Skip to content

fix: breaking change for pgvecto_rs #12

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
silver-ymz wants to merge 1 commit into
base: master
Choose a base branch
from
Open

fix: breaking change for pgvecto_rs #12

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions engine/clients/pgvector/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,9 +13,9 @@
}

DISTANCE_MAPPING_CREATE_RUST = {
Distance.L2: "l2_ops",
Distance.COSINE: "cosine_ops",
Distance.DOT: "dot_ops",
Distance.L2: "vector_l2_ops",
Distance.COSINE: "vector_cos_ops",
Distance.DOT: "vector_dot_ops",
}

DISTANCE_MAPPING_SEARCH = {
Expand Down
7 changes: 1 addition & 6 deletions engine/clients/pgvector/search.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,12 +33,7 @@ def search_one(cls, vector: List[float], meta_conditions, top: Optional[int], sc
cur.execute("BEGIN;")
# set index create parameter
for key in cls.search_params["params"].keys():
if cls.engine_type == "c":
cur.execute(f"SET LOCAL {key} = {cls.search_params['params'][key]};")
else:
# pgvector_rs only support hnsw
cur.execute(f"SET LOCAL vectors.k = {cls.search_params['params']['hnsw.ef_search']};")
break
cur.execute(f"SET LOCAL {key} = {cls.search_params['params'][key]};")

meta_conditions = cls.parser.parse(meta_conditions)
if meta_conditions:
Expand Down
16 changes: 9 additions & 7 deletions engine/clients/pgvector/upload.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,15 +30,15 @@ def upload_batch(cls, ids: List[int], vectors: List[list], metadata: List[Option
raise RuntimeError("PGVector batch upload unhealthy")
# Getting the names of structured data columns based on the first meta information.
col_name_tuple = ('id', 'vector')
col_type_tuple = ('%s', '%s::real[]')
col_type_tuple = ('%s', '%s')
if metadata[0] is not None:
for col_name in list(metadata[0].keys()):
col_name_tuple += (col_name,)
col_type_tuple += ('%s',)

insert_data = []
for i in range(0, len(ids)):
temp_tuple = (ids[i], vectors[i])
temp_tuple = (ids[i], str(vectors[i]))
if metadata[i] is not None:
for col_name in list(metadata[i].keys()):
value = metadata[i][col_name]
Expand Down Expand Up @@ -74,11 +74,6 @@ def post_upload(cls, distance):
if cls.engine_type == "rust":
create_index_command = f"""
CREATE INDEX ON {PGVECTOR_INDEX} USING vectors (vector {cls.distance}) WITH (options=$$
capacity = {int(cls.vector_count*1.2)}
[vectors]
memmap = "ram"
[algorithm.hnsw]
memmap = "ram"
{index_options_rust}
$$);
"""
Expand All @@ -92,3 +87,10 @@ def post_upload(cls, distance):
with cls.conn.cursor() as cur:
cur.execute("SELECT phase, tuples_done, tuples_total FROM pg_stat_progress_create_index;")
cls.conn.commit()
if cls.engine_type == "rust":
with cls.conn.cursor() as cur:
indexing = True
while indexing:
cur.execute("SELECT idx_indexing FROM pg_vector_index_stat;")
indexing = cur.fetchone()[0]
cls.conn.commit()
16 changes: 9 additions & 7 deletions experiments/needs_editing/pgvector_rust_HNSW_single_node_laion-768-5m-ip.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
"platform": "CloudTest_v0.0.4",
"index_type": "HNSW",
"dataset": "laion-768-5m-ip",
"version": "pg15-latest",
"version": "pg16-v0.2.0",
"branch": "master",
"commit": "sha256:38c7e5c4fa3afd48fb4911b3f96489ac287b59f8bfdd9b7b55c0eca898ffec21",
"remark": "pgvector implemented in RUST",
Expand All @@ -24,37 +24,39 @@
"parallel": 4,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100
}
},
{
"parallel": 4,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100
}
},
{
"parallel": 8,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100
}
},
{
"parallel": 8,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100
}
}
],
"upload_params": {
"parallel": 16,
"batch_size": 64,
"index_params": {
"m": 12,
"ef_construction": 100
"indexing.hnsw.m": 12,
"indexing.hnsw.ef_construction": 100,
"optimizing.optimizing_threads": 8,
"segment.max_sealed_segment_size": 5000000
},
"index_type": "hnsw",
"engine_type": "rust"
Expand Down
32 changes: 21 additions & 11 deletions experiments/needs_editing/pgvector_rust_HNSW_single_node_laion-768-5m-probability-ip.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
"platform": "CloudTest_v0.0.4",
"index_type": "HNSW",
"dataset": "laion-768-5m-ip-probability",
"version": "pg15-latest",
"version": "pg16-v0.2.0",
"branch": "master",
"commit": "sha256:38c7e5c4fa3afd48fb4911b3f96489ac287b59f8bfdd9b7b55c0eca898ffec21",
"remark": "pgvector implemented in RUST",
Expand All @@ -24,7 +24,8 @@
"parallel": 4,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.01
Expand All @@ -34,7 +35,8 @@
"parallel": 4,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.1
Expand All @@ -44,7 +46,8 @@
"parallel": 4,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.01
Expand All @@ -54,7 +57,8 @@
"parallel": 4,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.1
Expand All @@ -64,7 +68,8 @@
"parallel": 8,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.01
Expand All @@ -74,7 +79,8 @@
"parallel": 8,
"top": 10,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.1
Expand All @@ -84,7 +90,8 @@
"parallel": 8,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.01
Expand All @@ -94,7 +101,8 @@
"parallel": 8,
"top": 100,
"params": {
"hnsw.ef_search": 100
"vectors.hnsw_ef_search": 100,
"vectors.search_mode": "vbase"
},
"query_meta": {
"probability": 0.1
Expand All @@ -105,8 +113,10 @@
"parallel": 16,
"batch_size": 64,
"index_params": {
"m": 12,
"ef_construction": 100
"indexing.hnsw.m": 12,
"indexing.hnsw.ef_construction": 100,
"optimizing.optimizing_threads": 8,
"segment.max_sealed_segment_size": 5000000
},
"index_type": "hnsw",
"engine_type": "rust"
Expand Down