diff --git a/.dockerignore b/.dockerignore
deleted file mode 100644
index 6c60e6d5d..000000000
--- a/.dockerignore
+++ /dev/null
@@ -1,8 +0,0 @@
-/.git/
-/dist/
-/results/
-/tmp_check/
-/sql/vector--?.?.?.sql
-regression.*
-*.o
-*.so
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index e58c3b56d..9f541f131 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -8,6 +8,8 @@ jobs:
       fail-fast: false
       matrix:
         include:
+          - postgres: 19
+            os: ubuntu-24.04
           - postgres: 18
             os: ubuntu-24.04
           - postgres: 17
@@ -21,14 +23,14 @@ jobs:
           - postgres: 13
             os: ubuntu-22.04
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - uses: ankane/setup-postgres@v1
         with:
           postgres-version: ${{ matrix.postgres }}
           dev-files: true
       - run: make
         env:
-          PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
+          PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare ${{ matrix.postgres >= 18 && '-Wno-missing-field-initializers' || '' }}
       - run: |
           export PG_CONFIG=`which pg_config`
           sudo --preserve-env=PG_CONFIG make install
@@ -39,6 +41,38 @@ jobs:
           sudo apt-get update
           sudo apt-get install libipc-run-perl
       - run: make prove_installcheck
+  ubuntu_spr:
+    runs-on: ubuntu-24.04
+    if: ${{ !startsWith(github.ref_name, 'mac') && !startsWith(github.ref_name, 'windows') }}
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install Intel SDE
+        run: |
+          curl -o /tmp/sde.tar.xz https://downloadmirror.intel.com/831748/sde-external-9.44.0-2024-08-22-lin.tar.xz
+          mkdir /tmp/sde && tar -xvf /tmp/sde.tar.xz -C /tmp/sde/
+          sudo mv /tmp/sde/* /opt/sde && sudo ln -s /opt/sde/sde64 /usr/bin/sde
+      - name: Install Postgres
+        run: |
+          sudo apt-get install postgresql-16
+          sudo apt-get install postgresql-server-dev-16
+          sudo systemctl stop postgresql
+          pgdir=$(pg_config --bindir)
+          pgdata=/tmp/postgres_data
+          mkdir $pgdata
+          $pgdir/initdb -D $pgdata --username=$USER
+          sudo chmod 777 /var/run/postgresql/
+          sde -spr -mix -- $pgdir/pg_ctl -D $pgdata start
+      - run: make
+        env:
+          PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-sign-compare
+      - run: |
+          export PG_CONFIG=`which pg_config`
+          sudo --preserve-env=PG_CONFIG make install
+      - run: sde -spr -mix -- make installcheck
+      - if: ${{ failure() }}
+        run: cat regression.diffs
+      - name: Report AVX512 FP16 FMA instruction count
+        run: cat sde*.txt | grep -E "FMA.*PH"
   mac:
     runs-on: ${{ matrix.os }}
     if: ${{ !startsWith(github.ref_name, 'windows') }}
@@ -46,18 +80,18 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - postgres: 17
-            os: macos-15
+          - postgres: 18
+            os: macos-26
           - postgres: 14
-            os: macos-13
+            os: macos-15-intel
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - uses: ankane/setup-postgres@v1
         with:
           postgres-version: ${{ matrix.postgres }}
       - run: make
         env:
-          PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter
+          PG_CFLAGS: -DUSE_ASSERT_CHECKING -Wall -Wextra -Werror -Wno-unused-parameter -Wno-unknown-warning-option ${{ matrix.postgres >= 18 && '-Wno-missing-field-initializers' || '' }}
       - run: make install
       - run: make installcheck
       - if: ${{ failure() }}
@@ -70,27 +104,35 @@ jobs:
           tar xf $TAG.tar.gz
           mv postgres-$TAG postgres
         env:
-          TAG: ${{ matrix.postgres == 17 && 'REL_17_2' || 'REL_14_15' }}
+          TAG: ${{ matrix.postgres == 18 && 'REL_18_0' || 'REL_14_19' }}
       - run: make prove_installcheck PROVE_FLAGS="-I ./postgres/src/test/perl -I ./test/perl"
         env:
           PERL5LIB: /Users/runner/perl5/lib/perl5
       - run: make clean && $(brew --prefix llvm@$LLVM_VERSION)/bin/scan-build --status-bugs make
         env:
-          LLVM_VERSION: ${{ matrix.os == 'macos-15' && 18 || 15 }}
+          LLVM_VERSION: ${{ matrix.os == 'macos-26' && 20 || 18 }}
           PG_CFLAGS: -DUSE_ASSERT_CHECKING
   windows:
-    runs-on: windows-latest
+    runs-on: ${{ matrix.os }}
     if: ${{ !startsWith(github.ref_name, 'mac') }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - postgres: 17
+            os: windows-2025
+          - postgres: 14
+            os: windows-2022
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - uses: ankane/setup-postgres@v1
         with:
-          postgres-version: 14
+          postgres-version: ${{ matrix.postgres }}
       - run: |
           call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat" && ^
           nmake /NOLOGO /F Makefile.win && ^
           nmake /NOLOGO /F Makefile.win install && ^
-          nmake /NOLOGO /F Makefile.win installcheck && ^
+          nmake /NOLOGO /F Makefile.win installcheck ${{ matrix.postgres != 17 && 'PG_REGRESS=$(PGROOT)\bin\pg_regress' || '' }} && ^
           nmake /NOLOGO /F Makefile.win clean && ^
           nmake /NOLOGO /F Makefile.win uninstall
         shell: cmd
@@ -123,10 +165,10 @@ jobs:
     if: ${{ !startsWith(github.ref_name, 'mac') && !startsWith(github.ref_name, 'windows') }}
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - uses: ankane/setup-postgres-valgrind@v1
         with:
-          postgres-version: 17
+          postgres-version: 18
           check-ub: yes
       - run: make OPTFLAGS=""
       - run: sudo --preserve-env=PG_CONFIG make install
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 757f998d0..a3515593f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,3 +1,12 @@
+## 0.8.2 (unreleased)
+
+- Fixed `Index Searches` in `EXPLAIN` output for Postgres 18
+
+## 0.8.1 (2025-09-04)
+
+- Added support for Postgres 18 rc1
+- Improved performance of `binary_quantize` function
+
 ## 0.8.0 (2024-10-30)
 
 - Added support for iterative index scans
diff --git a/Dockerfile b/Dockerfile
index 936440928..7e06759d5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,8 +1,11 @@
+# syntax=docker/dockerfile:1
+
 ARG PG_MAJOR=17
-FROM postgres:$PG_MAJOR
+ARG DEBIAN_CODENAME=bookworm
+FROM postgres:$PG_MAJOR-$DEBIAN_CODENAME
 ARG PG_MAJOR
 
-COPY . /tmp/pgvector
+ADD https://github.com/pgvector/pgvector.git#v0.8.1 /tmp/pgvector
 
 RUN apt-get update && \
 		apt-mark hold locales && \
diff --git a/META.json b/META.json
index b9a68f62a..343518d1c 100644
--- a/META.json
+++ b/META.json
@@ -2,7 +2,7 @@
 	"name": "vector",
 	"abstract": "Open-source vector similarity search for Postgres",
 	"description": "Supports L2 distance, inner product, and cosine distance",
-	"version": "0.8.0",
+	"version": "0.8.1",
 	"maintainer": [
 		"Andrew Kane <andrew@ankane.org>"
 	],
@@ -20,7 +20,7 @@
 		"vector": {
 			"file": "sql/vector.sql",
 			"docfile": "README.md",
-			"version": "0.8.0",
+			"version": "0.8.1",
 			"abstract": "Open-source vector similarity search for Postgres"
 		}
 	},
diff --git a/Makefile b/Makefile
index 7a4b88caf..e89ee1342 100644
--- a/Makefile
+++ b/Makefile
@@ -1,10 +1,13 @@
 EXTENSION = vector
-EXTVERSION = 0.8.0
+EXTVERSION = 0.8.1
 
 MODULE_big = vector
 DATA = $(wildcard sql/*--*--*.sql)
 DATA_built = sql/$(EXTENSION)--$(EXTVERSION).sql
 OBJS = src/bitutils.o src/bitvec.o src/halfutils.o src/halfvec.o src/hnsw.o src/hnswbuild.o src/hnswinsert.o src/hnswscan.o src/hnswutils.o src/hnswvacuum.o src/ivfbuild.o src/ivfflat.o src/ivfinsert.o src/ivfkmeans.o src/ivfscan.o src/ivfutils.o src/ivfvacuum.o src/sparsevec.o src/vector.o
+ifneq ($(USE_AVX512), 0)
+	OBJS += src/halfutils_avx512.o
+endif
 HEADERS = src/halfvec.h src/sparsevec.h src/vector.h
 
 TESTS = $(wildcard test/sql/*.sql)
@@ -31,6 +34,9 @@ endif
 # - GCC (needs -ftree-vectorize OR -O3) - https://gcc.gnu.org/projects/tree-ssa/vectorization.html
 # - Clang (could use pragma instead) - https://llvm.org/docs/Vectorizers.html
 PG_CFLAGS += $(OPTFLAGS) -ftree-vectorize -fassociative-math -fno-signed-zeros -fno-trapping-math
+ifneq ($(USE_AVX512), 0)
+	PG_CFLAGS += -DUSE_AVX512
+endif
 
 # Debug GCC auto-vectorization
 # PG_CFLAGS += -fopt-info-vec
@@ -76,4 +82,9 @@ docker:
 .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=$(PG_MAJOR) --build-arg DEBIAN_CODENAME=bookworm -t pgvector/pgvector:pg$(PG_MAJOR) -t pgvector/pgvector:pg$(PG_MAJOR)-bookworm -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR) -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR)-bookworm .
+
+.PHONY: docker-release-trixie
+
+docker-release-trixie:
+	docker buildx build --push --pull --no-cache --platform linux/amd64,linux/arm64 --build-arg PG_MAJOR=$(PG_MAJOR) --build-arg DEBIAN_CODENAME=trixie -t pgvector/pgvector:pg$(PG_MAJOR)-trixie -t pgvector/pgvector:$(EXTVERSION)-pg$(PG_MAJOR)-trixie .
diff --git a/Makefile.win b/Makefile.win
index 8c62f9d5f..46ab3da64 100644
--- a/Makefile.win
+++ b/Makefile.win
@@ -1,5 +1,5 @@
 EXTENSION = vector
-EXTVERSION = 0.8.0
+EXTVERSION = 0.8.1
 
 DATA_built = sql\$(EXTENSION)--$(EXTVERSION).sql
 OBJS = src\bitutils.obj src\bitvec.obj src\halfutils.obj src\halfvec.obj src\hnsw.obj src\hnswbuild.obj src\hnswinsert.obj src\hnswscan.obj src\hnswutils.obj src\hnswvacuum.obj src\ivfbuild.obj src\ivfflat.obj src\ivfinsert.obj src\ivfkmeans.obj src\ivfscan.obj src\ivfutils.obj src\ivfvacuum.obj src\sparsevec.obj src\vector.obj
@@ -31,6 +31,9 @@ LIBDIR = $(PGROOT)\lib
 PKGLIBDIR = $(PGROOT)\lib
 SHAREDIR = $(PGROOT)\share
 
+# Use $(PGROOT)\bin\pg_regress for Postgres < 17
+PG_REGRESS = $(LIBDIR)\pgxs\src\test\regress\pg_regress
+
 CFLAGS = /nologo /I"$(INCLUDEDIR_SERVER)\port\win32_msvc" /I"$(INCLUDEDIR_SERVER)\port\win32" /I"$(INCLUDEDIR_SERVER)" /I"$(INCLUDEDIR)"
 
 CFLAGS = $(CFLAGS) $(PG_CFLAGS)
@@ -58,7 +61,7 @@ install: all
 	for %f in ($(HEADERS)) do copy %f "$(INCLUDEDIR_SERVER)\extension\$(EXTENSION)"
 
 installcheck:
-	"$(BINDIR)\pg_regress" --bindir="$(BINDIR)" $(REGRESS_OPTS) $(REGRESS)
+	"$(PG_REGRESS)" --bindir="$(BINDIR)" $(REGRESS_OPTS) $(REGRESS)
 
 uninstall:
 	del /f "$(PKGLIBDIR)\$(SHLIB)"
diff --git a/README.md b/README.md
index a1077cd45..0f7ee7a51 100644
--- a/README.md
+++ b/README.md
@@ -21,7 +21,7 @@ Compile and install the extension (supports Postgres 13+)
 
 ```sh
 cd /tmp
-git clone --branch v0.8.0 https://github.com/pgvector/pgvector.git
+git clone --branch v0.8.1 https://github.com/pgvector/pgvector.git
 cd pgvector
 make
 make install # may need sudo
@@ -33,20 +33,12 @@ You can also install it with [Docker](#docker), [Homebrew](#homebrew), [PGXN](#p
 
 ### Windows
 
-Ensure [C++ support in Visual Studio](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#download-and-install-the-tools) is installed, and run:
+Ensure [C++ support in Visual Studio](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#download-and-install-the-tools) is installed and run `x64 Native Tools Command Prompt for VS [version]` as administrator. Then use `nmake` to build:
 
 ```cmd
-call "C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvars64.bat"
-```
-
-Note: The exact path will vary depending on your Visual Studio version and edition
-
-Then use `nmake` to build:
-
-```cmd
-set "PGROOT=C:\Program Files\PostgreSQL\16"
+set "PGROOT=C:\Program Files\PostgreSQL\18"
 cd %TEMP%
-git clone --branch v0.8.0 https://github.com/pgvector/pgvector.git
+git clone --branch v0.8.1 https://github.com/pgvector/pgvector.git
 cd pgvector
 nmake /F Makefile.win
 nmake /F Makefile.win install
@@ -318,7 +310,9 @@ You can also speed up index creation by increasing the number of parallel worker
 SET max_parallel_maintenance_workers = 7; -- plus leader
 ```
 
-For a large number of workers, you may also need to increase `max_parallel_workers` (8 by default)
+For a large number of workers, you may need to increase `max_parallel_workers` (8 by default)
+
+The [index options](#index-options) also have a significant impact on build time (use the defaults unless seeing low recall)
 
 ### Indexing Progress
 
@@ -475,8 +469,6 @@ CREATE TABLE items (embedding vector(3), category_id int) PARTITION BY LIST(cate
 
 ## Iterative Index Scans
 
-*Added in 0.8.0*
-
 With approximate indexes, queries with filtering can return less results since filtering is applied *after* the index is scanned. Starting with 0.8.0, you can enable iterative index scans, which will automatically scan more of the index until enough results are found (or it reaches `hnsw.max_scan_tuples` or `ivfflat.max_probes`).
 
 Iterative scans can use strict or relaxed ordering.
@@ -500,9 +492,11 @@ With relaxed ordering, you can use a [materialized CTE](https://www.postgresql.o
 ```sql
 WITH relaxed_results AS MATERIALIZED (
     SELECT id, embedding <-> '[1,2,3]' AS distance FROM items WHERE category_id = 123 ORDER BY distance LIMIT 5
-) SELECT * FROM relaxed_results ORDER BY distance;
+) SELECT * FROM relaxed_results ORDER BY distance + 0;
 ```
 
+Note: `+ 0` is needed for Postgres 17+
+
 For queries that filter by distance, use a materialized CTE and place the distance filter outside of it for best performance (due to the [current behavior](https://www.postgresql.org/message-id/flat/CAOdR5yGUoMQ6j7M5hNUXrySzaqZVGf_Ne%2B8fwZMRKTFxU1nbJg%40mail.gmail.com) of the Postgres executor)
 
 ```sql
@@ -754,8 +748,6 @@ SELECT query, calls, ROUND((total_plan_time + total_exec_time) / calls) AS avg_t
     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
@@ -793,7 +785,7 @@ Go | [pgvector-go](https://github.com/pgvector/pgvector-go)
 Haskell | [pgvector-haskell](https://github.com/pgvector/pgvector-haskell)
 Java, Kotlin, Groovy, Scala | [pgvector-java](https://github.com/pgvector/pgvector-java)
 JavaScript, TypeScript | [pgvector-node](https://github.com/pgvector/pgvector-node)
-Julia | [pgvector-julia](https://github.com/pgvector/pgvector-julia)
+Julia | [Pgvector.jl](https://github.com/pgvector/Pgvector.jl)
 Lisp | [pgvector-lisp](https://github.com/pgvector/pgvector-lisp)
 Lua | [pgvector-lua](https://github.com/pgvector/pgvector-lua)
 Nim | [pgvector-nim](https://github.com/pgvector/pgvector-nim)
@@ -820,11 +812,11 @@ Yes, pgvector uses the write-ahead log (WAL), which allows for replication and p
 
 #### What if I want to index vectors with more than 2,000 dimensions?
 
-You can use [half-precision indexing](#half-precision-indexing) to index up to 4,000 dimensions or [binary quantization](#binary-quantization) to index up to 64,000 dimensions. Another option is [dimensionality reduction](https://en.wikipedia.org/wiki/Dimensionality_reduction).
+You can use [half-precision vectors](#half-precision-vectors) or [half-precision indexing](#half-precision-indexing) to index up to 4,000 dimensions or [binary quantization](#binary-quantization) to index up to 64,000 dimensions. Other options are [indexing subvectors](#indexing-subvectors) (for models that support it) or [dimensionality reduction](https://en.wikipedia.org/wiki/Dimensionality_reduction).
 
 #### Can I store vectors with different dimensions in the same column?
 
-You can use `vector` as the type (instead of `vector(3)`).
+You can use `vector` as the type (instead of `vector(n)`).
 
 ```sql
 CREATE TABLE embeddings (model_id bigint, item_id bigint, embedding vector, PRIMARY KEY (model_id, item_id));
@@ -1072,7 +1064,7 @@ l2_normalize(sparsevec) → sparsevec | Normalize with Euclidean norm | 0.7.0
 If your machine has multiple Postgres installations, specify the path to [pg_config](https://www.postgresql.org/docs/current/app-pgconfig.html) with:
 
 ```sh
-export PG_CONFIG=/Library/PostgreSQL/17/bin/pg_config
+export PG_CONFIG=/Library/PostgreSQL/18/bin/pg_config
 ```
 
 Then re-run the installation instructions (run `make clean` before `make` if needed). If `sudo` is needed for `make install`, use:
@@ -1083,11 +1075,11 @@ sudo --preserve-env=PG_CONFIG make install
 
 A few common paths on Mac are:
 
-- EDB installer - `/Library/PostgreSQL/17/bin/pg_config`
-- Homebrew (arm64) - `/opt/homebrew/opt/postgresql@17/bin/pg_config`
-- Homebrew (x86-64) - `/usr/local/opt/postgresql@17/bin/pg_config`
+- EDB installer - `/Library/PostgreSQL/18/bin/pg_config`
+- Homebrew (arm64) - `/opt/homebrew/opt/postgresql@18/bin/pg_config`
+- Homebrew (x86-64) - `/usr/local/opt/postgresql@18/bin/pg_config`
 
-Note: Replace `17` with your Postgres server version
+Note: Replace `18` with your Postgres server version
 
 ### Missing Header
 
@@ -1096,10 +1088,10 @@ If compilation fails with `fatal error: postgres.h: No such file or directory`,
 For Ubuntu and Debian, use:
 
 ```sh
-sudo apt install postgresql-server-dev-17
+sudo apt install postgresql-server-dev-18
 ```
 
-Note: Replace `17` with your Postgres server version
+Note: Replace `18` with your Postgres server version
 
 ### Missing SDK
 
@@ -1129,7 +1121,7 @@ If compilation fails with `Cannot open include file: 'postgres.h': No such file
 
 ### Mismatched Architecture
 
-If compilation fails with `error C2196: case value '4' already used`, make sure `vcvars64.bat` was called. Then run `nmake /F Makefile.win clean` and re-run the installation instructions.
+If compilation fails with `error C2196: case value '4' already used`, make sure you’re using the `x64 Native Tools Command Prompt`. Then run `nmake /F Makefile.win clean` and re-run the installation instructions.
 
 ### Missing Symbol
 
@@ -1146,17 +1138,32 @@ If installation fails with `Access is denied`, re-run the installation instructi
 Get the [Docker image](https://hub.docker.com/r/pgvector/pgvector) with:
 
 ```sh
-docker pull pgvector/pgvector:pg17
+docker pull pgvector/pgvector:pg18-trixie
 ```
 
-This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `17` with your Postgres server version, and run it the same way).
+This adds pgvector to the [Postgres image](https://hub.docker.com/_/postgres) (replace `18` with your Postgres server version, and run it the same way).
+
+Supported tags are:
+
+- `pg18-trixie`, `0.8.1-pg18-trixie`
+- `pg18-bookworm`, `0.8.1-pg18-bookworm`, `pg18`, `0.8.1-pg18`
+- `pg17-trixie`, `0.8.1-pg17-trixie`
+- `pg17-bookworm`, `0.8.1-pg17-bookworm`, `pg17`, `0.8.1-pg17`
+- `pg16-trixie`, `0.8.1-pg16-trixie`
+- `pg16-bookworm`, `0.8.1-pg16-bookworm`, `pg16`, `0.8.1-pg16`
+- `pg15-trixie`, `0.8.1-pg15-trixie`
+- `pg15-bookworm`, `0.8.1-pg15-bookworm`, `pg15`, `0.8.1-pg15`
+- `pg14-trixie`, `0.8.1-pg14-trixie`
+- `pg14-bookworm`, `0.8.1-pg14-bookworm`, `pg14`, `0.8.1-pg14`
+- `pg13-trixie`, `0.8.1-pg13-trixie`
+- `pg13-bookworm`, `0.8.1-pg13-bookworm`, `pg13`, `0.8.1-pg13`
 
 You can also build the image manually:
 
 ```sh
-git clone --branch v0.8.0 https://github.com/pgvector/pgvector.git
+git clone --branch v0.8.1 https://github.com/pgvector/pgvector.git
 cd pgvector
-docker build --pull --build-arg PG_MAJOR=17 -t myuser/pgvector .
+docker build --pull --build-arg PG_MAJOR=18 -t myuser/pgvector .
 ```
 
 If you increase `maintenance_work_mem`, make sure `--shm-size` is at least that size to avoid an error with parallel HNSW index builds.
@@ -1173,7 +1180,7 @@ With Homebrew Postgres, you can use:
 brew install pgvector
 ```
 
-Note: This only adds it to the `postgresql@17` and `postgresql@14` formulas
+Note: This only adds it to the `postgresql@18` and `postgresql@17` formulas
 
 ### PGXN
 
@@ -1188,29 +1195,29 @@ pgxn install vector
 Debian and Ubuntu packages are available from the [PostgreSQL APT Repository](https://wiki.postgresql.org/wiki/Apt). Follow the [setup instructions](https://wiki.postgresql.org/wiki/Apt#Quickstart) and run:
 
 ```sh
-sudo apt install postgresql-17-pgvector
+sudo apt install postgresql-18-pgvector
 ```
 
-Note: Replace `17` with your Postgres server version
+Note: Replace `18` with your Postgres server version
 
 ### Yum
 
 RPM packages are available from the [PostgreSQL Yum Repository](https://yum.postgresql.org/). Follow the [setup instructions](https://www.postgresql.org/download/linux/redhat/) for your distribution and run:
 
 ```sh
-sudo yum install pgvector_17
+sudo yum install pgvector_18
 # or
-sudo dnf install pgvector_17
+sudo dnf install pgvector_18
 ```
 
-Note: Replace `17` with your Postgres server version
+Note: Replace `18` with your Postgres server version
 
 ### pkg
 
 Install the FreeBSD package with:
 
 ```sh
-pkg install postgresql16-pgvector
+pkg install postgresql17-pgvector
 ```
 
 or the port with:
diff --git a/sql/vector--0.8.0--0.8.1.sql b/sql/vector--0.8.0--0.8.1.sql
new file mode 100644
index 000000000..547bd44a3
--- /dev/null
+++ b/sql/vector--0.8.0--0.8.1.sql
@@ -0,0 +1,2 @@
+-- complain if script is sourced in psql, rather than via CREATE EXTENSION
+\echo Use "ALTER EXTENSION vector UPDATE TO '0.8.1'" to load this file. \quit
diff --git a/src/halfutils.c b/src/halfutils.c
index d16909409..bdc370a19 100644
--- a/src/halfutils.c
+++ b/src/halfutils.c
@@ -5,6 +5,7 @@
 
 #ifdef HALFVEC_DISPATCH
 #include <immintrin.h>
+#include "halfutils_avx512.h"
 
 #if defined(USE__GET_CPUID)
 #include <cpuid.h>
@@ -24,6 +25,8 @@ float		(*HalfvecInnerProduct) (int dim, half * ax, half * bx);
 double		(*HalfvecCosineSimilarity) (int dim, half * ax, half * bx);
 float		(*HalfvecL1Distance) (int dim, half * ax, half * bx);
 
+void 		(*Float4ToHalfVector) (Vector * vec, HalfVector * result);
+
 static float
 HalfvecL2SquaredDistanceDefault(int dim, half * ax, half * bx)
 {
@@ -237,6 +240,13 @@ HalfvecL1DistanceF16c(int dim, half * ax, half * bx)
 }
 #endif
 
+static void
+Float4ToHalfVectorDefault(Vector * vec, HalfVector * result) {
+	for (int i = 0; i < vec->dim; i++)
+		result->x[i] = Float4ToHalf(vec->x[i]);
+}
+
+
 #ifdef HALFVEC_DISPATCH
 #define CPU_FEATURE_FMA     (1 << 12)
 #define CPU_FEATURE_OSXSAVE (1 << 27)
@@ -284,6 +294,7 @@ HalfvecInit(void)
 	HalfvecInnerProduct = HalfvecInnerProductDefault;
 	HalfvecCosineSimilarity = HalfvecCosineSimilarityDefault;
 	HalfvecL1Distance = HalfvecL1DistanceDefault;
+	Float4ToHalfVector = Float4ToHalfVectorDefault;
 
 #ifdef HALFVEC_DISPATCH
 	if (SupportsCpuFeature(CPU_FEATURE_AVX | CPU_FEATURE_F16C | CPU_FEATURE_FMA))
@@ -294,5 +305,16 @@ HalfvecInit(void)
 		/* Does not require FMA, but keep logic simple */
 		HalfvecL1Distance = HalfvecL1DistanceF16c;
 	}
+
+#if defined(USE_AVX512) && defined(HAVE_AVX512FP16)
+	if (SupportsAvx512Fp16())
+	{
+		HalfvecL2SquaredDistance = HalfvecL2SquaredDistanceAvx512;
+		HalfvecInnerProduct = HalfvecInnerProductAvx512;
+		HalfvecCosineSimilarity = HalfvecCosineSimilarityAvx512;
+		HalfvecL1Distance = HalfvecL1DistanceAvx512;
+		Float4ToHalfVector = Float4ToHalfVectorAvx512;
+	}
+#endif
 #endif
 }
diff --git a/src/halfutils.h b/src/halfutils.h
index c684f72d7..1a33d1748 100644
--- a/src/halfutils.h
+++ b/src/halfutils.h
@@ -5,6 +5,7 @@
 
 #include "common/shortest_dec.h"
 #include "halfvec.h"
+#include "vector.h"
 
 #ifdef F16C_SUPPORT
 #include <immintrin.h>
@@ -15,6 +16,8 @@ extern float (*HalfvecInnerProduct) (int dim, half * ax, half * bx);
 extern double (*HalfvecCosineSimilarity) (int dim, half * ax, half * bx);
 extern float (*HalfvecL1Distance) (int dim, half * ax, half * bx);
 
+extern void (*Float4ToHalfVector) (Vector * vec, HalfVector * result);
+
 void		HalfvecInit(void);
 
 /*
diff --git a/src/halfutils_avx512.c b/src/halfutils_avx512.c
new file mode 100644
index 000000000..27d907781
--- /dev/null
+++ b/src/halfutils_avx512.c
@@ -0,0 +1,449 @@
+#ifdef USE_AVX512
+#include "halfutils_avx512.h"
+
+#ifdef HAVE_AVX512FP16
+#include "common/shortest_dec.h"
+
+#include <immintrin.h>
+#include <math.h>
+
+#if defined(USE__GET_CPUID)
+#include <cpuid.h>
+#else
+#include <intrin.h>
+#endif
+
+#ifdef _MSC_VER
+#define TARGET_AVX512FP16
+#else
+#define TARGET_AVX512FP16 __attribute__((target("avx512fp16,avx512f,avx512dq,avx512vl,avx512bw")))
+#endif
+
+TARGET_AVX512FP16 static inline bool
+HasInfinity(__m512h val) {
+	/* Test for positive and negative infinity */
+	__mmask32 mask = _mm512_fpclass_ph_mask(val, 0x08 + 0x10);
+	return mask != 0;
+}
+
+TARGET_AVX512FP16 static inline __m512
+ConvertToFp32Sum(__m512h val) {
+	__m256h val_lower = _mm256_castsi256_ph(_mm512_extracti32x8_epi32(_mm512_castph_si512(val), 0));
+	__m256h val_upper = _mm256_castsi256_ph(_mm512_extracti32x8_epi32(_mm512_castph_si512(val), 1));
+	return _mm512_add_ps(_mm512_cvtxph_ps(val_lower), _mm512_cvtxph_ps(val_upper));
+}
+
+TARGET_AVX512FP16 float
+HalfvecL2SquaredDistanceAvx512(int dim, half * ax, half * bx)
+{
+	float		distance;
+	int			i;
+	unsigned long mask;
+
+	/* For FP16 computation */
+	__m512h		axi_512h;
+	__m512h		bxi_512h;
+	__m512h		diff_512h;
+	__m512h		dist_512h = _mm512_setzero_ph();
+	__m512h		dist_512h_temp;
+
+	/* For FP32 computation */
+	__m256h		axi_256h;
+	__m256h		bxi_256h;
+	__m512		axi_512;
+	__m512		bxi_512;
+	__m512		diff_512;
+	__m512		dist_512;
+
+	/* FP16 computation */
+	for (i = 0; i < dim; i += 32)
+	{
+		if (dim - i < 32)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, ax + i));
+			bxi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_512h = _mm512_loadu_ph(ax + i);
+			bxi_512h = _mm512_loadu_ph(bx + i);
+		}
+		diff_512h = _mm512_sub_ph(axi_512h, bxi_512h);
+		dist_512h_temp = _mm512_fmadd_ph(diff_512h, diff_512h, dist_512h);
+
+		/* if overflow, continue with FP32 */
+		if (HasInfinity(dist_512h_temp))
+			break;
+		else
+			dist_512h = dist_512h_temp;
+	}
+	dist_512 = ConvertToFp32Sum(dist_512h);
+
+	/* FP32 computation */
+	for (; i < dim; i += 16)
+	{
+		if (dim - i < 16)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, ax + i));
+			bxi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_256h = _mm256_loadu_ph(ax + i);
+			bxi_256h = _mm256_loadu_ph(bx + i);
+		}
+		axi_512 = _mm512_cvtxph_ps(axi_256h);
+		bxi_512 = _mm512_cvtxph_ps(bxi_256h);
+		diff_512 = _mm512_sub_ps(axi_512, bxi_512);
+		dist_512 = _mm512_fmadd_ps(diff_512, diff_512, dist_512);
+	}
+
+	distance = _mm512_reduce_add_ps(dist_512);
+	return distance;
+}
+
+TARGET_AVX512FP16 float
+HalfvecInnerProductAvx512(int dim, half * ax, half * bx)
+{
+	float		distance;
+	int			i;
+	unsigned int mask;
+
+	/* For FP16 computation */
+	__m512h		axi_512h;
+	__m512h		bxi_512h;
+	__m512h		dist_512h = _mm512_setzero_ph();
+	__m512h		dist_512h_temp;
+
+	/* For FP32 computation */
+	__m256h		axi_256h;
+	__m256h		bxi_256h;
+	__m512		axi_512;
+	__m512		bxi_512;
+	__m512		dist_512;
+
+	/* FP16 computation */
+	for (i = 0; i < dim; i += 32)
+	{
+		if (dim - i < 32)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, ax + i));
+			bxi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_512h = _mm512_loadu_ph(ax + i);
+			bxi_512h = _mm512_loadu_ph(bx + i);
+		}
+		dist_512h_temp = _mm512_fmadd_ph(axi_512h, bxi_512h, dist_512h);
+
+		/* if overflow, continue with FP32 */
+		if (HasInfinity(dist_512h_temp))
+			break;
+		else
+			dist_512h = dist_512h_temp;
+	}
+	dist_512 = ConvertToFp32Sum(dist_512h);
+
+	/* FP32 computation */
+	for (; i < dim; i += 16)
+	{
+		if (dim - i < 16)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, ax + i));
+			bxi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_256h = _mm256_loadu_ph(ax + i);
+			bxi_256h = _mm256_loadu_ph(bx + i);
+		}
+		axi_512 = _mm512_cvtxph_ps(axi_256h);
+		bxi_512 = _mm512_cvtxph_ps(bxi_256h);
+		dist_512 = _mm512_fmadd_ps(axi_512, bxi_512, dist_512);
+	}
+
+	distance = _mm512_reduce_add_ps(dist_512);
+	return distance;
+}
+
+TARGET_AVX512FP16 double
+HalfvecCosineSimilarityAvx512(int dim, half * ax, half * bx)
+{
+	float		similarity;
+	float		norma;
+	float		normb;
+	int			i;
+	unsigned int mask;
+
+	/* For FP16 computation */
+	__m512h		axi_512h;
+	__m512h		bxi_512h;
+	__m512h		sim_512h = _mm512_setzero_ph();
+	__m512h		na_512h = _mm512_setzero_ph();
+	__m512h		nb_512h = _mm512_setzero_ph();
+	__m512h		sim_512h_temp;
+	__m512h		na_512h_temp;
+	__m512h		nb_512h_temp;
+
+	/* For FP32 computation */
+	__m256h		axi_256h;
+	__m256h 	bxi_256h;
+	__m512		axi_512;
+	__m512		bxi_512;
+	__m512		sim_512;
+	__m512		na_512;
+	__m512		nb_512;
+
+	/* FP16 computation */
+	for (i = 0; i < dim; i += 32)
+	{
+		if (dim - i < 32) {
+			mask = (1 << (dim - i)) - 1;
+			axi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, ax + i));
+			bxi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, bx + i));
+		}
+		else {
+			axi_512h = _mm512_loadu_ph(ax + i);
+			bxi_512h = _mm512_loadu_ph(bx + i);
+		}
+		sim_512h_temp = _mm512_fmadd_ph(axi_512h, bxi_512h, sim_512h);
+		na_512h_temp = _mm512_fmadd_ph(axi_512h, axi_512h, na_512h);
+		nb_512h_temp = _mm512_fmadd_ph(bxi_512h, bxi_512h, nb_512h);
+
+		/* if overflow, continue with FP32 */
+		if (HasInfinity(sim_512h_temp) ||
+			HasInfinity(na_512h_temp) ||
+			HasInfinity(nb_512h_temp))
+			break;
+		else
+		{
+			sim_512h = sim_512h_temp;
+			na_512h = na_512h_temp;
+			nb_512h = nb_512h_temp;
+		}
+	}
+	sim_512 = ConvertToFp32Sum(sim_512h);
+	na_512 = ConvertToFp32Sum(na_512h);
+	nb_512 = ConvertToFp32Sum(nb_512h);
+
+	/* FP32 computation */
+	for (; i < dim; i += 16)
+	{
+		if (dim - i < 16)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, ax + i));
+			bxi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_256h = _mm256_loadu_ph(ax + i);
+			bxi_256h = _mm256_loadu_ph(bx + i);
+		}
+		axi_512 = _mm512_cvtxph_ps(axi_256h);
+		bxi_512 = _mm512_cvtxph_ps(bxi_256h);
+		sim_512 = _mm512_fmadd_ps(axi_512, bxi_512, sim_512);
+		na_512 = _mm512_fmadd_ps(axi_512, axi_512, na_512);
+		nb_512 = _mm512_fmadd_ps(bxi_512, bxi_512, nb_512);
+	}
+
+	similarity = _mm512_reduce_add_ps(sim_512);
+	norma = _mm512_reduce_add_ps(na_512);
+	normb = _mm512_reduce_add_ps(nb_512);
+
+	/* Use sqrt(a * b) over sqrt(a) * sqrt(b) */
+	return (double) similarity / sqrt((double) norma * (double) normb);
+}
+
+TARGET_AVX512FP16 float
+HalfvecL1DistanceAvx512(int dim, half * ax, half * bx)
+{
+	float		distance;
+	int			i;
+	unsigned long mask;
+
+	/* For FP16 computation */
+	__m512h		axi_512h;
+	__m512h		bxi_512h;
+	__m512h		dist_512h = _mm512_setzero_ph();
+	__m512h		dist_512h_temp;
+
+	/* For FP32 computation */
+	__m256h		axi_256h;
+	__m256h		bxi_256h;
+	__m512		axi_512;
+	__m512		bxi_512;
+	__m512		dist_512;
+
+	/* FP16 computation */
+	for (i = 0; i < dim; i += 32)
+	{
+		if (dim - i < 32)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, ax + i));
+			bxi_512h = _mm512_castsi512_ph(_mm512_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_512h = _mm512_loadu_ph(ax + i);
+			bxi_512h = _mm512_loadu_ph(bx + i);
+		}
+		dist_512h_temp = _mm512_add_ph(dist_512h, _mm512_abs_ph(_mm512_sub_ph(axi_512h, bxi_512h)));
+
+		/* if overflow, continue with FP32 */
+		if (HasInfinity(dist_512h_temp))
+			break;
+		else
+			dist_512h = dist_512h_temp;
+	}
+	dist_512 = ConvertToFp32Sum(dist_512h);
+
+	/* FP32 computation */
+	for (; i < dim; i += 16)
+	{
+		if (dim - i < 16)
+		{
+			mask = (1 << (dim - i)) - 1;
+			axi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, ax + i));
+			bxi_256h = _mm256_castsi256_ph(_mm256_maskz_loadu_epi16(mask, bx + i));
+		}
+		else
+		{
+			axi_256h = _mm256_loadu_ph(ax + i);
+			bxi_256h = _mm256_loadu_ph(bx + i);
+		}
+		axi_512 = _mm512_cvtxph_ps(axi_256h);
+		bxi_512 = _mm512_cvtxph_ps(bxi_256h);
+		dist_512 = _mm512_add_ps(dist_512, _mm512_abs_ps(_mm512_sub_ps(axi_512, bxi_512)));
+	}
+
+	distance = _mm512_reduce_add_ps(dist_512);
+
+	return distance;
+}
+
+TARGET_AVX512FP16 void
+Float4ToHalfVectorAvx512(Vector * vec, HalfVector * result)
+{
+	unsigned long mask;
+	__m512		vec_512;
+	__m256h		vec_256h;
+	__mmask16 	vec_512_inf;
+	__mmask16 	vec_256h_inf;
+
+	for (int i = 0; i < vec->dim; i += 16)
+	{
+		if (vec->dim - i < 16)
+		{
+			mask = (1 << (vec->dim - i)) - 1;
+			vec_512 = _mm512_maskz_loadu_ps(mask, vec->x + i);
+			vec_256h = _mm512_cvtxps_ph(vec_512);
+			_mm256_mask_storeu_epi16(result->x + i, mask, _mm256_castph_si256(vec_256h));
+		}
+		else
+		{
+			vec_512 = _mm512_loadu_ps(vec->x + i);
+			vec_256h = _mm512_cvtxps_ph(vec_512);
+			_mm256_storeu_ph(result->x + i, vec_256h);
+		}
+
+		/* Test for positive and negative infinity */
+		vec_512_inf = _mm512_fpclass_ps_mask(vec_512, 0x08 + 0x10);
+		vec_256h_inf = _mm256_fpclass_ph_mask(vec_256h, 0x08 + 0x10);
+		if (unlikely(vec_512_inf != vec_256h_inf))
+		{
+			float num;
+			char* buf;
+
+			__mmask16 diff = _kxor_mask16(vec_512_inf, vec_256h_inf);
+			/* Find first element in vector to overflow after conversion (first bit set) */
+			int count = 0;
+			while (diff % 2 == 0) {
+				diff >>= 1;
+				count++;
+			}
+			num = vec->x[i + count];
+
+			/* TODO Avoid duplicate code in Float4ToHalf */
+			buf = palloc(FLOAT_SHORTEST_DECIMAL_LEN);
+
+			float_to_shortest_decimal_buf(num, buf);
+
+			ereport(ERROR,
+					(errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
+					 errmsg("\"%s\" is out of range for type halfvec", buf)));
+		}
+	}
+}
+
+#define CPU_FEATURE_OSXSAVE     (1 << 27)
+#define CPU_FEATURE_AVX512F     (1 << 16)
+#define CPU_FEATURE_AVX512DQ    (1 << 17)
+#define CPU_FEATURE_AVX512_FP16 (1 << 23)
+#define CPU_FEATURE_AVX512BW    (1 << 30)
+#define CPU_FEATURE_AVX512VL    (1 << 31)
+
+#ifdef _MSC_VER
+#define TARGET_XSAVE
+#else
+#define TARGET_XSAVE __attribute__((target("xsave")))
+#endif
+
+TARGET_XSAVE static bool
+SupportsOsXsave()
+{
+	unsigned int exx[4] = {0, 0, 0, 0};
+
+#if defined(HAVE__GET_CPUID)
+	__get_cpuid(1, &exx[0], &exx[1], &exx[2], &exx[3]);
+#else
+	__cpuid(exx, 1);
+#endif
+
+	return (exx[2] & CPU_FEATURE_OSXSAVE) == CPU_FEATURE_OSXSAVE;
+}
+
+TARGET_XSAVE bool
+SupportsAvx512Fp16()
+{
+	unsigned int exx[4] = {0, 0, 0, 0};
+
+	/* AVX512 features required:
+	 *  AVX512F : sub/fma/add instructions
+	 *  AVX512DQ: _mm512_extracti32x8_epi32
+	 *  AVX512VL: _mm256_loadu_ph
+	 *  AVX512BW: masked loads
+	 */
+	unsigned int features = CPU_FEATURE_AVX512F |
+	                        CPU_FEATURE_AVX512DQ |
+	                        CPU_FEATURE_AVX512VL |
+	                        CPU_FEATURE_AVX512BW;
+
+	/* Check OS supports XSAVE */
+	if (!SupportsOsXsave())
+		return false;
+
+	/* Check XMM, YMM, and ZMM registers are enabled */
+	if ((_xgetbv(0) & 0xe6) != 0xe6)
+		return false;
+
+#if defined(HAVE__GET_CPUID)
+	__get_cpuid_count(7, 0, &exx[0], &exx[1], &exx[2], &exx[3]);
+#elif defined(HAVE__CPUID)
+	__cpuid(exx, 7, 0);
+#endif
+
+	if ((exx[1] & features) != features)
+		return false;
+
+	return (exx[3] & CPU_FEATURE_AVX512_FP16) == CPU_FEATURE_AVX512_FP16;
+}
+
+#endif
+#endif
diff --git a/src/halfutils_avx512.h b/src/halfutils_avx512.h
new file mode 100644
index 000000000..9f94f11dc
--- /dev/null
+++ b/src/halfutils_avx512.h
@@ -0,0 +1,25 @@
+#ifndef HALFUTILS_AVX512_H
+#define HALFUTILS_AVX512_H
+
+#ifdef USE_AVX512
+#include "postgres.h"
+#include "halfvec.h"
+#include "vector.h"
+
+#if (defined(__GNUC__) && (__GNUC__ >= 12)) || \
+	(defined(__clang__) && (__clang_major__ >= 16)) || \
+	(defined __AVX512FP16__)
+#define HAVE_AVX512FP16
+#endif
+
+#ifdef HAVE_AVX512FP16
+extern float HalfvecL2SquaredDistanceAvx512(int dim, half * ax, half * bx);
+extern float HalfvecInnerProductAvx512(int dim, half * ax, half * bx);
+extern double HalfvecCosineSimilarityAvx512(int dim, half * ax, half * bx);
+extern float HalfvecL1DistanceAvx512(int dim, half * ax, half * bx);
+extern void Float4ToHalfVectorAvx512(Vector * vec, HalfVector * result);
+
+extern bool SupportsAvx512Fp16(void);
+#endif
+#endif
+#endif
diff --git a/src/halfvec.c b/src/halfvec.c
index aad320b1c..f9a42a559 100644
--- a/src/halfvec.c
+++ b/src/halfvec.c
@@ -533,8 +533,7 @@ vector_to_halfvec(PG_FUNCTION_ARGS)
 
 	result = InitHalfVector(vec->dim);
 
-	for (int i = 0; i < vec->dim; i++)
-		result->x[i] = Float4ToHalf(vec->x[i]);
+	Float4ToHalfVector(vec, result);
 
 	PG_RETURN_POINTER(result);
 }
@@ -898,8 +897,21 @@ halfvec_binary_quantize(PG_FUNCTION_ARGS)
 	half	   *ax = a->x;
 	VarBit	   *result = InitBitVector(a->dim);
 	unsigned char *rx = VARBITS(result);
+	int			i = 0;
+	int			count = (a->dim / 8) * 8;
 
-	for (int i = 0; i < a->dim; i++)
+	/* Auto-vectorized on aarch64 */
+	for (; i < count; i += 8)
+	{
+		unsigned char result_byte = 0;
+
+		for (int j = 0; j < 8; j++)
+			result_byte |= (HalfToFloat4(ax[i + j]) > 0) << (7 - j);
+
+		rx[i / 8] = result_byte;
+	}
+
+	for (; i < a->dim; i++)
 		rx[i / 8] |= (HalfToFloat4(ax[i]) > 0) << (7 - (i % 8));
 
 	PG_RETURN_VARBIT_P(result);
diff --git a/src/hnsw.c b/src/hnsw.c
index 5bfc6193e..1d56ef6ea 100644
--- a/src/hnsw.c
+++ b/src/hnsw.c
@@ -52,12 +52,20 @@ HnswInitLockTranche(void)
 								  sizeof(int) * 1,
 								  &found);
 	if (!found)
+	{
+#if PG_VERSION_NUM >= 190000
+		tranche_ids[0] = LWLockNewTrancheId("HnswBuild");
+#else
 		tranche_ids[0] = LWLockNewTrancheId();
+#endif
+	}
 	hnsw_lock_tranche_id = tranche_ids[0];
 	LWLockRelease(AddinShmemInitLock);
 
+#if PG_VERSION_NUM < 190000
 	/* Per-backend registration of the tranche ID */
 	LWLockRegisterTranche(hnsw_lock_tranche_id, "HnswBuild");
+#endif
 }
 
 /*
@@ -130,7 +138,7 @@ hnswcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
 	Relation	index;
 
 	/* Never use index without order */
-	if (path->indexorderbys == NULL)
+	if (path->indexorderbys == NIL)
 	{
 		*indexStartupCost = get_float8_infinity();
 		*indexTotalCost = get_float8_infinity();
@@ -259,6 +267,11 @@ hnswhandler(PG_FUNCTION_ARGS)
 	amroutine->amoptsprocnum = 0;
 	amroutine->amcanorder = false;
 	amroutine->amcanorderbyop = true;
+#if PG_VERSION_NUM >= 180000
+	amroutine->amcanhash = false;
+	amroutine->amconsistentequality = false;
+	amroutine->amconsistentordering = false;
+#endif
 	amroutine->amcanbackward = false;	/* can change direction mid-scan */
 	amroutine->amcanunique = false;
 	amroutine->amcanmulticol = false;
@@ -291,6 +304,9 @@ hnswhandler(PG_FUNCTION_ARGS)
 	amroutine->amvacuumcleanup = hnswvacuumcleanup;
 	amroutine->amcanreturn = NULL;
 	amroutine->amcostestimate = hnswcostestimate;
+#if PG_VERSION_NUM >= 180000
+	amroutine->amgettreeheight = NULL;
+#endif
 	amroutine->amoptions = hnswoptions;
 	amroutine->amproperty = NULL;	/* TODO AMPROP_DISTANCE_ORDERABLE */
 	amroutine->ambuildphasename = hnswbuildphasename;
@@ -311,5 +327,10 @@ hnswhandler(PG_FUNCTION_ARGS)
 	amroutine->aminitparallelscan = NULL;
 	amroutine->amparallelrescan = NULL;
 
+#if PG_VERSION_NUM >= 180000
+	amroutine->amtranslatestrategy = NULL;
+	amroutine->amtranslatecmptype = NULL;
+#endif
+
 	PG_RETURN_POINTER(amroutine);
 }
diff --git a/src/hnswbuild.c b/src/hnswbuild.c
index b667478b6..03f0ef4a4 100644
--- a/src/hnswbuild.c
+++ b/src/hnswbuild.c
@@ -54,6 +54,10 @@
 #include "utils/datum.h"
 #include "utils/memutils.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 #if PG_VERSION_NUM >= 140000
 #include "utils/backend_progress.h"
 #else
@@ -398,7 +402,7 @@ UpdateNeighborsInMemory(char *base, HnswSupport * support, HnswElement e, int m)
  * Update graph in memory
  */
 static void
-UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, HnswBuildState * buildstate)
+UpdateGraphInMemory(HnswSupport * support, HnswElement element, int m, HnswElement entryPoint, HnswBuildState * buildstate)
 {
 	HnswGraph  *graph = buildstate->graph;
 	char	   *base = buildstate->hnswarea;
@@ -460,7 +464,7 @@ InsertTupleInMemory(HnswBuildState * buildstate, HnswElement element)
 	HnswFindElementNeighbors(base, element, entryPoint, NULL, support, m, efConstruction, false);
 
 	/* Update graph in memory */
-	UpdateGraphInMemory(support, element, m, efConstruction, entryPoint, buildstate);
+	UpdateGraphInMemory(support, element, m, entryPoint, buildstate);
 
 	/* Release entry lock */
 	LWLockRelease(entryLock);
@@ -1054,7 +1058,7 @@ ComputeParallelWorkers(Relation heap, Relation index)
  * Build graph
  */
 static void
-BuildGraph(HnswBuildState * buildstate, ForkNumber forkNum)
+BuildGraph(HnswBuildState * buildstate)
 {
 	int			parallel_workers = 0;
 
@@ -1102,7 +1106,7 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
 
 	InitBuildState(buildstate, heap, index, indexInfo, forkNum);
 
-	BuildGraph(buildstate, forkNum);
+	BuildGraph(buildstate);
 
 	if (RelationNeedsWAL(index) || forkNum == INIT_FORKNUM)
 		log_newpage_range(index, forkNum, 0, RelationGetNumberOfBlocksInFork(index, forkNum), true);
diff --git a/src/hnswinsert.c b/src/hnswinsert.c
index a5fac4eda..a4d288506 100644
--- a/src/hnswinsert.c
+++ b/src/hnswinsert.c
@@ -9,6 +9,10 @@
 #include "utils/datum.h"
 #include "utils/memutils.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 /*
  * Get the insert page
  */
@@ -660,7 +664,7 @@ FindDuplicateOnDisk(Relation index, HnswElement element, bool building)
  * Update graph on disk
  */
 static void
-UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, int m, int efConstruction, HnswElement entryPoint, bool building)
+UpdateGraphOnDisk(Relation index, HnswSupport * support, HnswElement element, int m, HnswElement entryPoint, bool building)
 {
 	BlockNumber newInsertPage = InvalidBlockNumber;
 
@@ -728,7 +732,7 @@ HnswInsertTupleOnDisk(Relation index, HnswSupport * support, Datum value, ItemPo
 	HnswFindElementNeighbors(base, element, entryPoint, index, support, m, efConstruction, false);
 
 	/* Update graph on disk */
-	UpdateGraphOnDisk(index, support, element, m, efConstruction, entryPoint, building);
+	UpdateGraphOnDisk(index, support, element, m, entryPoint, building);
 
 	/* Release lock */
 	UnlockPage(index, HNSW_UPDATE_LOCK, lockmode);
diff --git a/src/hnswscan.c b/src/hnswscan.c
index 955998a52..5c526f4a6 100644
--- a/src/hnswscan.c
+++ b/src/hnswscan.c
@@ -193,6 +193,10 @@ hnswgettuple(IndexScanDesc scan, ScanDirection dir)
 
 		/* Count index scan for stats */
 		pgstat_count_index_scan(scan->indexRelation);
+#if PG_VERSION_NUM >= 180000
+		if (scan->instrument)
+			scan->instrument->nsearches++;
+#endif
 
 		/* Safety check */
 		if (scan->orderByData == NULL)
diff --git a/src/hnswutils.c b/src/hnswutils.c
index c52d2c78a..8e2a42c1d 100644
--- a/src/hnswutils.c
+++ b/src/hnswutils.c
@@ -15,6 +15,10 @@
 #include "utils/memdebug.h"
 #include "utils/rel.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 #if PG_VERSION_NUM < 170000
 static inline uint64
 murmurhash64(uint64 data)
diff --git a/src/hnswvacuum.c b/src/hnswvacuum.c
index 2f7b2f372..3a8ee26c0 100644
--- a/src/hnswvacuum.c
+++ b/src/hnswvacuum.c
@@ -9,6 +9,10 @@
 #include "storage/lmgr.h"
 #include "utils/memutils.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 #if PG_VERSION_NUM >= 180000
 #define vacuum_delay_point() vacuum_delay_point(false)
 #endif
diff --git a/src/ivfbuild.c b/src/ivfbuild.c
index 944f07b55..7166b7da3 100644
--- a/src/ivfbuild.c
+++ b/src/ivfbuild.c
@@ -20,6 +20,10 @@
 #include "utils/memutils.h"
 #include "vector.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 #if PG_VERSION_NUM >= 140000
 #include "utils/backend_progress.h"
 #else
@@ -138,7 +142,7 @@ SampleRows(IvfflatBuildState * buildstate)
  * Add tuple to sort
  */
 static void
-AddTupleToSort(Relation index, ItemPointer tid, Datum *values, IvfflatBuildState * buildstate)
+AddTupleToSort(ItemPointer tid, Datum *values, IvfflatBuildState * buildstate)
 {
 	double		distance;
 	double		minDistance = DBL_MAX;
@@ -215,7 +219,7 @@ BuildCallback(Relation index, ItemPointer tid, Datum *values,
 	oldCtx = MemoryContextSwitchTo(buildstate->tmpCtx);
 
 	/* Add tuple to sort */
-	AddTupleToSort(index, tid, values, buildstate);
+	AddTupleToSort(tid, values, buildstate);
 
 	/* Reset memory context */
 	MemoryContextSwitchTo(oldCtx);
@@ -470,8 +474,8 @@ CreateMetaPage(Relation index, int dimensions, int lists, ForkNumber forkNum)
  * Create list pages
  */
 static void
-CreateListPages(Relation index, VectorArray centers, int dimensions,
-				int lists, ForkNumber forkNum, ListInfo * *listInfo)
+CreateListPages(Relation index, VectorArray centers, int lists,
+				ForkNumber forkNum, ListInfo * *listInfo)
 {
 	Buffer		buf;
 	Page		page;
@@ -1004,7 +1008,7 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
 
 	/* Create pages */
 	CreateMetaPage(index, buildstate->dimensions, buildstate->lists, forkNum);
-	CreateListPages(index, buildstate->centers, buildstate->dimensions, buildstate->lists, forkNum, &buildstate->listInfo);
+	CreateListPages(index, buildstate->centers, buildstate->lists, forkNum, &buildstate->listInfo);
 	CreateEntryPages(buildstate, forkNum);
 
 	/* Write WAL for initialization fork since GenericXLog functions do not */
@@ -1023,6 +1027,10 @@ ivfflatbuild(Relation heap, Relation index, IndexInfo *indexInfo)
 	IndexBuildResult *result;
 	IvfflatBuildState buildstate;
 
+#ifdef IVFFLAT_BENCH
+	SeedRandom(42);
+#endif
+
 	BuildIndex(heap, index, indexInfo, &buildstate, MAIN_FORKNUM);
 
 	result = (IndexBuildResult *) palloc(sizeof(IndexBuildResult));
diff --git a/src/ivfflat.c b/src/ivfflat.c
index 9e8370ffe..31c2f7d5e 100644
--- a/src/ivfflat.c
+++ b/src/ivfflat.c
@@ -92,7 +92,7 @@ ivfflatcostestimate(PlannerInfo *root, IndexPath *path, double loop_count,
 	Relation	index;
 
 	/* Never use index without order */
-	if (path->indexorderbys == NULL)
+	if (path->indexorderbys == NIL)
 	{
 		*indexStartupCost = get_float8_infinity();
 		*indexTotalCost = get_float8_infinity();
@@ -186,6 +186,11 @@ ivfflathandler(PG_FUNCTION_ARGS)
 	amroutine->amoptsprocnum = 0;
 	amroutine->amcanorder = false;
 	amroutine->amcanorderbyop = true;
+#if PG_VERSION_NUM >= 180000
+	amroutine->amcanhash = false;
+	amroutine->amconsistentequality = false;
+	amroutine->amconsistentordering = false;
+#endif
 	amroutine->amcanbackward = false;	/* can change direction mid-scan */
 	amroutine->amcanunique = false;
 	amroutine->amcanmulticol = false;
@@ -218,6 +223,9 @@ ivfflathandler(PG_FUNCTION_ARGS)
 	amroutine->amvacuumcleanup = ivfflatvacuumcleanup;
 	amroutine->amcanreturn = NULL;	/* tuple not included in heapsort */
 	amroutine->amcostestimate = ivfflatcostestimate;
+#if PG_VERSION_NUM >= 180000
+	amroutine->amgettreeheight = NULL;
+#endif
 	amroutine->amoptions = ivfflatoptions;
 	amroutine->amproperty = NULL;	/* TODO AMPROP_DISTANCE_ORDERABLE */
 	amroutine->ambuildphasename = ivfflatbuildphasename;
@@ -238,5 +246,10 @@ ivfflathandler(PG_FUNCTION_ARGS)
 	amroutine->aminitparallelscan = NULL;
 	amroutine->amparallelrescan = NULL;
 
+#if PG_VERSION_NUM >= 180000
+	amroutine->amtranslatestrategy = NULL;
+	amroutine->amtranslatecmptype = NULL;
+#endif
+
 	PG_RETURN_POINTER(amroutine);
 }
diff --git a/src/ivfflat.h b/src/ivfflat.h
index c296b6677..c4e64e1dc 100644
--- a/src/ivfflat.h
+++ b/src/ivfflat.h
@@ -13,6 +13,10 @@
 #include "utils/tuplesort.h"
 #include "vector.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 #if PG_VERSION_NUM >= 150000
 #include "common/pg_prng.h"
 #endif
@@ -73,9 +77,11 @@
 #if PG_VERSION_NUM >= 150000
 #define RandomDouble() pg_prng_double(&pg_global_prng_state)
 #define RandomInt() pg_prng_uint32(&pg_global_prng_state)
+#define SeedRandom(seed) pg_prng_seed(&pg_global_prng_state, seed)
 #else
 #define RandomDouble() (((double) random()) / MAX_RANDOM_VALUE)
 #define RandomInt() random()
+#define SeedRandom(seed) srandom(seed)
 #endif
 
 /* Variables */
diff --git a/src/ivfinsert.c b/src/ivfinsert.c
index 014c9be82..ebc61e49d 100644
--- a/src/ivfinsert.c
+++ b/src/ivfinsert.c
@@ -65,7 +65,7 @@ FindInsertPage(Relation index, Datum *values, BlockNumber *insertPage, ListInfo
  * Insert a tuple into the index
  */
 static void
-InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid, Relation heapRel)
+InsertTuple(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid)
 {
 	const		IvfflatTypeInfo *typeInfo = IvfflatGetTypeInfo(index);
 	IndexTuple	itup;
@@ -204,7 +204,7 @@ ivfflatinsert(Relation index, Datum *values, bool *isnull, ItemPointer heap_tid,
 	oldCtx = MemoryContextSwitchTo(insertCtx);
 
 	/* Insert tuple */
-	InsertTuple(index, values, isnull, heap_tid, heap);
+	InsertTuple(index, values, isnull, heap_tid);
 
 	/* Delete memory context */
 	MemoryContextSwitchTo(oldCtx);
diff --git a/src/ivfkmeans.c b/src/ivfkmeans.c
index 4b6d14f1a..9c79e64ad 100644
--- a/src/ivfkmeans.c
+++ b/src/ivfkmeans.c
@@ -13,6 +13,10 @@
 #include "utils/memutils.h"
 #include "vector.h"
 
+#if PG_VERSION_NUM >= 160000
+#include "varatt.h"
+#endif
+
 /*
  * Initialize with kmeans++
  *
diff --git a/src/ivfscan.c b/src/ivfscan.c
index 6cc5d2efd..b42ada15d 100644
--- a/src/ivfscan.c
+++ b/src/ivfscan.c
@@ -355,6 +355,10 @@ ivfflatgettuple(IndexScanDesc scan, ScanDirection dir)
 
 		/* Count index scan for stats */
 		pgstat_count_index_scan(scan->indexRelation);
+#if PG_VERSION_NUM >= 180000
+		if (scan->instrument)
+			scan->instrument->nsearches++;
+#endif
 
 		/* Safety check */
 		if (scan->orderByData == NULL)
diff --git a/src/ivfutils.c b/src/ivfutils.c
index da241ee0a..9596391be 100644
--- a/src/ivfutils.c
+++ b/src/ivfutils.c
@@ -259,8 +259,8 @@ VectorUpdateCenter(Pointer v, int dimensions, float *x)
 	SET_VARSIZE(vec, VECTOR_SIZE(dimensions));
 	vec->dim = dimensions;
 
-	for (int k = 0; k < dimensions; k++)
-		vec->x[k] = x[k];
+	for (int i = 0; i < dimensions; i++)
+		vec->x[i] = x[i];
 }
 
 static void
@@ -271,8 +271,8 @@ HalfvecUpdateCenter(Pointer v, int dimensions, float *x)
 	SET_VARSIZE(vec, HALFVEC_SIZE(dimensions));
 	vec->dim = dimensions;
 
-	for (int k = 0; k < dimensions; k++)
-		vec->x[k] = Float4ToHalfUnchecked(x[k]);
+	for (int i = 0; i < dimensions; i++)
+		vec->x[i] = Float4ToHalfUnchecked(x[i]);
 }
 
 static void
@@ -284,29 +284,33 @@ BitUpdateCenter(Pointer v, int dimensions, float *x)
 	SET_VARSIZE(vec, VARBITTOTALLEN(dimensions));
 	VARBITLEN(vec) = dimensions;
 
-	for (uint32 k = 0; k < VARBITBYTES(vec); k++)
-		nx[k] = 0;
+	for (uint32 i = 0; i < VARBITBYTES(vec); i++)
+		nx[i] = 0;
 
-	for (int k = 0; k < dimensions; k++)
-		nx[k / 8] |= (x[k] > 0.5 ? 1 : 0) << (7 - (k % 8));
+	for (int i = 0; i < dimensions; i++)
+		nx[i / 8] |= (x[i] > 0.5 ? 1 : 0) << (7 - (i % 8));
 }
 
 static void
 VectorSumCenter(Pointer v, float *x)
 {
 	Vector	   *vec = (Vector *) v;
+	int			dim = vec->dim;
 
-	for (int k = 0; k < vec->dim; k++)
-		x[k] += vec->x[k];
+	/* Auto-vectorized */
+	for (int i = 0; i < dim; i++)
+		x[i] += vec->x[i];
 }
 
 static void
 HalfvecSumCenter(Pointer v, float *x)
 {
 	HalfVector *vec = (HalfVector *) v;
+	int			dim = vec->dim;
 
-	for (int k = 0; k < vec->dim; k++)
-		x[k] += HalfToFloat4(vec->x[k]);
+	/* Auto-vectorized on aarch64 */
+	for (int i = 0; i < dim; i++)
+		x[i] += HalfToFloat4(vec->x[i]);
 }
 
 static void
@@ -314,8 +318,8 @@ BitSumCenter(Pointer v, float *x)
 {
 	VarBit	   *vec = (VarBit *) v;
 
-	for (int k = 0; k < VARBITLEN(vec); k++)
-		x[k] += (float) (((VARBITS(vec)[k / 8]) >> (7 - (k % 8))) & 0x01);
+	for (int i = 0; i < VARBITLEN(vec); i++)
+		x[i] += (float) (((VARBITS(vec)[i / 8]) >> (7 - (i % 8))) & 0x01);
 }
 
 /*
diff --git a/src/vector.c b/src/vector.c
index a5b2aac36..03f70d601 100644
--- a/src/vector.c
+++ b/src/vector.c
@@ -35,7 +35,11 @@
 #define VECTOR_TARGET_CLONES
 #endif
 
+#if PG_VERSION_NUM >= 180000
+PG_MODULE_MAGIC_EXT(.name = "vector",.version = "0.8.1");
+#else
 PG_MODULE_MAGIC;
+#endif
 
 /*
  * Initialize index options and variables
@@ -920,11 +924,13 @@ vector_concat(PG_FUNCTION_ARGS)
 	CheckDim(dim);
 	result = InitVector(dim);
 
-	for (int i = 0; i < a->dim; i++)
+	/* Auto-vectorized */
+	for (int i = 0, imax = a->dim; i < imax; i++)
 		result->x[i] = a->x[i];
 
-	for (int i = 0; i < b->dim; i++)
-		result->x[i + a->dim] = b->x[i];
+	/* Auto-vectorized */
+	for (int i = 0, imax = b->dim, start = a->dim; i < imax; i++)
+		result->x[i + start] = b->x[i];
 
 	PG_RETURN_POINTER(result);
 }
@@ -940,8 +946,21 @@ binary_quantize(PG_FUNCTION_ARGS)
 	float	   *ax = a->x;
 	VarBit	   *result = InitBitVector(a->dim);
 	unsigned char *rx = VARBITS(result);
+	int			i = 0;
+	int			count = (a->dim / 8) * 8;
 
-	for (int i = 0; i < a->dim; i++)
+	/* Auto-vectorized */
+	for (; i < count; i += 8)
+	{
+		unsigned char result_byte = 0;
+
+		for (int j = 0; j < 8; j++)
+			result_byte |= (ax[i + j] > 0) << (7 - j);
+
+		rx[i / 8] = result_byte;
+	}
+
+	for (; i < a->dim; i++)
 		rx[i / 8] |= (ax[i] > 0) << (7 - (i % 8));
 
 	PG_RETURN_VARBIT_P(result);
diff --git a/test/expected/halfvec.out b/test/expected/halfvec.out
index a3ce8931f..c3856c512 100644
--- a/test/expected/halfvec.out
+++ b/test/expected/halfvec.out
@@ -378,6 +378,24 @@ SELECT '[0,0]'::halfvec <-> '[3,4]';
         5
 (1 row)
 
+SELECT l2_distance('[501]'::halfvec, '[1]');
+ l2_distance 
+-------------
+         500
+(1 row)
+
+SELECT l2_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ l2_distance 
+-------------
+           0
+(1 row)
+
+SELECT l2_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ l2_distance 
+-------------
+           6
+(1 row)
+
 SELECT inner_product('[1,2]'::halfvec, '[3,4]');
  inner_product 
 ---------------
@@ -404,6 +422,24 @@ SELECT '[1,2]'::halfvec <#> '[3,4]';
       -11
 (1 row)
 
+SELECT inner_product('[50000,50000]'::halfvec, '[1,1]');
+ inner_product 
+---------------
+         99968
+(1 row)
+
+SELECT inner_product('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ inner_product 
+---------------
+            36
+(1 row)
+
+SELECT inner_product('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ inner_product 
+---------------
+            72
+(1 row)
+
 SELECT cosine_distance('[1,2]'::halfvec, '[2,4]');
  cosine_distance 
 -----------------
@@ -466,6 +502,24 @@ SELECT '[1,2]'::halfvec <=> '[2,4]';
         0
 (1 row)
 
+SELECT cosine_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ cosine_distance 
+-----------------
+               0
+(1 row)
+
+SELECT cosine_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ cosine_distance 
+-----------------
+               0
+(1 row)
+
+SELECT cosine_distance('[1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0]'::halfvec, '[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]');
+ cosine_distance 
+-----------------
+               1
+(1 row)
+
 SELECT l1_distance('[0,0]'::halfvec, '[3,4]');
  l1_distance 
 -------------
@@ -498,6 +552,18 @@ SELECT '[0,0]'::halfvec <+> '[3,4]';
         7
 (1 row)
 
+SELECT l1_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ l1_distance 
+-------------
+           0
+(1 row)
+
+SELECT l1_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+ l1_distance 
+-------------
+          36
+(1 row)
+
 SELECT l2_normalize('[3,4]'::halfvec);
       l2_normalize      
 ------------------------
@@ -540,6 +606,12 @@ SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::halfvec)
  01001110101
 (1 row)
 
+SELECT binary_quantize('[1,2,3,-4,5,6,-7,8,1,-2,-3,4,5,-6,7,8,-1,2,3]'::halfvec);
+   binary_quantize   
+---------------------
+ 1110110110011011011
+(1 row)
+
 SELECT subvector('[1,2,3,4,5]'::halfvec, 1, 3);
  subvector 
 -----------
diff --git a/test/expected/vector_type.out b/test/expected/vector_type.out
index 674865822..f4c85d035 100644
--- a/test/expected/vector_type.out
+++ b/test/expected/vector_type.out
@@ -576,6 +576,12 @@ SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::vector);
  01001110101
 (1 row)
 
+SELECT binary_quantize('[1,2,3,-4,5,6,-7,8,1,-2,-3,4,5,-6,7,8,-1,2,3]'::vector);
+   binary_quantize   
+---------------------
+ 1110110110011011011
+(1 row)
+
 SELECT subvector('[1,2,3,4,5]'::vector, 1, 3);
  subvector 
 -----------
diff --git a/test/sql/halfvec.sql b/test/sql/halfvec.sql
index 1a3fd1b82..744d03868 100644
--- a/test/sql/halfvec.sql
+++ b/test/sql/halfvec.sql
@@ -87,12 +87,18 @@ SELECT l2_distance('[0,0]'::halfvec, '[0,1]');
 SELECT l2_distance('[1,2]'::halfvec, '[3]');
 SELECT l2_distance('[1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,4,5]');
 SELECT '[0,0]'::halfvec <-> '[3,4]';
+SELECT l2_distance('[501]'::halfvec, '[1]');
+SELECT l2_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+SELECT l2_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
 
 SELECT inner_product('[1,2]'::halfvec, '[3,4]');
 SELECT inner_product('[1,2]'::halfvec, '[3]');
 SELECT inner_product('[65504]'::halfvec, '[65504]');
 SELECT inner_product('[1,1,1,1,1,1,1,1,1]'::halfvec, '[1,2,3,4,5,6,7,8,9]');
 SELECT '[1,2]'::halfvec <#> '[3,4]';
+SELECT inner_product('[50000,50000]'::halfvec, '[1,1]');
+SELECT inner_product('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+SELECT inner_product('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
 
 SELECT cosine_distance('[1,2]'::halfvec, '[2,4]');
 SELECT cosine_distance('[1,2]'::halfvec, '[0,0]');
@@ -105,6 +111,9 @@ SELECT cosine_distance('[1,1]'::halfvec, '[-1.1,-1.1]');
 SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::halfvec, '[1,2,3,4,5,6,7,8,9]');
 SELECT cosine_distance('[1,2,3,4,5,6,7,8,9]'::halfvec, '[-1,-2,-3,-4,-5,-6,-7,-8,-9]');
 SELECT '[1,2]'::halfvec <=> '[2,4]';
+SELECT cosine_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+SELECT cosine_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+SELECT cosine_distance('[1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0]'::halfvec, '[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]');
 
 SELECT l1_distance('[0,0]'::halfvec, '[3,4]');
 SELECT l1_distance('[0,0]'::halfvec, '[0,1]');
@@ -112,6 +121,8 @@ SELECT l1_distance('[1,2]'::halfvec, '[3]');
 SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::halfvec, '[1,2,3,4,5,6,7,8,9]');
 SELECT l1_distance('[1,2,3,4,5,6,7,8,9]'::halfvec, '[0,3,2,5,4,7,6,9,8]');
 SELECT '[0,0]'::halfvec <+> '[3,4]';
+SELECT l1_distance('[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
+SELECT l1_distance('[2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]'::halfvec, '[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]');
 
 SELECT l2_normalize('[3,4]'::halfvec);
 SELECT l2_normalize('[3,0]'::halfvec);
@@ -121,6 +132,7 @@ SELECT l2_normalize('[65504]'::halfvec);
 
 SELECT binary_quantize('[1,0,-1]'::halfvec);
 SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::halfvec);
+SELECT binary_quantize('[1,2,3,-4,5,6,-7,8,1,-2,-3,4,5,-6,7,8,-1,2,3]'::halfvec);
 
 SELECT subvector('[1,2,3,4,5]'::halfvec, 1, 3);
 SELECT subvector('[1,2,3,4,5]'::halfvec, 3, 2);
diff --git a/test/sql/vector_type.sql b/test/sql/vector_type.sql
index 088b040aa..086a39bcf 100644
--- a/test/sql/vector_type.sql
+++ b/test/sql/vector_type.sql
@@ -128,6 +128,7 @@ SELECT l2_normalize('[3e38]'::vector);
 
 SELECT binary_quantize('[1,0,-1]'::vector);
 SELECT binary_quantize('[0,0.1,-0.2,-0.3,0.4,0.5,0.6,-0.7,0.8,-0.9,1]'::vector);
+SELECT binary_quantize('[1,2,3,-4,5,6,-7,8,1,-2,-3,4,5,-6,7,8,-1,2,3]'::vector);
 
 SELECT subvector('[1,2,3,4,5]'::vector, 1, 3);
 SELECT subvector('[1,2,3,4,5]'::vector, 3, 2);
diff --git a/vector.control b/vector.control
index 7bfc0f1fa..2ad02286a 100644
--- a/vector.control
+++ b/vector.control
@@ -1,4 +1,4 @@
 comment = 'vector data type and ivfflat and hnsw access methods'
-default_version = '0.8.0'
+default_version = '0.8.1'
 module_pathname = '$libdir/vector'
 relocatable = true