Skip to content

Speed up vector index tests by reducing dataset size #790

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

Merged
changhiskhan merged 2 commits into from
Apr 20, 2023
Merged

Speed up vector index tests by reducing dataset size #790

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bd4ff0d
Speed up vector index tests by reducing dataset size
changhiskhan Apr 20, 2023
cac789a
Merge branch 'main' into changhiskhan/speedup-tests
changhiskhan Apr 20, 2023
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
32 changes: 8 additions & 24 deletions python/python/tests/test_vector_index.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,14 +36,14 @@
from lance.vector import vec_to_table


def create_table(nvec=10000, ndim=768):
def create_table(nvec=1000, ndim=128):
mat = np.random.randn(nvec, ndim)
price = np.random.rand(nvec) * 100

def gen_str(n):
return "".join(random.choices(string.ascii_letters + string.digits, k=n))

meta = np.array([gen_str(1000) for _ in range(nvec)])
meta = np.array([gen_str(100) for _ in range(nvec)])
tbl = (
vec_to_table(data=mat)
.append_column("price", pa.array(price))
Expand All @@ -64,13 +64,13 @@ def indexed_dataset(tmp_path):
tbl = create_table()
dataset = lance.write_dataset(tbl, tmp_path)
yield dataset.create_index(
"vector", index_type="IVF_PQ", num_partitions=32, num_sub_vectors=16
"vector", index_type="IVF_PQ", num_partitions=4, num_sub_vectors=16
)


def run(ds, q=None, assert_func=None):
if q is None:
q = np.random.randn(768)
q = np.random.randn(128)
project = [None, ["price"], ["vector", "price"], ["vector", "meta", "price"]]
refine = [None, 1, 2]
filters = [None, pc.field("price") > 50.0]
Expand Down Expand Up @@ -119,23 +119,15 @@ def test_flat(dataset):
print(run(dataset))


@pytest.mark.skipif(
(platform.system() == "Darwin") and (platform.machine() != "arm64"),
reason="no neon on GHA",
)
def test_ann(indexed_dataset):
print(run(indexed_dataset))


@pytest.mark.skipif(
(platform.system() == "Darwin") and (platform.machine() != "arm64"),
reason="no neon on GHA",
)
def test_ann_append(tmp_path):
tbl = create_table()
dataset = lance.write_dataset(tbl, tmp_path)
dataset = dataset.create_index(
"vector", index_type="IVF_PQ", num_partitions=32, num_sub_vectors=16
"vector", index_type="IVF_PQ", num_partitions=4, num_sub_vectors=16
)
new_data = create_table(nvec=100)
dataset = lance.write_dataset(new_data, dataset.uri, mode="append")
Expand All @@ -147,16 +139,12 @@ def func(rs: pa.Table):
print(run(dataset, q=np.array(q), assert_func=func))


@pytest.mark.skipif(
(platform.system() == "Darwin") and (platform.machine() != "arm64"),
reason="no neon on GHA",
)
def test_use_index(dataset, tmp_path):
ann_ds = lance.write_dataset(dataset.to_table(), tmp_path / "indexed.lance")
ann_ds = ann_ds.create_index(
"vector", index_type="IVF_PQ", num_partitions=32, num_sub_vectors=16
"vector", index_type="IVF_PQ", num_partitions=4, num_sub_vectors=16
)
q = np.random.randn(768)
q = np.random.randn(128)
expected = dataset.to_table(
columns=["id"],
nearest={
Expand All @@ -174,15 +162,11 @@ def test_use_index(dataset, tmp_path):
assert np.all(expected == actual)


@pytest.mark.skipif(
(platform.system() == "Darwin") and (platform.machine() != "arm64"),
reason="no neon on GHA",
)
def test_has_index(dataset, tmp_path):
assert not dataset.has_index
ann_ds = lance.write_dataset(dataset.to_table(), tmp_path / "indexed.lance")
ann_ds = ann_ds.create_index(
"vector", index_type="IVF_PQ", num_partitions=32, num_sub_vectors=16
"vector", index_type="IVF_PQ", num_partitions=4, num_sub_vectors=16
)
assert ann_ds.has_index

Expand Down