Compare commits

...

48 Commits

Author SHA1 Message Date
Andrew Kane
5af768752f Version bump to 0.3.0 [skip ci] 2022-10-15 14:38:43 -07:00
Andrew Kane
f03bcc4fd8 Updated Dockerfile to Postgres 15 [skip ci] 2022-10-14 16:01:25 -07:00
Andrew Kane
74fc4b9aaa Added .pm to .editorconfig [skip ci] 2022-10-14 14:05:20 -07:00
Andrew Kane
6870ba96ba Removed Postgres 9.6 from CI 2022-10-14 13:47:07 -07:00
Andrew Kane
2575cfdaa8 Fixed CI 2022-10-14 13:45:16 -07:00
Andrew Kane
9ff5ceacca Dropped support for Postgres 9.6 [skip ci] 2022-10-14 13:44:13 -07:00
Andrew Kane
9a58ee609a Added Postgres 15 to CI 2022-10-14 13:42:04 -07:00
Andrew Kane
66a31bef36 Simplified define [skip ci] 2022-10-10 15:01:16 -07:00
Andrew Kane
163ecdf063 Use define for RandomDouble 2022-10-10 14:59:10 -07:00
Andrew Kane
c9c6b96eed Added support for Postgres 15 [skip ci] 2022-10-05 16:11:37 -07:00
Andrew Kane
48e9a6830b Added link to pgvector-elixir [skip ci] 2022-08-03 23:12:25 -07:00
Andrew Kane
f22b47ca46 Added link to pgvector-php - #30 [skip ci] 2022-08-03 13:50:34 -07:00
Andrew Kane
2a057c055c Updated example [skip ci] 2022-08-01 00:06:45 -07:00
Andrew Kane
b89b58b893 Updated examples to be runnable - fixes #24 [skip ci] 2022-07-31 22:36:46 -07:00
Andrew Kane
2e77650e1d Updated example [skip ci] 2022-07-31 22:34:03 -07:00
Andrew Kane
0a68b040fb DRY code 2022-07-31 21:56:44 -07:00
Andrew Kane
434dc904c1 Version bump to 0.2.7 [skip ci] 2022-07-31 21:42:15 -07:00
Andrew Kane
92a671eeb1 Revert "Commit pages with tuple"
This reverts commit 94eeb83abe.
2022-07-31 21:36:46 -07:00
Andrew Kane
94eeb83abe Commit pages with tuple 2022-07-31 21:29:29 -07:00
Andrew Kane
6866939090 Fixed indentation [skip ci] 2022-07-31 21:21:06 -07:00
Andrew Kane
ba2fb6ad8c Fixed order in IvfflatAppendPage 2022-07-31 21:03:57 -07:00
Andrew Kane
51fe8aff02 Fixed unexpected data beyond EOF error - fixes #29 2022-07-31 20:48:15 -07:00
Andrew Kane
df3dc151d5 Updated readme [skip ci] 2022-07-27 17:38:13 -07:00
Andrew Kane
6687430589 Fixed CI 2022-07-10 16:30:45 -04:00
Andrew Kane
f128a455d1 Made dimensions configurable for WAL test - #29 2022-07-10 16:01:05 -04:00
Andrew Kane
56cb5f3503 Version bump to 0.2.6 [skip ci] 2022-05-22 18:21:15 -07:00
Andrew Kane
ecbf46938f Moved metrics to separate function [skip ci] 2022-02-15 19:13:23 -08:00
Andrew Kane
4ca264ba02 Added Davies-Bouldin index [skip ci] 2022-02-15 19:08:47 -08:00
Andrew Kane
f5458414b8 Made recall test less flaky 2022-02-15 18:17:05 -08:00
Andrew Kane
e64ed39acb Disabled scan progress for sampling 2022-02-15 18:14:59 -08:00
Andrew Kane
0d025be9d3 Improved performance of index creation for Postgres < 12 2022-02-15 18:05:33 -08:00
Andrew Kane
fed60dce78 Added tests for lists 2022-02-15 17:04:20 -08:00
Andrew Kane
a37f5eea4a Improved compare method [skip ci] 2022-02-15 11:50:35 -08:00
Andrew Kane
4bdb27e85a Fixed style [skip ci] 2022-02-14 23:15:27 -08:00
Andrew Kane
38f869e0bd Added test for 100% recall 2022-02-14 23:13:36 -08:00
Andrew Kane
ad8acc00d4 Fixed spacing [skip ci] 2022-02-14 21:58:20 -08:00
Andrew Kane
01926a418e Print operator name for failed recall test [skip ci] 2022-02-14 21:52:31 -08:00
Andrew Kane
9658d3c1ad Made recall test less flaky 2022-02-14 21:49:08 -08:00
Andrew Kane
6b9c6516f4 Improved code 2022-02-14 21:41:27 -08:00
Andrew Kane
88be03a3fa Removed unused code 2022-02-14 21:37:53 -08:00
Andrew Kane
bf5b2c8d7e Use tuple id directly 2022-02-14 21:31:08 -08:00
Andrew Kane
ac65ec2856 Improved code [skip ci] 2022-02-14 19:38:59 -08:00
Andrew Kane
7bba0e2a01 Test recall for all operators 2022-02-14 15:22:53 -08:00
Andrew Kane
c35e9f3b84 Removed comment [skip ci] 2022-02-13 03:53:30 -08:00
Andrew Kane
9549d93260 Improved recall test 2022-02-13 03:08:39 -08:00
Andrew Kane
310809d0e5 Fixed warnings 2022-02-13 03:03:17 -08:00
Andrew Kane
ce72ca8620 Switched to heap for nearest lists for performance 2022-02-13 02:59:27 -08:00
Andrew Kane
4f2c937a1f Improved recall test 2022-02-13 02:35:28 -08:00
24 changed files with 445 additions and 155 deletions

View File

@@ -1,6 +1,6 @@
root = true
[*.{c,h,pl}]
[*.{c,h,pl,pm}]
indent_style = tab
indent_size = tab
tab_width = 4

View File

@@ -7,12 +7,12 @@ jobs:
fail-fast: false
matrix:
os: [ubuntu-latest]
postgres: [14, 13, 12, 11, 10, 9.6]
postgres: [15, 14, 13, 12, 11, 10]
include:
- os: macos-latest
postgres: 14
steps:
- uses: actions/checkout@v2
- uses: actions/checkout@v3
- uses: ankane/setup-postgres@v1
with:
postgres-version: ${{ matrix.postgres }}
@@ -33,6 +33,6 @@ jobs:
- if: ${{ startsWith(matrix.os, 'macos') }}
run: |
brew install cpanm && cpanm IPC::Run
wget -q https://github.com/postgres/postgres/archive/refs/tags/REL_14_1.tar.gz
tar xf REL_14_1.tar.gz
make prove_installcheck PROVE=prove PERL5LIB=postgres-REL_14_1/src/test/perl
wget -q https://github.com/postgres/postgres/archive/refs/tags/REL_14_5.tar.gz
tar xf REL_14_5.tar.gz
make prove_installcheck PROVE=prove PROVE_FLAGS="-I ./postgres-REL_14_5/src/test/perl" PERL5LIB="/Users/runner/perl5/lib/perl5"

View File

@@ -1,3 +1,16 @@
## 0.3.0 (2022-10-15)
- Added support for Postgres 15
- Dropped support for Postgres 9.6
## 0.2.7 (2022-07-31)
- Fixed `unexpected data beyond EOF` error
## 0.2.6 (2022-05-22)
- Improved performance of index creation for Postgres < 12
## 0.2.5 (2022-02-11)
- Reduced memory usage during index creation

View File

@@ -1,9 +1,9 @@
FROM postgres:14
FROM postgres:15
COPY . /tmp/pgvector
RUN apt-get update && \
apt-get install -y --no-install-recommends build-essential postgresql-server-dev-14 && \
apt-get install -y --no-install-recommends build-essential postgresql-server-dev-15 && \
cd /tmp/pgvector && \
make clean && \
make OPTFLAGS="" && \
@@ -11,6 +11,6 @@ RUN apt-get update && \
mkdir /usr/share/doc/pgvector && \
cp LICENSE README.md /usr/share/doc/pgvector && \
rm -r /tmp/pgvector && \
apt-get remove -y build-essential postgresql-server-dev-14 && \
apt-get remove -y build-essential postgresql-server-dev-15 && \
apt-get autoremove -y && \
rm -rf /var/lib/apt/lists/*

View File

@@ -2,7 +2,7 @@
"name": "vector",
"abstract": "Open-source vector similarity search for Postgres",
"description": "Supports L2 distance, inner product, and cosine distance",
"version": "0.2.5",
"version": "0.3.0",
"maintainer": [
"Andrew Kane <andrew@ankane.org>"
],
@@ -12,7 +12,7 @@
"prereqs": {
"runtime": {
"requires": {
"PostgreSQL": "9.6.0"
"PostgreSQL": "10.0"
}
}
},
@@ -20,7 +20,7 @@
"vector": {
"file": "sql/vector.sql",
"docfile": "README.md",
"version": "0.2.5",
"version": "0.3.0",
"abstract": "Open-source vector similarity search for Postgres"
}
},

View File

@@ -1,5 +1,5 @@
EXTENSION = vector
EXTVERSION = 0.2.5
EXTVERSION = 0.3.0
MODULE_big = vector
DATA = $(wildcard sql/*--*.sql)
@@ -40,6 +40,9 @@ PG_CONFIG ?= pg_config
PGXS := $(shell $(PG_CONFIG) --pgxs)
include $(PGXS)
# for Postgres 15
PROVE_FLAGS += -I ./test/perl
prove_installcheck:
rm -rf $(CURDIR)/tmp_check
cd $(srcdir) && TESTDIR='$(CURDIR)' PATH="$(bindir):$$PATH" PGPORT='6$(DEF_PGPORT)' PG_REGRESS='$(top_builddir)/src/test/regress/pg_regress' $(PROVE) $(PG_PROVE_FLAGS) $(PROVE_FLAGS) $(if $(PROVE_TESTS),$(PROVE_TESTS),test/t/*.pl)

View File

@@ -3,9 +3,9 @@
Open-source vector similarity search for Postgres
```sql
CREATE TABLE table (column vector(3));
CREATE INDEX ON table USING ivfflat (column vector_l2_ops);
SELECT * FROM table ORDER BY column <-> '[1,2,3]' LIMIT 5;
CREATE TABLE items (embedding vector(3));
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops);
SELECT * FROM items ORDER BY embedding <-> '[1,2,3]' LIMIT 5;
```
Supports L2 distance, inner product, and cosine distance
@@ -14,10 +14,10 @@ Supports L2 distance, inner product, and cosine distance
## Installation
Compile and install the extension (supports Postgres 9.6+)
Compile and install the extension (supports Postgres 10+)
```sh
git clone --branch v0.2.5 https://github.com/pgvector/pgvector.git
git clone --branch v0.3.0 https://github.com/pgvector/pgvector.git
cd pgvector
make
make install # may need sudo
@@ -33,22 +33,22 @@ You can also install it with [Docker](#docker), [Homebrew](#homebrew), or [PGXN]
## Getting Started
Create a vector column with 3 dimensions (replace `table` and `column` with non-reserved names)
Create a vector column with 3 dimensions
```sql
CREATE TABLE table (column vector(3));
CREATE TABLE items (embedding vector(3));
```
Insert values
```sql
INSERT INTO table VALUES ('[1,2,3]'), ('[4,5,6]');
INSERT INTO items VALUES ('[1,2,3]'), ('[4,5,6]');
```
Get the nearest neighbor by L2 distance
```sql
SELECT * FROM table ORDER BY column <-> '[3,1,2]' LIMIT 1;
SELECT * FROM items ORDER BY embedding <-> '[3,1,2]' LIMIT 1;
```
Also supports inner product (`<#>`) and cosine distance (`<=>`)
@@ -62,19 +62,19 @@ Speed up queries with an approximate index. Add an index for each distance funct
L2 distance
```sql
CREATE INDEX ON table USING ivfflat (column vector_l2_ops);
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops);
```
Inner product
```sql
CREATE INDEX ON table USING ivfflat (column vector_ip_ops);
CREATE INDEX ON items USING ivfflat (embedding vector_ip_ops);
```
Cosine distance
```sql
CREATE INDEX ON table USING ivfflat (column vector_cosine_ops);
CREATE INDEX ON items USING ivfflat (embedding vector_cosine_ops);
```
Indexes should be created after the table has some data for optimal clustering. Also, unlike typical indexes which only affect performance, you may see different results for queries after adding an approximate index.
@@ -84,7 +84,7 @@ Indexes should be created after the table has some data for optimal clustering.
Specify the number of inverted lists (100 by default)
```sql
CREATE INDEX ON table USING ivfflat (column opclass) WITH (lists = 100);
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100);
```
A [good place to start](https://github.com/facebookresearch/faiss/issues/112) is `4 * sqrt(rows)`
@@ -131,10 +131,20 @@ Note: `tuples_done` and `tuples_total` are only populated during the `loading tu
Consider [partial indexes](https://www.postgresql.org/docs/current/indexes-partial.html) for queries with a `WHERE` clause
```sql
CREATE INDEX ON table USING ivfflat (column opclass) WHERE (other_column = 123);
SELECT * FROM items WHERE category_id = 123 ORDER BY embedding <-> '[3,1,2]' LIMIT 5;
```
To index many different values of `other_column`, consider [partitioning](https://www.postgresql.org/docs/current/ddl-partitioning.html) on `other_column`.
can be indexed with:
```sql
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WHERE (category_id = 123);
```
To index many different values of `category_id`, consider [partitioning](https://www.postgresql.org/docs/current/ddl-partitioning.html) on `category_id`.
```sql
CREATE TABLE items (embedding vector(3), category_id int) PARTITION BY LIST(category_id);
```
## Performance
@@ -147,7 +157,7 @@ SET max_parallel_workers_per_gather = 4;
To speed up queries with an index, increase the number of inverted lists (at the expense of recall).
```sql
CREATE INDEX ON table USING ivfflat (column opclass) WITH (lists = 1000);
CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 1000);
```
## Reference
@@ -185,8 +195,10 @@ Libraries that use pgvector:
- [pgvector-ruby](https://github.com/pgvector/pgvector-ruby) (Ruby)
- [pgvector-node](https://github.com/pgvector/pgvector-node) (Node.js)
- [pgvector-go](https://github.com/pgvector/pgvector-go) (Go)
- [pgvector-php](https://github.com/pgvector/pgvector-php) (PHP)
- [pgvector-rust](https://github.com/pgvector/pgvector-rust) (Rust)
- [pgvector-cpp](https://github.com/pgvector/pgvector-cpp) (C++)
- [pgvector-elixir](https://github.com/pgvector/pgvector-elixir) (Elixir)
## Frequently Asked Questions
@@ -220,14 +232,14 @@ This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres).
You can also build the image manually
```sh
git clone --branch v0.2.5 https://github.com/pgvector/pgvector.git
git clone --branch v0.3.0 https://github.com/pgvector/pgvector.git
cd pgvector
docker build -t pgvector .
```
### Homebrew
On Mac with Homebrew Postgres, you can use:
With Homebrew Postgres, you can use:
```sh
brew install pgvector/brew/pgvector

View File

@@ -0,0 +1,2 @@
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
\echo Use "ALTER EXTENSION vector UPDATE TO '0.2.6'" to load this file. \quit

View File

@@ -0,0 +1,2 @@
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
\echo Use "ALTER EXTENSION vector UPDATE TO '0.2.7'" to load this file. \quit

View File

@@ -0,0 +1,2 @@
-- complain if script is sourced in psql, rather than via CREATE EXTENSION
\echo Use "ALTER EXTENSION vector UPDATE TO '0.3.0'" to load this file. \quit

View File

@@ -80,7 +80,11 @@ SampleCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
if (buildstate->rowstoskip <= 0)
{
#if PG_VERSION_NUM >= 150000
int k = (int) (targsamples * sampler_random_fract(&buildstate->rstate.randstate));
#else
int k = (int) (targsamples * sampler_random_fract(buildstate->rstate.randstate));
#endif
Assert(k >= 0 && k < targsamples);
VectorArraySet(samples, k, DatumGetVector(value));
@@ -112,13 +116,13 @@ SampleRows(IvfflatBuildState * buildstate)
#if PG_VERSION_NUM >= 120000
table_index_build_range_scan(buildstate->heap, buildstate->index, buildstate->indexInfo,
false, true, true, targblock, 1, SampleCallback, (void *) buildstate, NULL);
false, true, false, targblock, 1, SampleCallback, (void *) buildstate, NULL);
#elif PG_VERSION_NUM >= 110000
IndexBuildHeapRangeScan(buildstate->heap, buildstate->index, buildstate->indexInfo,
true, true, targblock, 1, SampleCallback, (void *) buildstate, NULL);
false, true, targblock, 1, SampleCallback, (void *) buildstate, NULL);
#else
IndexBuildHeapRangeScan(buildstate->heap, buildstate->index, buildstate->indexInfo,
true, true, targblock, 1, SampleCallback, (void *) buildstate);
false, true, targblock, 1, SampleCallback, (void *) buildstate);
#endif
}
}
@@ -167,18 +171,18 @@ BuildCallback(Relation index, CALLBACK_ITEM_POINTER, Datum *values,
#ifdef IVFFLAT_KMEANS_DEBUG
buildstate->inertia += minDistance;
buildstate->listSums[closestCenter] += minDistance;
buildstate->listCounts[closestCenter]++;
#endif
/* Create a virtual tuple */
ExecClearTuple(slot);
slot->tts_values[0] = Int32GetDatum(closestCenter);
slot->tts_isnull[0] = false;
slot->tts_values[1] = Int32GetDatum(ItemPointerGetBlockNumberNoCheck(tid));
slot->tts_values[1] = PointerGetDatum(tid);
slot->tts_isnull[1] = false;
slot->tts_values[2] = Int32GetDatum(ItemPointerGetOffsetNumberNoCheck(tid));
slot->tts_values[2] = value;
slot->tts_isnull[2] = false;
slot->tts_values[3] = value;
slot->tts_isnull[3] = false;
ExecStoreVirtualTuple(slot);
/*
@@ -200,8 +204,6 @@ GetNextTuple(Tuplesortstate *sortstate, TupleDesc tupdesc, TupleTableSlot *slot,
{
Datum value;
bool isnull;
int tupblk;
int tupoff;
#if PG_VERSION_NUM >= 100000
if (tuplesort_gettupleslot(sortstate, true, false, slot, NULL))
@@ -210,13 +212,11 @@ GetNextTuple(Tuplesortstate *sortstate, TupleDesc tupdesc, TupleTableSlot *slot,
#endif
{
*list = DatumGetInt32(slot_getattr(slot, 1, &isnull));
tupblk = DatumGetInt32(slot_getattr(slot, 2, &isnull));
tupoff = DatumGetInt32(slot_getattr(slot, 3, &isnull));
value = slot_getattr(slot, 4, &isnull);
value = slot_getattr(slot, 3, &isnull);
/* Form the index tuple */
*itup = index_form_tuple(tupdesc, &value, &isnull);
ItemPointerSet(&(*itup)->t_tid, tupblk, tupoff);
(*itup)->t_tid = *((ItemPointer) DatumGetPointer(slot_getattr(slot, 2, &isnull)));
}
else
*list = -1;
@@ -259,7 +259,7 @@ InsertTuples(Relation index, IvfflatBuildState * buildstate, ForkNumber forkNum)
CHECK_FOR_INTERRUPTS();
buf = IvfflatNewBuffer(index, forkNum);
IvfflatInitPage(index, &buf, &page, &state);
IvfflatInitRegisterPage(index, &buf, &page, &state);
startPage = BufferGetBlockNumber(buf);
@@ -325,17 +325,16 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
/* Create tuple description for sorting */
#if PG_VERSION_NUM >= 120000
buildstate->tupdesc = CreateTemplateTupleDesc(4);
buildstate->tupdesc = CreateTemplateTupleDesc(3);
#else
buildstate->tupdesc = CreateTemplateTupleDesc(4, false);
buildstate->tupdesc = CreateTemplateTupleDesc(3, false);
#endif
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 1, "list", INT4OID, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 2, "blkno", INT4OID, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 3, "offset", INT4OID, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 2, "tid", TIDOID, -1, 0);
#if PG_VERSION_NUM >= 110000
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 4, "vector", RelationGetDescr(index)->attrs[0].atttypid, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 3, "vector", RelationGetDescr(index)->attrs[0].atttypid, -1, 0);
#else
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 4, "vector", RelationGetDescr(index)->attrs[0]->atttypid, -1, 0);
TupleDescInitEntry(buildstate->tupdesc, (AttrNumber) 3, "vector", RelationGetDescr(index)->attrs[0]->atttypid, -1, 0);
#endif
#if PG_VERSION_NUM >= 120000
@@ -352,6 +351,8 @@ InitBuildState(IvfflatBuildState * buildstate, Relation heap, Relation index, In
#ifdef IVFFLAT_KMEANS_DEBUG
buildstate->inertia = 0;
buildstate->listSums = palloc0(sizeof(double) * buildstate->lists);
buildstate->listCounts = palloc0(sizeof(int) * buildstate->lists);
#endif
}
@@ -364,6 +365,11 @@ FreeBuildState(IvfflatBuildState * buildstate)
pfree(buildstate->centers);
pfree(buildstate->listInfo);
pfree(buildstate->normvec);
#ifdef IVFFLAT_KMEANS_DEBUG
pfree(buildstate->listSums);
pfree(buildstate->listCounts);
#endif
}
/*
@@ -410,7 +416,7 @@ CreateMetaPage(Relation index, int dimensions, int lists, ForkNumber forkNum)
IvfflatMetaPage metap;
buf = IvfflatNewBuffer(index, forkNum);
IvfflatInitPage(index, &buf, &page, &state);
IvfflatInitRegisterPage(index, &buf, &page, &state);
/* Set metapage data */
metap = IvfflatPageGetMeta(page);
@@ -443,7 +449,7 @@ CreateListPages(Relation index, VectorArray centers, int dimensions,
list = palloc(itemsz);
buf = IvfflatNewBuffer(index, forkNum);
IvfflatInitPage(index, &buf, &page, &state);
IvfflatInitRegisterPage(index, &buf, &page, &state);
for (i = 0; i < lists; i++)
{
@@ -471,6 +477,51 @@ CreateListPages(Relation index, VectorArray centers, int dimensions,
pfree(list);
}
/*
* Print k-means metrics
*/
#ifdef IVFFLAT_KMEANS_DEBUG
static void
PrintKmeansMetrics(IvfflatBuildState * buildstate)
{
elog(INFO, "inertia: %.3e", buildstate->inertia);
/* Calculate Davies-Bouldin index */
if (buildstate->lists > 1)
{
double db = 0.0;
/* Calculate average distance */
for (int i = 0; i < buildstate->lists; i++)
{
if (buildstate->listCounts[i] > 0)
buildstate->listSums[i] /= buildstate->listCounts[i];
}
for (int i = 0; i < buildstate->lists; i++)
{
double max = 0.0;
double distance;
for (int j = 0; j < buildstate->lists; j++)
{
if (j == i)
continue;
distance = DatumGetFloat8(FunctionCall2Coll(buildstate->procinfo, buildstate->collation, PointerGetDatum(VectorArrayGet(buildstate->centers, i)), PointerGetDatum(VectorArrayGet(buildstate->centers, j))));
distance = (buildstate->listSums[i] + buildstate->listSums[j]) / distance;
if (distance > max)
max = distance;
}
db += max;
}
db /= buildstate->lists;
elog(INFO, "davies-bouldin: %.3f", db);
}
}
#endif
/*
* Create entry pages
*/
@@ -509,7 +560,7 @@ CreateEntryPages(IvfflatBuildState * buildstate, ForkNumber forkNum)
tuplesort_performsort(buildstate->sortstate);
#ifdef IVFFLAT_KMEANS_DEBUG
elog(INFO, "inertia: %.3e", buildstate->inertia);
PrintKmeansMetrics(buildstate);
#endif
/* Insert */

View File

@@ -14,8 +14,8 @@
#include "portability/instr_time.h"
#endif
#if PG_VERSION_NUM < 90600
#error "Requires PostgreSQL 9.6+"
#if PG_VERSION_NUM < 100000
#error "Requires PostgreSQL 10+"
#endif
/* Support functions */
@@ -62,11 +62,6 @@
#define IvfflatBench(name, code) (code)
#endif
#if PG_VERSION_NUM < 100000
#define ItemPointerGetBlockNumberNoCheck ItemPointerGetBlockNumber
#define ItemPointerGetOffsetNumberNoCheck ItemPointerGetOffsetNumber
#endif
/* Variables */
extern int ivfflat_probes;
@@ -122,6 +117,8 @@ typedef struct IvfflatBuildState
#ifdef IVFFLAT_KMEANS_DEBUG
double inertia;
double *listSums;
int *listCounts;
#endif
/* Sampling */
@@ -165,6 +162,7 @@ typedef IvfflatListData * IvfflatList;
typedef struct IvfflatScanList
{
pairingheap_node ph_node;
BlockNumber startPage;
double distance;
} IvfflatScanList;
@@ -186,6 +184,8 @@ typedef struct IvfflatScanOpaqueData
FmgrInfo *normprocinfo;
Oid collation;
/* Lists */
pairingheap *listQueue;
IvfflatScanList lists[FLEXIBLE_ARRAY_MEMBER]; /* must come last */
} IvfflatScanOpaqueData;
@@ -208,7 +208,8 @@ void IvfflatUpdateList(Relation index, GenericXLogState *state, ListInfo listIn
void IvfflatCommitBuffer(Buffer buf, GenericXLogState *state);
void IvfflatAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state, ForkNumber forkNum);
Buffer IvfflatNewBuffer(Relation index, ForkNumber forkNum);
void IvfflatInitPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state);
void IvfflatInitPage(Buffer buf, Page page);
void IvfflatInitRegisterPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state);
/* Index access methods */
IndexBuildResult *ivfflatbuild(Relation heap, Relation index, IndexInfo *indexInfo);

View File

@@ -53,18 +53,6 @@ FindInsertPage(Relation rel, Datum *values, BlockNumber *insertPage, ListInfo *
}
}
/*
* Prepare to insert an index tuple
*/
static void
LoadInsertPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state, BlockNumber insertPage)
{
*buf = ReadBuffer(index, insertPage);
LockBuffer(*buf, BUFFER_LOCK_EXCLUSIVE);
*state = GenericXLogStart(index);
*page = GenericXLogRegisterBuffer(*state, *buf, 0);
}
/*
* Insert a tuple into the index
*/
@@ -87,11 +75,18 @@ InsertTuple(Relation rel, IndexTuple itup, Relation heapRel, Datum *values)
itemsz = MAXALIGN(IndexTupleSize(itup));
Assert(itemsz <= BLCKSZ - MAXALIGN(SizeOfPageHeaderData) - MAXALIGN(sizeof(IvfflatPageOpaqueData)));
LoadInsertPage(rel, &buf, &page, &state, insertPage);
/* Find a page to insert the item */
while (PageGetFreeSpace(page) < itemsz)
for (;;)
{
buf = ReadBuffer(rel, insertPage);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
state = GenericXLogStart(rel);
page = GenericXLogRegisterBuffer(state, buf, 0);
if (PageGetFreeSpace(page) >= itemsz)
break;
insertPage = IvfflatPageGetOpaque(page)->nextblkno;
if (BlockNumberIsValid(insertPage))
@@ -99,15 +94,29 @@ InsertTuple(Relation rel, IndexTuple itup, Relation heapRel, Datum *values)
/* Move to next page */
GenericXLogAbort(state);
UnlockReleaseBuffer(buf);
LoadInsertPage(rel, &buf, &page, &state, insertPage);
}
else
{
/* Add a new page */
IvfflatAppendPage(rel, &buf, &page, &state, MAIN_FORKNUM);
Buffer newbuf = IvfflatNewBuffer(rel, MAIN_FORKNUM);
Page newpage = GenericXLogRegisterBuffer(state, buf, GENERIC_XLOG_FULL_IMAGE);
insertPage = BufferGetBlockNumber(buf);
insertPage = BufferGetBlockNumber(newbuf);
/* Update previous buffer */
IvfflatPageGetOpaque(page)->nextblkno = insertPage;
/* Init page */
IvfflatInitPage(newbuf, newpage);
/* Commit */
MarkBufferDirty(buf);
MarkBufferDirty(newbuf);
GenericXLogFinish(state);
/* Unlock */
UnlockReleaseBuffer(buf);
UnlockReleaseBuffer(newbuf);
}
}

View File

@@ -5,6 +5,16 @@
#include "ivfflat.h"
#include "miscadmin.h"
#if PG_VERSION_NUM >= 150000
#include "common/pg_prng.h"
#endif
#if PG_VERSION_NUM >= 150000
#define RandomDouble() pg_prng_double(&pg_global_prng_state)
#else
#define RandomDouble() (((double) random()) / MAX_RANDOM_VALUE)
#endif
/*
* Initialize with kmeans++
*
@@ -66,7 +76,7 @@ InitCenters(Relation index, VectorArray samples, VectorArray centers, float *low
break;
/* Choose new center using weighted probability distribution. */
choice = sum * (((double) random()) / MAX_RANDOM_VALUE);
choice = sum * RandomDouble();
for (j = 0; j < numSamples - 1; j++)
{
choice -= weight[j];
@@ -145,7 +155,7 @@ QuickCenters(Relation index, VectorArray samples, VectorArray centers)
vec->dim = dimensions;
for (j = 0; j < dimensions; j++)
vec->x[j] = ((double) random()) / MAX_RANDOM_VALUE;
vec->x[j] = RandomDouble();
/* Normalize if needed (only needed for random centers) */
if (normprocinfo != NULL)
@@ -403,7 +413,7 @@ ElkanKmeans(Relation index, VectorArray samples, VectorArray centers)
{
/* TODO Handle empty centers properly */
for (k = 0; k < dimensions; k++)
vec->x[k] = ((double) random()) / MAX_RANDOM_VALUE;
vec->x[k] = RandomDouble();
}
/* Normalize if needed */

View File

@@ -1,5 +1,7 @@
#include "postgres.h"
#include <float.h>
#include "access/relscan.h"
#include "ivfflat.h"
#include "miscadmin.h"
@@ -17,14 +19,12 @@
* Compare list distances
*/
static int
CompareLists(const void *a, const void *b)
CompareLists(const pairingheap_node *a, const pairingheap_node *b, void *arg)
{
double diff = (((IvfflatScanList *) a)->distance - ((IvfflatScanList *) b)->distance);
if (diff > 0)
if (((const IvfflatScanList *) a)->distance > ((const IvfflatScanList *) b)->distance)
return 1;
if (diff < 0)
if (((const IvfflatScanList *) a)->distance < ((const IvfflatScanList *) b)->distance)
return -1;
return 0;
@@ -45,6 +45,8 @@ GetScanLists(IndexScanDesc scan, Datum value)
int listCount = 0;
IvfflatScanOpaque so = (IvfflatScanOpaque) scan->opaque;
double distance;
IvfflatScanList *scanlist;
double maxDistance = DBL_MAX;
/* Search all list pages */
while (BlockNumberIsValid(nextblkno))
@@ -62,22 +64,39 @@ GetScanLists(IndexScanDesc scan, Datum value)
/* Use procinfo from the index instead of scan key for performance */
distance = DatumGetFloat8(FunctionCall2Coll(so->procinfo, so->collation, PointerGetDatum(&list->center), value));
so->lists[listCount].startPage = list->startPage;
so->lists[listCount].distance = distance;
listCount++;
if (listCount < so->probes)
{
scanlist = &so->lists[listCount];
scanlist->startPage = list->startPage;
scanlist->distance = distance;
listCount++;
/* Add to heap */
pairingheap_add(so->listQueue, &scanlist->ph_node);
/* Calculate max distance */
if (listCount == so->probes)
maxDistance = ((IvfflatScanList *) pairingheap_first(so->listQueue))->distance;
}
else if (distance < maxDistance)
{
/* Remove */
scanlist = (IvfflatScanList *) pairingheap_remove_first(so->listQueue);
/* Reuse */
scanlist->startPage = list->startPage;
scanlist->distance = distance;
pairingheap_add(so->listQueue, &scanlist->ph_node);
/* Update max distance */
maxDistance = ((IvfflatScanList *) pairingheap_first(so->listQueue))->distance;
}
}
nextblkno = IvfflatPageGetOpaque(cpage)->nextblkno;
UnlockReleaseBuffer(cbuf);
}
/* Sort by distance */
/* TODO Use heap for performance */
qsort(so->lists, listCount, sizeof(IvfflatScanList), CompareLists);
if (so->probes > listCount)
so->probes = listCount;
}
/*
@@ -95,7 +114,6 @@ GetScanItems(IndexScanDesc scan, Datum value)
OffsetNumber maxoffno;
Datum datum;
bool isnull;
int i;
TupleDesc tupdesc = RelationGetDescr(scan->indexRelation);
#if PG_VERSION_NUM >= 120000
@@ -112,9 +130,9 @@ GetScanItems(IndexScanDesc scan, Datum value)
BufferAccessStrategy bas = GetAccessStrategy(BAS_BULKREAD);
/* Search closest probes lists */
for (i = 0; i < so->probes; i++)
while (!pairingheap_is_empty(so->listQueue))
{
searchPage = so->lists[i].startPage;
searchPage = ((IvfflatScanList *) pairingheap_remove_first(so->listQueue))->startPage;
/* Search all entry pages for list */
while (BlockNumberIsValid(searchPage))
@@ -138,12 +156,10 @@ GetScanItems(IndexScanDesc scan, Datum value)
ExecClearTuple(slot);
slot->tts_values[0] = FunctionCall2Coll(so->procinfo, so->collation, datum, value);
slot->tts_isnull[0] = false;
slot->tts_values[1] = Int32GetDatum((int) ItemPointerGetBlockNumberNoCheck(&itup->t_tid));
slot->tts_values[1] = PointerGetDatum(&itup->t_tid);
slot->tts_isnull[1] = false;
slot->tts_values[2] = Int32GetDatum((int) ItemPointerGetOffsetNumberNoCheck(&itup->t_tid));
slot->tts_values[2] = Int32GetDatum((int) searchPage);
slot->tts_isnull[2] = false;
slot->tts_values[3] = Int32GetDatum((int) searchPage);
slot->tts_isnull[3] = false;
ExecStoreVirtualTuple(slot);
tuplesort_puttupleslot(so->sortstate, slot);
@@ -171,13 +187,18 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
Oid sortOperators[] = {Float8LessOperator};
Oid sortCollations[] = {InvalidOid};
bool nullsFirstFlags[] = {false};
int probes = ivfflat_probes;
scan = RelationGetIndexScan(index, nkeys, norderbys);
lists = IvfflatGetLists(scan->indexRelation);
so = (IvfflatScanOpaque) palloc(offsetof(IvfflatScanOpaqueData, lists) + lists * sizeof(IvfflatScanList));
if (probes > lists)
probes = lists;
so = (IvfflatScanOpaque) palloc(offsetof(IvfflatScanOpaqueData, lists) + probes * sizeof(IvfflatScanList));
so->buf = InvalidBuffer;
so->first = true;
so->probes = probes;
/* Set support functions */
so->procinfo = index_getprocinfo(index, 1, IVFFLAT_DISTANCE_PROC);
@@ -186,14 +207,13 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
/* Create tuple description for sorting */
#if PG_VERSION_NUM >= 120000
so->tupdesc = CreateTemplateTupleDesc(4);
so->tupdesc = CreateTemplateTupleDesc(3);
#else
so->tupdesc = CreateTemplateTupleDesc(4, false);
so->tupdesc = CreateTemplateTupleDesc(3, false);
#endif
TupleDescInitEntry(so->tupdesc, (AttrNumber) 1, "distance", FLOAT8OID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 2, "blkno", INT4OID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 3, "offset", INT4OID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 4, "indexblkno", INT4OID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 2, "tid", TIDOID, -1, 0);
TupleDescInitEntry(so->tupdesc, (AttrNumber) 3, "indexblkno", INT4OID, -1, 0);
/* Prep sort */
#if PG_VERSION_NUM >= 110000
@@ -208,6 +228,8 @@ ivfflatbeginscan(Relation index, int nkeys, int norderbys)
so->slot = MakeSingleTupleTableSlot(so->tupdesc);
#endif
so->listQueue = pairingheap_allocate(CompareLists, scan);
scan->opaque = so;
return scan;
@@ -227,7 +249,7 @@ ivfflatrescan(IndexScanDesc scan, ScanKey keys, int nkeys, ScanKey orderbys, int
#endif
so->first = true;
so->probes = ivfflat_probes;
pairingheap_reset(so->listQueue);
if (keys && scan->numberOfKeys > 0)
memmove(scan->keyData, keys, scan->numberOfKeys * sizeof(ScanKeyData));
@@ -286,14 +308,13 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
if (tuplesort_gettupleslot(so->sortstate, true, so->slot, NULL))
#endif
{
BlockNumber blkno = DatumGetInt32(slot_getattr(so->slot, 2, &so->isnull));
OffsetNumber offset = DatumGetInt32(slot_getattr(so->slot, 3, &so->isnull));
BlockNumber indexblkno = DatumGetInt32(slot_getattr(so->slot, 4, &so->isnull));
ItemPointer tid = (ItemPointer) DatumGetPointer(slot_getattr(so->slot, 2, &so->isnull));
BlockNumber indexblkno = DatumGetInt32(slot_getattr(so->slot, 3, &so->isnull));
#if PG_VERSION_NUM >= 120000
ItemPointerSet(&scan->xs_heaptid, blkno, offset);
scan->xs_heaptid = *tid;
#else
ItemPointerSet(&scan->xs_ctup.t_self, blkno, offset);
scan->xs_ctup.t_self = *tid;
#endif
if (BufferIsValid(so->buf))
@@ -326,6 +347,7 @@ ivfflatendscan(IndexScanDesc scan)
if (BufferIsValid(so->buf))
ReleaseBuffer(so->buf);
pairingheap_free(so->listQueue);
tuplesort_end(so->sortstate);
pfree(so);

View File

@@ -107,13 +107,22 @@ IvfflatNewBuffer(Relation index, ForkNumber forkNum)
* Init page
*/
void
IvfflatInitPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state)
IvfflatInitPage(Buffer buf, Page page)
{
PageInit(page, BufferGetPageSize(buf), sizeof(IvfflatPageOpaqueData));
IvfflatPageGetOpaque(page)->nextblkno = InvalidBlockNumber;
IvfflatPageGetOpaque(page)->page_id = IVFFLAT_PAGE_ID;
}
/*
* Init and register page
*/
void
IvfflatInitRegisterPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state)
{
*state = GenericXLogStart(index);
*page = GenericXLogRegisterBuffer(*state, *buf, GENERIC_XLOG_FULL_IMAGE);
PageInit(*page, BufferGetPageSize(*buf), sizeof(IvfflatPageOpaqueData));
IvfflatPageGetOpaque(*page)->nextblkno = InvalidBlockNumber;
IvfflatPageGetOpaque(*page)->page_id = IVFFLAT_PAGE_ID;
IvfflatInitPage(*buf, *page);
}
/*
@@ -135,17 +144,27 @@ IvfflatCommitBuffer(Buffer buf, GenericXLogState *state)
void
IvfflatAppendPage(Relation index, Buffer *buf, Page *page, GenericXLogState **state, ForkNumber forkNum)
{
Buffer prevbuf = *buf;
/* Get new buffer */
*buf = IvfflatNewBuffer(index, forkNum);
Buffer newbuf = IvfflatNewBuffer(index, forkNum);
Page newpage = GenericXLogRegisterBuffer(*state, newbuf, GENERIC_XLOG_FULL_IMAGE);
/* Update and commit previous buffer */
IvfflatPageGetOpaque(*page)->nextblkno = BufferGetBlockNumber(*buf);
IvfflatCommitBuffer(prevbuf, *state);
/* Update the previous buffer */
IvfflatPageGetOpaque(*page)->nextblkno = BufferGetBlockNumber(newbuf);
/* Init new page */
IvfflatInitPage(index, buf, page, state);
IvfflatInitPage(newbuf, newpage);
/* Commit */
MarkBufferDirty(*buf);
MarkBufferDirty(newbuf);
GenericXLogFinish(*state);
/* Unlock */
UnlockReleaseBuffer(*buf);
*state = GenericXLogStart(index);
*page = GenericXLogRegisterBuffer(*state, newbuf, GENERIC_XLOG_FULL_IMAGE);
*buf = newbuf;
}
/*

View File

@@ -0,0 +1,8 @@
use PostgreSQL::Test::Cluster;
sub get_new_node
{
return PostgreSQL::Test::Cluster->new(@_);
}
1;

3
test/perl/TestLib.pm Normal file
View File

@@ -0,0 +1,3 @@
use PostgreSQL::Test::Utils;
1;

View File

@@ -7,6 +7,8 @@ use PostgresNode;
use TestLib;
use Test::More tests => 31;
my $dim = 32;
my $node_primary;
my $node_replica;
@@ -30,13 +32,15 @@ sub test_index_replay
$node_primary->poll_query_until('postgres', $caughtup_query)
or die "Timed out while waiting for replica 1 to catch up";
my $r1 = rand();
my $r2 = rand();
my $r3 = rand();
my @r = ();
for (1 .. $dim) {
push(@r, rand());
}
my $sql = join(",", @r);
my $queries = qq(
SET enable_seqscan = off;
SELECT * FROM tst ORDER BY v <-> '[$r1,$r2,$r3]' LIMIT 10;
SELECT * FROM tst ORDER BY v <-> '[$sql]' LIMIT 10;
);
# Run test queries and compare their result
@@ -50,6 +54,10 @@ sub test_index_replay
# Initialize primary node
$node_primary = get_new_node('primary');
$node_primary->init(allows_streaming => 1);
if ($dim > 32) {
# TODO use wal_keep_segments for Postgres < 13
$node_primary->append_conf('postgresql.conf', qq(wal_keep_size = 1GB));
}
$node_primary->start;
my $backup_name = 'my_backup';
@@ -64,9 +72,9 @@ $node_replica->start;
# Create ivfflat index on primary
$node_primary->safe_psql("postgres", "CREATE EXTENSION vector;");
$node_primary->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector(3));");
$node_primary->safe_psql("postgres", "CREATE TABLE tst (i int4, v vector($dim));");
$node_primary->safe_psql("postgres",
"INSERT INTO tst SELECT i % 10, ARRAY[random(), random(), random()] FROM generate_series(1, 100000) i;"
"INSERT INTO tst SELECT i % 10, (SELECT array_agg(random()) FROM generate_series(1, $dim)) FROM generate_series(1, 100000) i;"
);
$node_primary->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v);");
@@ -82,7 +90,7 @@ for my $i (1 .. 10)
test_index_replay("vacuum $i");
my ($start, $end) = (100001 + ($i - 1) * 10000, 100000 + $i * 10000);
$node_primary->safe_psql("postgres",
"INSERT INTO tst SELECT i % 10, ARRAY[random(), random(), random()] FROM generate_series($start, $end) i;"
"INSERT INTO tst SELECT i % 10, (SELECT array_agg(random()) FROM generate_series(1, $dim)) FROM generate_series($start, $end) i;"
);
test_index_replay("insert $i");
}

View File

@@ -2,15 +2,16 @@ use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More tests => 2;
use Test::More tests => 9;
my $node;
my @queries = ();
my @expected = ();
my @expected;
my $limit = 20;
sub test_recall
{
my ($probes, $min) = @_;
my ($probes, $min, $operator) = @_;
my $correct = 0;
my $total = 0;
@@ -18,7 +19,7 @@ sub test_recall
my $actual = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SET ivfflat.probes = $probes;
SELECT i FROM tst ORDER BY v <-> '$queries[$i]' LIMIT 10;
SELECT i FROM tst ORDER BY v $operator '$queries[$i]' LIMIT $limit;
));
my @actual_ids = split("\n", $actual);
my %actual_set = map { $_ => 1 } @actual_ids;
@@ -33,7 +34,7 @@ sub test_recall
}
}
cmp_ok($correct / $total, ">=", $min);
cmp_ok($correct / $total, ">=", $min, $operator);
}
# Initialize node
@@ -56,17 +57,32 @@ for (1..20) {
push(@queries, "[$r1,$r2,$r3]");
}
# Get exact results
foreach (@queries) {
my $res = $node->safe_psql("postgres", "SELECT i FROM tst ORDER BY v <-> '$_' LIMIT 10;");
push(@expected, $res);
# Check each index type
my @operators = ("<->", "<#>", "<=>");
foreach (@operators) {
my $operator = $_;
# 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);
}
# Add index
my $opclass;
if ($operator == "<->") {
$opclass = "vector_l2_ops";
} elsif ($operator == "<#>") {
$opclass = "vector_ip_ops";
} else {
$opclass = "vector_cosine_ops";
}
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v $opclass);");
# Test approximate results
test_recall(1, 0.75, $operator);
test_recall(10, 0.95, $operator);
test_recall(100, 1.0, $operator);
}
# Add index
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v);");
# Test approximate results
test_recall(1, 0.8);
# Test probes
test_recall(100, 1.0);

View File

@@ -0,0 +1,45 @@
use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More tests => 60;
# Initialize node
my $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 primary key, v vector(3));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT i, ARRAY[random(), random(), random()] FROM generate_series(1, 100000) i;"
);
# Check each index type
my @operators = ("<->", "<#>", "<=>");
foreach (@operators) {
my $operator = $_;
# Add index
my $opclass;
if ($operator == "<->") {
$opclass = "vector_l2_ops";
} elsif ($operator == "<#>") {
$opclass = "vector_ip_ops";
} else {
$opclass = "vector_cosine_ops";
}
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v $opclass);");
# Test 100% recall
for (1..20) {
my $i = int(rand() * 100000);
my $query = $node->safe_psql("postgres", "SELECT v FROM tst WHERE i = $i;");
my $res = $node->safe_psql("postgres", qq(
SET enable_seqscan = off;
SELECT v FROM tst ORDER BY v <-> '$query' LIMIT 1;
));
is($res, $query);
}
}

31
test/t/006_lists.pl Normal file
View File

@@ -0,0 +1,31 @@
use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More tests => 3;
# Initialize node
my $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 (v vector(3));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT ARRAY[random(), random(), random()] FROM generate_series(1, 100000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX lists50 ON tst USING ivfflat (v) WITH (lists = 50);");
$node->safe_psql("postgres", "CREATE INDEX lists100 ON tst USING ivfflat (v) WITH (lists = 100);");
# Test prefers more lists
my $res = $node->safe_psql("postgres", "EXPLAIN SELECT v FROM tst ORDER BY v <-> '[0.5,0.5,0.5]' LIMIT 10;");
like($res, qr/lists100/);
unlike($res, qr/lists50/);
# Test errors with too much memory
my ($ret, $stdout, $stderr) = $node->psql("postgres",
"CREATE INDEX lists10000 ON tst USING ivfflat (v) WITH (lists = 10000);"
);
like($stderr, qr/memory required is/);

33
test/t/007_inserts.pl Normal file
View File

@@ -0,0 +1,33 @@
use strict;
use warnings;
use PostgresNode;
use TestLib;
use Test::More tests => 3;
# Initialize node
my $node = get_new_node('node');
$node->init;
$node->start;
# Create table and index
$node->safe_psql("postgres", "CREATE EXTENSION vector;");
$node->safe_psql("postgres", "CREATE TABLE tst (v vector(768));");
$node->safe_psql("postgres",
"INSERT INTO tst SELECT (SELECT array_agg(random()) FROM generate_series(1, 768)) FROM generate_series(1, 10000) i;"
);
$node->safe_psql("postgres", "CREATE INDEX ON tst USING ivfflat (v);");
$node->pgbench(
"--no-vacuum --client=5 --transactions=100",
0,
[qr{actually processed}],
[qr{^$}],
"concurrent INSERTs",
{
"007_concurrent" => q(
BEGIN;
INSERT INTO tst SELECT (SELECT array_agg(random()) FROM generate_series(1, 768)) FROM generate_series(1, 10) i;
COMMIT;
),
}
);

View File

@@ -1,4 +1,4 @@
comment = 'vector data type and ivfflat access method'
default_version = '0.2.5'
default_version = '0.3.0'
module_pathname = '$libdir/vector'
relocatable = true