Skip to content

feat: enlarge indice size limit for sparse vectors #229

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
zhourrr merged 7 commits into from
Mar 16, 2026
Merged

feat: enlarge indice size limit for sparse vectors #229

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
8d099f9
feat: enlarge the limit of sparse vectors
zhourrr Mar 13, 2026
7263b38
fix: revise logging
zhourrr Mar 16, 2026
9859d3e
fix: adjust query limit
zhourrr Mar 16, 2026
8eea0c4
fix: uses greater instead of greater equal
zhourrr Mar 16, 2026
bc33e81
fix: revise
zhourrr Mar 16, 2026
db86ab0
fix: validation [skip ci]
zhourrr Mar 16, 2026
c2c1aee
fix: unit test
zhourrr Mar 16, 2026
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
12 changes: 8 additions & 4 deletions src/core/algorithm/flat_sparse/flat_sparse_streamer.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -195,8 +195,10 @@ int FlatSparseStreamer::add_impl(uint64_t pkey, const uint32_t sparse_count,
}

if (ailego_unlikely(sparse_count > PARAM_FLAT_SPARSE_MAX_DIM_SIZE)) {
LOG_ERROR("Add vector failed, dim size too larg, dim_size=%u, key=%zu",
sparse_count, (size_t)pkey);
LOG_ERROR(
"Failed to add sparse vector: number of non-zero elements (%u) exceeds "
"maximum allowed (%u), key=%zu",
sparse_count, PARAM_FLAT_SPARSE_MAX_DIM_SIZE, (size_t)pkey);
(*stats_.mutable_discarded_count())++;
return IndexError_InvalidValue;
}
Expand Down Expand Up @@ -252,8 +254,10 @@ int FlatSparseStreamer::add_with_id_impl(uint32_t pkey,
}

if (ailego_unlikely(sparse_count > PARAM_FLAT_SPARSE_MAX_DIM_SIZE)) {
LOG_ERROR("Add vector failed, dim size too larg, dim_size=%u, key=%zu",
sparse_count, (size_t)pkey);
LOG_ERROR(
"Failed to add sparse vector: number of non-zero elements (%u) exceeds "
"maximum allowed (%u), key=%zu",
sparse_count, PARAM_FLAT_SPARSE_MAX_DIM_SIZE, (size_t)pkey);
(*stats_.mutable_discarded_count())++;
return IndexError_InvalidValue;
}
Expand Down
2 changes: 1 addition & 1 deletion src/core/algorithm/flat_sparse/flat_sparse_utility.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@
namespace zvec {
namespace core {

static constexpr uint32_t PARAM_FLAT_SPARSE_MAX_DIM_SIZE = 4096;
static constexpr uint32_t PARAM_FLAT_SPARSE_MAX_DIM_SIZE = 16384;
static const std::string PARAM_FLAT_SPARSE_META_SEG_ID =
"bruteforce_sparse_meta";

Expand Down
8 changes: 5 additions & 3 deletions src/core/algorithm/hnsw_sparse/hnsw_sparse_builder_entity.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,9 +88,11 @@ int HnswSparseBuilderEntity::add_vector(level_t level, key_t key,
const uint32_t sparse_count,
const uint32_t *sparse_indices,
const void *sparse_vec, node_id_t *id) {
if (ailego_unlikely(sparse_count >= HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN("Add vector failed, dim size too larg, dim_size=%u, key=%zu",
sparse_count, (size_t)key);
if (ailego_unlikely(sparse_count > HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN(
"Failed to add sparse vector: number of non-zero elements (%u) exceeds "
"maximum allowed (%u), key=%zu",
sparse_count, HnswSparseEntity::kSparseMaxDimSize, (size_t)key);
return IndexError_InvalidValue;
}

Expand Down
2 changes: 1 addition & 1 deletion src/core/algorithm/hnsw_sparse/hnsw_sparse_entity.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -610,7 +610,7 @@ class HnswSparseEntity {

constexpr static uint32_t kSparseMetaSize = 2u * sizeof(uint64_t);
constexpr static float kDefaultSparseNeighborRatio = 0.5f;
constexpr static uint32_t kSparseMaxDimSize = 4096;
constexpr static uint32_t kSparseMaxDimSize = 16384;
constexpr static float kDefaultQueryFilteringRatio = 0.0f; // turn off

protected:
Expand Down
16 changes: 10 additions & 6 deletions src/core/algorithm/hnsw_sparse/hnsw_sparse_streamer.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -438,9 +438,11 @@ int HnswSparseStreamer::add_with_id_impl(uint32_t id,
return ret;
}

if (ailego_unlikely(sparse_count >= HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN("Add vector failed, dim size too larg, dim_size=%u, id=%u",
sparse_count, id);
if (ailego_unlikely(sparse_count > HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN(
"Failed to add sparse vector: number of non-zero elements (%u) exceeds "
"maximum allowed (%u), id=%u",
sparse_count, HnswSparseEntity::kSparseMaxDimSize, id);
return IndexError_InvalidValue;
}

Expand Down Expand Up @@ -523,9 +525,11 @@ int HnswSparseStreamer::add_impl(uint64_t pkey, const uint32_t sparse_count,
return ret;
}

if (ailego_unlikely(sparse_count >= HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN("Add vector failed, dim size too larg, dim_size=%u, key=%zu",
sparse_count, (size_t)pkey);
if (ailego_unlikely(sparse_count > HnswSparseEntity::kSparseMaxDimSize)) {
LOG_WARN(
"Failed to add sparse vector: number of non-zero elements (%u) exceeds "
"maximum allowed (%u), key=%zu",
sparse_count, HnswSparseEntity::kSparseMaxDimSize, (size_t)pkey);
return IndexError_InvalidValue;
}

Expand Down
2 changes: 1 addition & 1 deletion src/db/common/constants.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,7 +32,7 @@ const std::string GLOBAL_DOC_ID = "_zvec_g_doc_id_";

const std::string USER_ID = "_zvec_uid_";

const int kSparseMaxDimSize = 4096;
const int kSparseMaxDimSize = 16384;

const int64_t kMaxRecordBatchNumRows = 4096;

Expand Down
18 changes: 16 additions & 2 deletions src/db/index/common/doc.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -866,6 +866,12 @@ Status Doc::validate(const CollectionSchema::Ptr &schema,
"doc validate failed: field[", field_name,
"]'s sparse vector indices and values size not match");
}
if (sparse_indices.size() > kSparseMaxDimSize) {
return Status::InvalidArgument(
"doc validate failed: vector[", field_name,
"], the number of sparse indices exceeds the maximum limit ",
kSparseMaxDimSize);
}
}
break;
}
Expand All @@ -881,6 +887,12 @@ Status Doc::validate(const CollectionSchema::Ptr &schema,
"doc validate failed: field[", field_name,
"]'s sparse vector indices and values size not match");
}
if (sparse_indices.size() > kSparseMaxDimSize) {
return Status::InvalidArgument(
"doc validate failed: vector[", field_name,
"], the number of sparse indices exceeds the maximum limit ",
kSparseMaxDimSize);
}
}
break;
}
Expand Down Expand Up @@ -1251,9 +1263,11 @@ Status VectorQuery::validate(const FieldSchema *schema) const {
}
} else if (schema->is_sparse_vector()) {
// validate sparse indices size
if (query_sparse_indices_.size() >= kSparseMaxDimSize * sizeof(uint32_t)) {
if (query_sparse_indices_.size() > kSparseMaxDimSize * sizeof(uint32_t)) {
return Status::InvalidArgument(
"query validate failed: sparse indices size is too large");
"query validate failed: the number of sparse indices exceeds the "
"maximum limit ",
kSparseMaxDimSize);
}
} else {
return Status::InvalidArgument(
Expand Down
2 changes: 1 addition & 1 deletion tests/db/index/common/doc_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1255,7 +1255,7 @@ TEST(VectorQuery, Validate) {
VectorQuery query;
query.field_name_ = "field_name";
query.topk_ = 100;
std::vector<uint32_t> query_indices = std::vector<uint32_t>(4097);
std::vector<uint32_t> query_indices = std::vector<uint32_t>(16385);
std::string query_indices_str =
std::string(reinterpret_cast<char *>(query_indices.data()),
query_indices.size() * sizeof(uint32_t));
Expand Down
Loading