_node('node');
|
||||
$node->init;
|
||||
$node->start;
|
||||
|
||||
# Create table
|
||||
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
|
||||
$node->safe_psql("postgres", "CREATE TABLE tst (i int4, v bit($dim));");
|
||||
$node->safe_psql("postgres",
|
||||
"INSERT INTO tst SELECT i, (random() * $max)::bigint::bit($dim) FROM generate_series(1, 10000) i;"
|
||||
);
|
||||
|
||||
# Generate queries
|
||||
for (1 .. 20)
|
||||
{
|
||||
my $r = int(rand() * $max);
|
||||
push(@queries, "${r}::bigint::bit($dim)");
|
||||
}
|
||||
|
||||
# Check each index type
|
||||
my @operators = ("<~>", "<\%>");
|
||||
my @opclasses = ("bit_hamming_ops", "bit_jaccard_ops");
|
||||
|
||||
for my $i (0 .. $#operators)
|
||||
{
|
||||
my $operator = $operators[$i];
|
||||
my $opclass = $opclasses[$i];
|
||||
|
||||
# Get exact results
|
||||
@expected = ();
|
||||
foreach (@queries)
|
||||
{
|
||||
# Handle ties
|
||||
my $res = $node->safe_psql("postgres", qq(
|
||||
WITH top AS (
|
||||
SELECT v $operator $_ AS distance FROM tst ORDER BY v $operator $_ 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 hnsw (v $opclass);
|
||||
));
|
||||
|
||||
# Test approximate results
|
||||
my $min = $operator eq "<\%>" ? 0.96 : 0.99;
|
||||
test_recall($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/);
|
||||
|
||||
$node->safe_psql("postgres", "DROP INDEX idx;");
|
||||
}
|
||||
|
||||
done_testing();
|
||||
@@ -1,4 +1,4 @@
|
||||
comment = 'vector data type and ivfflat and hnsw access methods'
|
||||
default_version = '0.6.2'
|
||||
default_version = '0.6.0'
|
||||
module_pathname = '$libdir/vector'
|
||||
relocatable = true
|
||||
|
||||
Reference in New Issue
Block a user