admin/milvus
1
0
Fork 0
mirror of https://gitee.com/milvus-io/milvus.git synced 2025-12-07 01:28:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

Upgrade to knowhere-v1.3.0, remove following index support: (#18935)

Browse Source 
- IVF_SQ8H
- RHNSW_FLAT/RHNSW_PQ/RHNSW_SQ
- NGT
- NSG
- SPTAG

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>

Signed-off-by: yudong.cai <yudong.cai@zilliz.com>
This commit is contained in:
Cai Yudong 2022-09-05 10:41:11 +08:00 committed by GitHub
parent 9d8c3bf348
commit 686b0ce796
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
17 changed files with 55 additions and 485 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
internal/core/src/indexbuilder/VecIndexCreator.cpp
View File

@ -75,14 +75,6 @@ VecIndexCreator::parse_impl(const std::string& serialized_params_str, knowhere::
/************************** PQ Params *****************************/ /************************** PQ Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::PQ_M, stoi_closure, std::nullopt); check_parameter<int>(conf, knowhere::indexparam::PQ_M, stoi_closure, std::nullopt);
#ifdef MILVUS_SUPPORT_NSG
/************************** NSG Parameter **************************/
check_parameter<int>(conf, knowhere::indexparam::KNNG, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::SEARCH_LENGTH, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::OUT_DEGREE, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::CANDIDATE, stoi_closure, std::nullopt);
#endif
/************************** HNSW Params *****************************/ /************************** HNSW Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::EFCONSTRUCTION, stoi_closure, std::nullopt); check_parameter<int>(conf, knowhere::indexparam::EFCONSTRUCTION, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::HNSW_M, stoi_closure, std::nullopt); check_parameter<int>(conf, knowhere::indexparam::HNSW_M, stoi_closure, std::nullopt);
@ -91,23 +83,6 @@ VecIndexCreator::parse_impl(const std::string& serialized_params_str, knowhere::
/************************** Annoy Params *****************************/ /************************** Annoy Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::N_TREES, stoi_closure, std::nullopt); check_parameter<int>(conf, knowhere::indexparam::N_TREES, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::SEARCH_K, stoi_closure, std::nullopt); check_parameter<int>(conf, knowhere::indexparam::SEARCH_K, stoi_closure, std::nullopt);
#ifdef MILVUS_SUPPORT_NGT
/************************** NGT Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::EDGE_SIZE, stoi_closure, std::nullopt);
/************************** NGT Search Params *****************************/
check_parameter<float>(conf, knowhere::indexparam::EPSILON, stof_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::MAX_SEARCH_EDGES, stoi_closure, std::nullopt);
/************************** NGT_PANNG Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::FORCEDLY_PRUNED_EDGE_SIZE, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::SELECTIVELY_PRUNED_EDGE_SIZE, stoi_closure, std::nullopt);
/************************** NGT_ONNG Params *****************************/
check_parameter<int>(conf, knowhere::indexparam::OUTGOING_EDGE_SIZE, stoi_closure, std::nullopt);
check_parameter<int>(conf, knowhere::indexparam::INCOMING_EDGE_SIZE, stoi_closure, std::nullopt);
#endif
} }
void void

20
internal/core/src/indexbuilder/utils.cpp
View File

@ -22,10 +22,6 @@ std::vector<knowhere::IndexType>
NM_List() { NM_List() {
static std::vector<knowhere::IndexType> ret{ static std::vector<knowhere::IndexType> ret{
knowhere::IndexEnum::INDEX_FAISS_IVFFLAT, knowhere::IndexEnum::INDEX_FAISS_IVFFLAT,
#ifdef MILVUS_SUPPORT_NSG
knowhere::IndexEnum::INDEX_NSG,
#endif
knowhere::IndexEnum::INDEX_RHNSWFlat,
}; };
return ret; return ret;
} }
@ -43,22 +39,11 @@ std::vector<knowhere::IndexType>
Need_ID_List() { Need_ID_List() {
static std::vector<knowhere::IndexType> ret{ static std::vector<knowhere::IndexType> ret{
// knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, // knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT,
// knowhere::IndexEnum::INDEX_NSG,
}; };
return ret; return ret;
} }
std::vector<knowhere::IndexType>
Need_BuildAll_list() {
static std::vector<knowhere::IndexType> ret{
#ifdef MILVUS_SUPPORT_NSG
knowhere::IndexEnum::INDEX_NSG,
#endif
};
return ret;
}
std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>> std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>>
unsupported_index_combinations() { unsupported_index_combinations() {
static std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>> ret{ static std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>> ret{
@ -84,11 +69,6 @@ is_in_nm_list(const knowhere::IndexType& index_type) {
return is_in_list<knowhere::IndexType>(index_type, NM_List); return is_in_list<knowhere::IndexType>(index_type, NM_List);
} }
bool
is_in_need_build_all_list(const knowhere::IndexType& index_type) {
return is_in_list<knowhere::IndexType>(index_type, Need_BuildAll_list);
}
bool bool
is_in_need_id_list(const knowhere::IndexType& index_type) { is_in_need_id_list(const knowhere::IndexType& index_type) {
return is_in_list<knowhere::IndexType>(index_type, Need_ID_List); return is_in_list<knowhere::IndexType>(index_type, Need_ID_List);

6
internal/core/src/indexbuilder/utils.h
View File

@ -32,9 +32,6 @@ BIN_List();
std::vector<knowhere::IndexType> std::vector<knowhere::IndexType>
Need_ID_List(); Need_ID_List();
std::vector<knowhere::IndexType>
Need_BuildAll_list();
std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>> std::vector<std::tuple<knowhere::IndexType, knowhere::MetricType>>
unsupported_index_combinations(); unsupported_index_combinations();
@ -48,9 +45,6 @@ is_in_bin_list(const knowhere::IndexType& index_type);
bool bool
is_in_nm_list(const knowhere::IndexType& index_type); is_in_nm_list(const knowhere::IndexType& index_type);
bool
is_in_need_build_all_list(const knowhere::IndexType& index_type);
bool bool
is_in_need_id_list(const knowhere::IndexType& index_type); is_in_need_id_list(const knowhere::IndexType& index_type);

67
internal/core/src/query/deprecated/ValidationUtil.cpp
View File

@ -187,17 +187,8 @@ ValidateVectorIndexType(std::string& index_type, bool is_binary) {
knowhere::IndexEnum::INDEX_FAISS_IVFFLAT, knowhere::IndexEnum::INDEX_FAISS_IVFFLAT,
knowhere::IndexEnum::INDEX_FAISS_IVFPQ, knowhere::IndexEnum::INDEX_FAISS_IVFPQ,
knowhere::IndexEnum::INDEX_FAISS_IVFSQ8, knowhere::IndexEnum::INDEX_FAISS_IVFSQ8,
#ifdef MILVUS_GPU_VERSION
knowhere::IndexEnum::INDEX_FAISS_IVFSQ8H,
#endif
knowhere::IndexEnum::INDEX_NSG,
knowhere::IndexEnum::INDEX_HNSW, knowhere::IndexEnum::INDEX_HNSW,
knowhere::IndexEnum::INDEX_ANNOY, knowhere::IndexEnum::INDEX_ANNOY,
knowhere::IndexEnum::INDEX_RHNSWFlat,
knowhere::IndexEnum::INDEX_RHNSWPQ,
knowhere::IndexEnum::INDEX_RHNSWSQ,
knowhere::IndexEnum::INDEX_NGTPANNG,
knowhere::IndexEnum::INDEX_NGTONNG,
}; };
static std::set<std::string> s_binary_index_types = { static std::set<std::string> s_binary_index_types = {
@ -264,7 +255,6 @@ ValidateIndexParams(const milvus::json& index_params, int64_t dimension, const s
return Status::OK(); return Status::OK();
} else if (index_type == knowhere::IndexEnum::INDEX_FAISS_IVFFLAT || } else if (index_type == knowhere::IndexEnum::INDEX_FAISS_IVFFLAT ||
index_type == knowhere::IndexEnum::INDEX_FAISS_IVFSQ8 || index_type == knowhere::IndexEnum::INDEX_FAISS_IVFSQ8 ||
index_type == knowhere::IndexEnum::INDEX_FAISS_IVFSQ8H ||
index_type == knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT) { index_type == knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT) {
auto status = CheckParameterRange(index_params, knowhere::IndexParams::nlist, 1, 65536); auto status = CheckParameterRange(index_params, knowhere::IndexParams::nlist, 1, 65536);
if (!status.ok()) { if (!status.ok()) {
@ -311,26 +301,7 @@ ValidateIndexParams(const milvus::json& index_params, int64_t dimension, const s
LOG_SERVER_ERROR_ << msg; LOG_SERVER_ERROR_ << msg;
return Status(SERVER_INVALID_ARGUMENT, msg); return Status(SERVER_INVALID_ARGUMENT, msg);
}*/ }*/
} else if (index_type == knowhere::IndexEnum::INDEX_NSG) { } else if (index_type == knowhere::IndexEnum::INDEX_HNSW) {
auto status = CheckParameterRange(index_params, knowhere::IndexParams::search_length, 10, 300);
if (!status.ok()) {
return status;
}
status = CheckParameterRange(index_params, knowhere::IndexParams::out_degree, 5, 300);
if (!status.ok()) {
return status;
}
status = CheckParameterRange(index_params, knowhere::IndexParams::candidate, 50, 1000);
if (!status.ok()) {
return status;
}
status = CheckParameterRange(index_params, knowhere::IndexParams::knng, 5, 300);
if (!status.ok()) {
return status;
}
} else if (index_type == knowhere::IndexEnum::INDEX_HNSW || index_type == knowhere::IndexEnum::INDEX_RHNSWFlat ||
index_type == knowhere::IndexEnum::INDEX_RHNSWPQ || index_type == knowhere::IndexEnum::INDEX_RHNSWSQ ||
index_type == knowhere::IndexEnum::INDEX_RHNSWFlat) {
auto status = CheckParameterRange(index_params, knowhere::IndexParams::M, 4, 64); auto status = CheckParameterRange(index_params, knowhere::IndexParams::M, 4, 64);
if (!status.ok()) { if (!status.ok()) {
return status; return status;
@ -339,42 +310,6 @@ ValidateIndexParams(const milvus::json& index_params, int64_t dimension, const s
if (!status.ok()) { if (!status.ok()) {
return status; return status;
} }
if (index_type == knowhere::IndexEnum::INDEX_RHNSWPQ) {
status = CheckParameterExistence(index_params, knowhere::IndexParams::PQM);
if (!status.ok()) {
return status;
}
// special check for 'PQM' parameter
int64_t pqm_value = index_params[knowhere::IndexParams::PQM];
if (!milvus::knowhere::IVFPQConfAdapter::GetValidCPUM(dimension, pqm_value)) {
std::string msg = "Invalid m, dimension can't not be divided by m ";
LOG_SERVER_ERROR_ << msg;
return Status(SERVER_INVALID_ARGUMENT, msg);
}
/*int64_t pqm_value = index_params[knowhere::IndexParams::PQM];
if (resset.empty()) {
std::string msg = "Invalid collection dimension, unable to get reasonable values for 'PQM'";
LOG_SERVER_ERROR_ << msg;
return Status(SERVER_INVALID_COLLECTION_DIMENSION, msg);
}
auto iter = std::find(std::begin(resset), std::end(resset), pqm_value);
if (iter == std::end(resset)) {
std::string msg =
"Invalid " + std::string(knowhere::IndexParams::PQM) + ", must be one of the following values: ";
for (size_t i = 0; i < resset.size(); i++) {
if (i != 0) {
msg += ",";
}
msg += std::to_string(resset[i]);
}
LOG_SERVER_ERROR_ << msg;
return Status(SERVER_INVALID_ARGUMENT, msg);
}*/
}
} else if (index_type == knowhere::IndexEnum::INDEX_ANNOY) { } else if (index_type == knowhere::IndexEnum::INDEX_ANNOY) {
auto status = CheckParameterRange(index_params, knowhere::IndexParams::n_trees, 1, 1024); auto status = CheckParameterRange(index_params, knowhere::IndexParams::n_trees, 1, 1024);
if (!status.ok()) { if (!status.ok()) {

11
internal/core/src/query/visitors/VerifyPlanNodeVisitor.cpp
View File

@ -32,24 +32,15 @@ class VerifyPlanNodeVisitor : PlanNodeVisitor {
static knowhere::IndexType static knowhere::IndexType
InferIndexType(const Json& search_params) { InferIndexType(const Json& search_params) {
// ivf -> nprobe // ivf -> nprobe
// nsg -> search_length // hnsw -> ef
// hnsw/rhnsw/*pq/*sq -> ef
// annoy -> search_k // annoy -> search_k
// ngtpanng / ngtonng -> max_search_edges / epsilon
static const std::map<std::string, knowhere::IndexType> key_list = [] { static const std::map<std::string, knowhere::IndexType> key_list = [] {
std::map<std::string, knowhere::IndexType> list; std::map<std::string, knowhere::IndexType> list;
namespace ip = knowhere::indexparam; namespace ip = knowhere::indexparam;
namespace ie = knowhere::IndexEnum; namespace ie = knowhere::IndexEnum;
list.emplace(ip::NPROBE, ie::INDEX_FAISS_IVFFLAT); list.emplace(ip::NPROBE, ie::INDEX_FAISS_IVFFLAT);
#ifdef MILVUS_SUPPORT_NSG
list.emplace(ip::search_length, ie::INDEX_NSG);
#endif
list.emplace(ip::EF, ie::INDEX_HNSW); list.emplace(ip::EF, ie::INDEX_HNSW);
list.emplace(ip::SEARCH_K, ie::INDEX_ANNOY); list.emplace(ip::SEARCH_K, ie::INDEX_ANNOY);
#ifdef MILVUS_SUPPORT_NGT
list.emplace(ip::max_search_edges, ie::INDEX_NGTONNG);
list.emplace(ip::epsilon, ie::INDEX_NGTONNG);
#endif
return list; return list;
}(); }();
auto dbg_str = search_params.dump(); auto dbg_str = search_params.dump();

5
internal/core/thirdparty/knowhere/CMakeLists.txt vendored
View File

@ -11,8 +11,8 @@
# or implied. See the License for the specific language governing permissions and limitations under the License. # or implied. See the License for the specific language governing permissions and limitations under the License.
#------------------------------------------------------------------------------- #-------------------------------------------------------------------------------
set( KNOWHERE_VERSION v1.2.0 ) set( KNOWHERE_VERSION v1.3.0 )
set( KNOWHERE_SOURCE_MD5 "acc955a3b90c03f0afd0ff855e594777") set( KNOWHERE_SOURCE_MD5 "e43de07b5a5603bda1e340cc7309c115")
if ( DEFINED ENV{MILVUS_KNOWHERE_URL} ) if ( DEFINED ENV{MILVUS_KNOWHERE_URL} )
set( KNOWHERE_SOURCE_URL "$ENV{MILVUS_KNOWHERE_URL}" ) set( KNOWHERE_SOURCE_URL "$ENV{MILVUS_KNOWHERE_URL}" )
@ -29,6 +29,7 @@ macro(build_knowhere)
${EP_COMMON_CMAKE_ARGS} ${EP_COMMON_CMAKE_ARGS}
-DCMAKE_BUILD_TYPE=Release -DCMAKE_BUILD_TYPE=Release
-DCMAKE_INSTALL_PREFIX=${KNOWHERE_INSTALL_PREFIX} -DCMAKE_INSTALL_PREFIX=${KNOWHERE_INSTALL_PREFIX}
-DKNOWHERE_WITH_DISKANN=false
) )
externalproject_add(knowhere_ep externalproject_add(knowhere_ep

52
internal/core/unittest/test_index_wrapper.cpp
View File

@ -48,22 +48,8 @@ class IndexWrapperTest : public ::testing::TestWithParam<Param> {
{knowhere::IndexEnum::INDEX_FAISS_IVFSQ8, false}, {knowhere::IndexEnum::INDEX_FAISS_IVFSQ8, false},
{knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, true}, {knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, true},
{knowhere::IndexEnum::INDEX_FAISS_BIN_IDMAP, true}, {knowhere::IndexEnum::INDEX_FAISS_BIN_IDMAP, true},
#ifdef MILVUS_SUPPORT_SPTAG
{knowhere::IndexEnum::INDEX_SPTAG_KDT_RNT, false},
{knowhere::IndexEnum::INDEX_SPTAG_BKT_RNT, false},
#endif
{knowhere::IndexEnum::INDEX_HNSW, false}, {knowhere::IndexEnum::INDEX_HNSW, false},
{knowhere::IndexEnum::INDEX_ANNOY, false}, {knowhere::IndexEnum::INDEX_ANNOY, false},
{knowhere::IndexEnum::INDEX_RHNSWFlat, false},
{knowhere::IndexEnum::INDEX_RHNSWPQ, false},
{knowhere::IndexEnum::INDEX_RHNSWSQ, false},
#ifdef MILVUS_SUPPORT_NGT
{knowhere::IndexEnum::INDEX_NGTPANNG, false},
{knowhere::IndexEnum::INDEX_NGTONNG, false},
#endif
#ifdef MILVUS_SUPPORT_NSG
{knowhere::IndexEnum::INDEX_NSG, false},
#endif
}; };
is_binary = is_binary_map[index_type]; is_binary = is_binary_map[index_type];
@ -160,28 +146,6 @@ TEST(IVFFLATNM, Query) {
index->ClearStatistics(); index->ClearStatistics();
} }
#ifdef MILVUS_SUPPORT_NSG
TEST(NSG, Query) {
auto index_type = knowhere::IndexEnum::INDEX_NSG;
auto metric_type = knowhere::metric::L2;
auto conf = generate_conf(index_type, metric_type);
auto index = knowhere::VecIndexFactory::GetInstance().CreateVecIndex(index_type);
auto dataset = GenDataset(NB, metric_type, false);
auto xb_data = dataset.get_col<float>(0);
auto xb_dataset = knowhere::GenDataset(NB, DIM, xb_data.data());
index->BuildAll(xb_dataset, conf);
auto bs = index->Serialize(conf);
auto bptr = std::make_shared<knowhere::Binary>();
bptr->data = std::shared_ptr<uint8_t[]>((uint8_t*)xb_data.data(), [&](uint8_t*) {});
bptr->size = DIM * NB * sizeof(float);
bs.Append(RAW_DATA, bptr);
index->Load(bs);
auto xq_data = dataset.get_col<float>(0);
auto xq_dataset = knowhere::GenDataset(NQ, DIM, xq_data.data());
auto result = index->Query(xq_dataset, conf, nullptr);
}
#endif
TEST(BINFLAT, Build) { TEST(BINFLAT, Build) {
auto index_type = knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT; auto index_type = knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT;
auto metric_type = knowhere::metric::JACCARD; auto metric_type = knowhere::metric::JACCARD;
@ -394,22 +358,8 @@ INSTANTIATE_TEST_CASE_P(
std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, knowhere::metric::JACCARD), std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, knowhere::metric::JACCARD),
std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, knowhere::metric::TANIMOTO), std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IVFFLAT, knowhere::metric::TANIMOTO),
std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IDMAP, knowhere::metric::JACCARD), std::pair(knowhere::IndexEnum::INDEX_FAISS_BIN_IDMAP, knowhere::metric::JACCARD),
#ifdef MILVUS_SUPPORT_SPTAG
std::pair(knowhere::IndexEnum::INDEX_SPTAG_KDT_RNT, knowhere::metric::L2),
std::pair(knowhere::IndexEnum::INDEX_SPTAG_BKT_RNT, knowhere::metric::L2),
#endif
std::pair(knowhere::IndexEnum::INDEX_HNSW, knowhere::metric::L2), std::pair(knowhere::IndexEnum::INDEX_HNSW, knowhere::metric::L2),
std::pair(knowhere::IndexEnum::INDEX_ANNOY, knowhere::metric::L2), std::pair(knowhere::IndexEnum::INDEX_ANNOY, knowhere::metric::L2)
std::pair(knowhere::IndexEnum::INDEX_RHNSWFlat, knowhere::metric::L2),
std::pair(knowhere::IndexEnum::INDEX_RHNSWPQ, knowhere::metric::L2),
std::pair(knowhere::IndexEnum::INDEX_RHNSWSQ, knowhere::metric::L2)
#ifdef MILVUS_SUPPORT_NGT
std::pair(knowhere::IndexEnum::INDEX_NGTPANNG, knowhere::metric::L2),
std::pair(knowhere::IndexEnum::INDEX_NGTONNG, knowhere::metric::L2),
#endif
#ifdef MILVUS_SUPPORT_NSG
std::pair(knowhere::IndexEnum::INDEX_NSG, knowhere::metric::L2)
#endif
)); ));
TEST_P(IndexWrapperTest, Constructor) { TEST_P(IndexWrapperTest, Constructor) {

87
internal/core/unittest/test_utils/indexbuilder_test_utils.h
View File

@ -109,36 +109,6 @@ generate_conf(const knowhere::IndexType& index_type, const knowhere::MetricType&
{knowhere::meta::DIM, DIM}, {knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K}, {knowhere::meta::TOPK, K},
}; };
#ifdef MILVUS_SUPPORT_NSG
} else if (index_type == knowhere::IndexEnum::INDEX_NSG) {
return knowhere::Config{
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::NLIST, 163},
{knowhere::indexparam::NPROBE, 8},
{knowhere::indexparam::KNNG, 20},
{knowhere::indexparam::SEARCH_LENGTH, 40},
{knowhere::indexparam::OUT_DEGREE, 30},
{knowhere::indexparam::CANDIDATE, 100},
};
#endif
#ifdef MILVUS_SUPPORT_SPTAG
} else if (index_type == knowhere::IndexEnum::INDEX_SPTAG_KDT_RNT) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, 10},
};
} else if (index_type == knowhere::IndexEnum::INDEX_SPTAG_BKT_RNT) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, 10},
};
#endif
} else if (index_type == knowhere::IndexEnum::INDEX_HNSW) { } else if (index_type == knowhere::IndexEnum::INDEX_HNSW) {
return knowhere::Config{ return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size}, {knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
@ -158,63 +128,6 @@ generate_conf(const knowhere::IndexType& index_type, const knowhere::MetricType&
{knowhere::indexparam::N_TREES, 4}, {knowhere::indexparam::N_TREES, 4},
{knowhere::indexparam::SEARCH_K, 100}, {knowhere::indexparam::SEARCH_K, 100},
}; };
} else if (index_type == knowhere::IndexEnum::INDEX_RHNSWFlat) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::HNSW_M, 16},
{knowhere::indexparam::EFCONSTRUCTION, 200},
{knowhere::indexparam::EF, 200},
};
} else if (index_type == knowhere::IndexEnum::INDEX_RHNSWPQ) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::HNSW_M, 16},
{knowhere::indexparam::EFCONSTRUCTION, 200},
{knowhere::indexparam::EF, 200},
{knowhere::indexparam::PQ_M, 8},
};
} else if (index_type == knowhere::IndexEnum::INDEX_RHNSWSQ) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::HNSW_M, 16},
{knowhere::indexparam::EFCONSTRUCTION, 200},
{knowhere::indexparam::EF, 200},
};
#ifdef MILVUS_SUPPORT_NGT
} else if (index_type == knowhere::IndexEnum::INDEX_NGTPANNG) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::EDGE_SIZE, 10},
{knowhere::indexparam::EPSILON, 0.1},
{knowhere::indexparam::MAX_SEARCH_EDGES, 50},
{knowhere::indexparam::FORCEDLY_PRUNED_EDGE_SIZE, 60},
{knowhere::indexparam::SELECTIVELY_PRUNED_EDGE_SIZE, 30},
};
} else if (index_type == knowhere::IndexEnum::INDEX_NGTONNG) {
return knowhere::Config{
{knowhere::meta::SLICE_SIZE, knowhere::index_file_slice_size},
{knowhere::meta::METRIC_TYPE, metric_type},
{knowhere::meta::DIM, DIM},
{knowhere::meta::TOPK, K},
{knowhere::indexparam::EDGE_SIZE, 20},
{knowhere::indexparam::EPSILON, 0.1},
{knowhere::indexparam::MAX_SEARCH_EDGES, 50},
{knowhere::indexparam::OUTGOING_EDGE_SIZE, 5},
{knowhere::indexparam::INCOMING_EDGE_SIZE, 40},
};
#endif
} }
return knowhere::Config(); return knowhere::Config();
} }

72
internal/util/indexcgowrapper/index_test.go
View File

@ -4,12 +4,10 @@
package indexcgowrapper package indexcgowrapper
import ( import (
"fmt"
"strconv" "strconv"
"testing" "testing"
"github.com/milvus-io/milvus/internal/proto/schemapb" "github.com/milvus-io/milvus/internal/proto/schemapb"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
) )
@ -19,18 +17,11 @@ const (
IndexFaissIVFFlat = "IVF_FLAT" IndexFaissIVFFlat = "IVF_FLAT"
IndexFaissIVFPQ = "IVF_PQ" IndexFaissIVFPQ = "IVF_PQ"
IndexFaissIVFSQ8 = "IVF_SQ8" IndexFaissIVFSQ8 = "IVF_SQ8"
IndexFaissIVFSQ8H = "IVF_SQ8_HYBRID"
IndexFaissBinIDMap = "BIN_FLAT" IndexFaissBinIDMap = "BIN_FLAT"
IndexFaissBinIVFFlat = "BIN_IVF_FLAT" IndexFaissBinIVFFlat = "BIN_IVF_FLAT"
IndexNsg = "NSG"
IndexHNSW = "HNSW" IndexHNSW = "HNSW"
IndexRHNSWFlat = "RHNSW_FLAT"
IndexRHNSWPQ = "RHNSW_PQ"
IndexRHNSWSQ = "RHNSW_SQ"
IndexANNOY = "ANNOY" IndexANNOY = "ANNOY"
IndexNGTPANNG = "NGT_PANNG"
IndexNGTONNG = "NGT_ONNG"
// metric type // metric type
L2 = "L2" L2 = "L2"
@ -44,13 +35,9 @@ const (
m = 4 m = 4
nbits = 8 nbits = 8
nb = 1000 nb = 1000
nprobe = 8
sliceSize = 4 sliceSize = 4
efConstruction = 200 efConstruction = 200
ef = 200 ef = 200
edgeSize = 10
epsilon = 0.1
maxSearchEdges = 50
) )
type vecTestCase struct { type vecTestCase struct {
@ -70,24 +57,10 @@ func generateFloatVectorTestCases() []vecTestCase {
{IndexFaissIVFPQ, IP, false, schemapb.DataType_FloatVector}, {IndexFaissIVFPQ, IP, false, schemapb.DataType_FloatVector},
{IndexFaissIVFSQ8, L2, false, schemapb.DataType_FloatVector}, {IndexFaissIVFSQ8, L2, false, schemapb.DataType_FloatVector},
{IndexFaissIVFSQ8, IP, false, schemapb.DataType_FloatVector}, {IndexFaissIVFSQ8, IP, false, schemapb.DataType_FloatVector},
//{IndexFaissIVFSQ8H, L2, false, schemapb.DataType_FloatVector}, // TODO: enable gpu
//{IndexFaissIVFSQ8H, IP, false, schemapb.DataType_FloatVector},
//{IndexNsg, L2, false, schemapb.DataType_FloatVector},
//{IndexNsg, IP, false, schemapb.DataType_FloatVector},
{IndexHNSW, L2, false, schemapb.DataType_FloatVector}, {IndexHNSW, L2, false, schemapb.DataType_FloatVector},
{IndexHNSW, IP, false, schemapb.DataType_FloatVector}, {IndexHNSW, IP, false, schemapb.DataType_FloatVector},
{IndexRHNSWFlat, L2, false, schemapb.DataType_FloatVector},
{IndexRHNSWFlat, IP, false, schemapb.DataType_FloatVector},
{IndexRHNSWPQ, L2, false, schemapb.DataType_FloatVector},
{IndexRHNSWPQ, IP, false, schemapb.DataType_FloatVector},
{IndexRHNSWSQ, L2, false, schemapb.DataType_FloatVector},
{IndexRHNSWSQ, IP, false, schemapb.DataType_FloatVector},
{IndexANNOY, L2, false, schemapb.DataType_FloatVector}, {IndexANNOY, L2, false, schemapb.DataType_FloatVector},
{IndexANNOY, IP, false, schemapb.DataType_FloatVector}, {IndexANNOY, IP, false, schemapb.DataType_FloatVector},
//{IndexNGTPANNG, L2, false, schemapb.DataType_FloatVector},
//{IndexNGTPANNG, IP, false, schemapb.DataType_FloatVector},
//{IndexNGTONNG, L2, false, schemapb.DataType_FloatVector},
//{IndexNGTONNG, IP, false, schemapb.DataType_FloatVector},
} }
} }
@ -127,61 +100,16 @@ func generateParams(indexType, metricType string) (map[string]string, map[string
indexParams["nlist"] = strconv.Itoa(nlist) indexParams["nlist"] = strconv.Itoa(nlist)
indexParams["nbits"] = strconv.Itoa(nbits) indexParams["nbits"] = strconv.Itoa(nbits)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize) indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexFaissIVFSQ8H {
// TODO: enable gpu
} else if indexType == IndexNsg {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["nlist"] = strconv.Itoa(163)
indexParams["nprobe"] = strconv.Itoa(nprobe)
indexParams["knng"] = strconv.Itoa(20)
indexParams["search_length"] = strconv.Itoa(40)
indexParams["out_degree"] = strconv.Itoa(30)
indexParams["candidate_pool_size"] = strconv.Itoa(100)
} else if indexType == IndexHNSW { } else if indexType == IndexHNSW {
indexParams["dim"] = strconv.Itoa(dim) indexParams["dim"] = strconv.Itoa(dim)
indexParams["M"] = strconv.Itoa(16) indexParams["M"] = strconv.Itoa(16)
indexParams["efConstruction"] = strconv.Itoa(efConstruction) indexParams["efConstruction"] = strconv.Itoa(efConstruction)
indexParams["ef"] = strconv.Itoa(ef) indexParams["ef"] = strconv.Itoa(ef)
} else if indexType == IndexRHNSWFlat {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["M"] = strconv.Itoa(16)
indexParams["efConstruction"] = strconv.Itoa(efConstruction)
indexParams["ef"] = strconv.Itoa(ef)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexRHNSWPQ {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["M"] = strconv.Itoa(16)
indexParams["efConstruction"] = strconv.Itoa(efConstruction)
indexParams["ef"] = strconv.Itoa(ef)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
indexParams["PQM"] = strconv.Itoa(8)
} else if indexType == IndexRHNSWSQ {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["M"] = strconv.Itoa(16)
indexParams["efConstruction"] = strconv.Itoa(efConstruction)
indexParams["ef"] = strconv.Itoa(ef)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexANNOY { } else if indexType == IndexANNOY {
indexParams["dim"] = strconv.Itoa(dim) indexParams["dim"] = strconv.Itoa(dim)
indexParams["n_trees"] = strconv.Itoa(4) indexParams["n_trees"] = strconv.Itoa(4)
indexParams["search_k"] = strconv.Itoa(100) indexParams["search_k"] = strconv.Itoa(100)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize) indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexNGTPANNG {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["edge_size"] = strconv.Itoa(edgeSize)
indexParams["epsilon"] = fmt.Sprint(epsilon)
indexParams["max_search_edges"] = strconv.Itoa(maxSearchEdges)
indexParams["forcedly_pruned_edge_size"] = strconv.Itoa(60)
indexParams["selectively_pruned_edge_size"] = strconv.Itoa(30)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexNGTONNG {
indexParams["dim"] = strconv.Itoa(dim)
indexParams["edge_size"] = strconv.Itoa(edgeSize)
indexParams["epsilon"] = fmt.Sprint(epsilon)
indexParams["max_search_edges"] = strconv.Itoa(maxSearchEdges)
indexParams["outgoing_edge_size"] = strconv.Itoa(5)
indexParams["incoming_edge_size"] = strconv.Itoa(40)
indexParams["SLICE_SIZE"] = strconv.Itoa(sliceSize)
} else if indexType == IndexFaissBinIVFFlat { // binary vector } else if indexType == IndexFaissBinIVFFlat { // binary vector
indexParams["dim"] = strconv.Itoa(dim) indexParams["dim"] = strconv.Itoa(dim)
indexParams["nlist"] = strconv.Itoa(nlist) indexParams["nlist"] = strconv.Itoa(nlist)

4
tests/benchmark/milvus_benchmark/client.py
View File

@ -16,15 +16,11 @@ INDEX_MAP = {
"flat": "FLAT", "flat": "FLAT",
"ivf_flat": "IVF_FLAT", "ivf_flat": "IVF_FLAT",
"ivf_sq8": "IVF_SQ8", "ivf_sq8": "IVF_SQ8",
"nsg": "NSG",
"ivf_sq8h": "IVF_SQ8_HYBRID",
"ivf_pq": "IVF_PQ", "ivf_pq": "IVF_PQ",
"hnsw": "HNSW", "hnsw": "HNSW",
"annoy": "ANNOY", "annoy": "ANNOY",
"bin_flat": "BIN_FLAT", "bin_flat": "BIN_FLAT",
"bin_ivf_flat": "BIN_IVF_FLAT", "bin_ivf_flat": "BIN_IVF_FLAT",
"rhnsw_pq": "RHNSW_PQ",
"rhnsw_sq": "RHNSW_SQ"
} }
epsilon = 0.1 epsilon = 0.1
DEFAULT_WARM_QUERY_TOPK = 1 DEFAULT_WARM_QUERY_TOPK = 1

25
tests/python_client/common/common_func.py
View File

@ -386,19 +386,14 @@ def gen_invaild_search_params_type():
if index_type == "FLAT": if index_type == "FLAT":
continue continue
search_params.append({"index_type": index_type, "search_params": {"invalid_key": invalid_search_key}}) search_params.append({"index_type": index_type, "search_params": {"invalid_key": invalid_search_key}})
if index_type in ["IVF_FLAT", "IVF_SQ8", "IVF_SQ8H", "IVF_PQ"]: if index_type in ["IVF_FLAT", "IVF_SQ8", "IVF_PQ"]:
for nprobe in ct.get_invalid_ints: for nprobe in ct.get_invalid_ints:
ivf_search_params = {"index_type": index_type, "search_params": {"nprobe": nprobe}} ivf_search_params = {"index_type": index_type, "search_params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in ct.get_invalid_ints: for ef in ct.get_invalid_ints:
hnsw_search_param = {"index_type": index_type, "search_params": {"ef": ef}} hnsw_search_param = {"index_type": index_type, "search_params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type in ["NSG", "RNSG"]:
for search_length in ct.get_invalid_ints:
nsg_search_param = {"index_type": index_type, "search_params": {"search_length": search_length}}
search_params.append(nsg_search_param)
search_params.append({"index_type": index_type, "search_params": {"invalid_key": invalid_search_key}})
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in ct.get_invalid_ints: for search_k in ct.get_invalid_ints:
if isinstance(search_k, int): if isinstance(search_k, int):
@ -410,7 +405,7 @@ def gen_invaild_search_params_type():
def gen_search_param(index_type, metric_type="L2"): def gen_search_param(index_type, metric_type="L2"):
search_params = [] search_params = []
if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8H", "IVF_PQ"]: if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]:
for nprobe in [64, 128]: for nprobe in [64, 128]:
ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}} ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
@ -422,14 +417,10 @@ def gen_search_param(index_type, metric_type="L2"):
for nprobe in [64, 128]: for nprobe in [64, 128]:
binary_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}} binary_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}}
search_params.append(binary_search_params) search_params.append(binary_search_params)
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in [64, 32768]: for ef in [64, 32768]:
hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}} hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type in ["NSG", "RNSG"]:
for search_length in [100, 300]:
nsg_search_param = {"metric_type": metric_type, "params": {"search_length": search_length}}
search_params.append(nsg_search_param)
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in [1000, 5000]: for search_k in [1000, 5000]:
annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}} annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}}
@ -442,19 +433,15 @@ def gen_search_param(index_type, metric_type="L2"):
def gen_invalid_search_param(index_type, metric_type="L2"): def gen_invalid_search_param(index_type, metric_type="L2"):
search_params = [] search_params = []
if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8H", "IVF_PQ"] \ if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"] \
or index_type in ["BIN_FLAT", "BIN_IVF_FLAT"]: or index_type in ["BIN_FLAT", "BIN_IVF_FLAT"]:
for nprobe in [-1]: for nprobe in [-1]:
ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}} ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in [-1]: for ef in [-1]:
hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}} hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type in ["NSG", "RNSG"]:
for search_length in [100, 300]:
nsg_search_param = {"metric_type": metric_type, "params": {"search_length": search_length}}
search_params.append(nsg_search_param)
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in ["-1"]: for search_k in ["-1"]:
annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}} annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}}

14
tests/python_client/common/common_type.py
View File

@ -164,20 +164,16 @@ get_wrong_format_dict = [
] ]
""" Specially defined list """ """ Specially defined list """
all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ", all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY", "BIN_FLAT", "BIN_IVF_FLAT"]
"BIN_FLAT", "BIN_IVF_FLAT"]
default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8}, default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8},
{"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"M": 48, "efConstruction": 500}, {"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"nlist": 128}, {"nlist": 128}]
{"M": 48, "efConstruction": 500, "PQM": 8}, {"M": 48, "efConstruction": 500}, {"nlist": 128},
{"nlist": 128}]
Handler_type = ["GRPC", "HTTP"] Handler_type = ["GRPC", "HTTP"]
index_cpu_not_support = ["IVF_SQ8_HYBRID"]
binary_support = ["BIN_FLAT", "BIN_IVF_FLAT"] binary_support = ["BIN_FLAT", "BIN_IVF_FLAT"]
delete_support = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8_HYBRID", "IVF_PQ"] delete_support = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
ivf = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8_HYBRID", "IVF_PQ"] ivf = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
skip_pq = ["IVF_PQ", "RHNSW_PQ", "RHNSW_SQ"] skip_pq = ["IVF_PQ"]
binary_metrics = ["JACCARD", "HAMMING", "TANIMOTO", "SUBSTRUCTURE", "SUPERSTRUCTURE"] binary_metrics = ["JACCARD", "HAMMING", "TANIMOTO", "SUBSTRUCTURE", "SUPERSTRUCTURE"]
structure_metrics = ["SUBSTRUCTURE", "SUPERSTRUCTURE"] structure_metrics = ["SUBSTRUCTURE", "SUPERSTRUCTURE"]

15
tests/python_client/deploy/common.py
View File

@ -1,13 +1,10 @@
import json import json
from utils.util_log import test_log as log from utils.util_log import test_log as log
all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ", all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY", "BIN_FLAT", "BIN_IVF_FLAT"]
"BIN_FLAT", "BIN_IVF_FLAT"]
default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8}, default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8},
{"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"M": 48, "efConstruction": 500}, {"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"nlist": 128}, {"nlist": 128}]
{"M": 48, "efConstruction": 500, "PQM": 8}, {"M": 48, "efConstruction": 500}, {"nlist": 128},
{"nlist": 128}]
index_params_map = dict(zip(all_index_types, default_index_params)) index_params_map = dict(zip(all_index_types, default_index_params))
@ -25,7 +22,7 @@ def gen_index_param(index_type):
def gen_search_param(index_type, metric_type="L2"): def gen_search_param(index_type, metric_type="L2"):
search_params = [] search_params = []
if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8H", "IVF_PQ"]: if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]:
for nprobe in [10]: for nprobe in [10]:
ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}} ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
@ -33,14 +30,10 @@ def gen_search_param(index_type, metric_type="L2"):
for nprobe in [10]: for nprobe in [10]:
bin_search_params = {"metric_type": "HAMMING", "params": {"nprobe": nprobe}} bin_search_params = {"metric_type": "HAMMING", "params": {"nprobe": nprobe}}
search_params.append(bin_search_params) search_params.append(bin_search_params)
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in [64]: for ef in [64]:
hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}} hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type in ["NSG", "RNSG"]:
for search_length in [100]:
nsg_search_param = {"metric_type": metric_type, "params": {"search_length": search_length}}
search_params.append(nsg_search_param)
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in [1000]: for search_k in [1000]:
annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}} annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}}

14
tests/python_client/deploy/scripts/utils.py
View File

@ -5,12 +5,10 @@ from pymilvus import (
Collection, list_collections, Collection, list_collections,
) )
all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"] all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "ANNOY"]
default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8}, default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8},
{"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"M": 48, "efConstruction": 500}, {"M": 48, "efConstruction": 500}, {"n_trees": 50}, {"nlist": 128}, {"nlist": 128}]
{"M": 48, "efConstruction": 500, "PQM": 8}, {"M": 48, "efConstruction": 500}, {"nlist": 128},
{"nlist": 128}]
index_params_map = dict(zip(all_index_types, default_index_params)) index_params_map = dict(zip(all_index_types, default_index_params))
@ -28,7 +26,7 @@ def filter_collections_by_prefix(prefix):
def gen_search_param(index_type, metric_type="L2"): def gen_search_param(index_type, metric_type="L2"):
search_params = [] search_params = []
if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8H", "IVF_PQ"]: if index_type in ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]:
for nprobe in [10]: for nprobe in [10]:
ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}} ivf_search_params = {"metric_type": metric_type, "params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
@ -36,14 +34,10 @@ def gen_search_param(index_type, metric_type="L2"):
for nprobe in [10]: for nprobe in [10]:
bin_search_params = {"metric_type": "HAMMING", "params": {"nprobe": nprobe}} bin_search_params = {"metric_type": "HAMMING", "params": {"nprobe": nprobe}}
search_params.append(bin_search_params) search_params.append(bin_search_params)
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in [64]: for ef in [64]:
hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}} hnsw_search_param = {"metric_type": metric_type, "params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type in ["NSG", "RNSG"]:
for search_length in [100]:
nsg_search_param = {"metric_type": metric_type, "params": {"search_length": search_length}}
search_params.append(nsg_search_param)
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in [1000]: for search_k in [1000]:
annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}} annoy_search_param = {"metric_type": metric_type, "params": {"search_k": search_k}}

3
tests/python_client/testcases/test_collection.py
View File

@ -2584,9 +2584,6 @@ class TestLoadPartition(TestcaseBase):
params=gen_simple_index() params=gen_simple_index()
) )
def get_simple_index(self, request, connect): def get_simple_index(self, request, connect):
# if str(connect._cmd("mode")) == "CPU":
# if request.param["index_type"] in index_cpu_not_support():
# pytest.skip("sq8h not support in cpu mode")
return request.param return request.param
@pytest.fixture( @pytest.fixture(

40
tests/python_client/testcases/test_search.py
View File

@ -298,8 +298,8 @@ class TestCollectionSearchInvalid(TestcaseBase):
@pytest.mark.tags(CaseLabel.L1) @pytest.mark.tags(CaseLabel.L1)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_invalid_params_type(self, index, params): def test_search_invalid_params_type(self, index, params):
""" """
target: test search with invalid search params target: test search with invalid search params
@ -678,8 +678,8 @@ class TestCollectionSearchInvalid(TestcaseBase):
@pytest.mark.tags(CaseLabel.L1) @pytest.mark.tags(CaseLabel.L1)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[1:9], zip(ct.all_index_types[1:6],
ct.default_index_params[1:9])) ct.default_index_params[1:6]))
def test_search_different_index_invalid_params(self, index, params): def test_search_different_index_invalid_params(self, index, params):
""" """
target: test search with different index target: test search with different index
@ -1295,8 +1295,8 @@ class TestCollectionSearch(TestcaseBase):
@pytest.mark.tags(CaseLabel.L1) @pytest.mark.tags(CaseLabel.L1)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_after_different_index_with_params(self, dim, index, params, auto_id, _async): def test_search_after_different_index_with_params(self, dim, index, params, auto_id, _async):
""" """
target: test search after different index target: test search after different index
@ -1335,8 +1335,8 @@ class TestCollectionSearch(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_after_different_index_with_min_dim(self, index, params, auto_id, _async): def test_search_after_different_index_with_min_dim(self, index, params, auto_id, _async):
""" """
target: test search after different index with min dim target: test search after different index with min dim
@ -1373,8 +1373,8 @@ class TestCollectionSearch(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_after_index_different_metric_type(self, dim, index, params, auto_id, _async): def test_search_after_index_different_metric_type(self, dim, index, params, auto_id, _async):
""" """
target: test search with different metric type target: test search with different metric type
@ -2672,8 +2672,8 @@ class TestSearchBase(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_index_empty_partition(self, index, params): def test_search_index_empty_partition(self, index, params):
""" """
target: test basic search function, all the search params are correct, test all index params, and build target: test basic search function, all the search params are correct, test all index params, and build
@ -2721,8 +2721,8 @@ class TestSearchBase(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_index_partitions(self, index, params, get_top_k): def test_search_index_partitions(self, index, params, get_top_k):
""" """
target: test basic search function, all the search params are correct, test all index params, and build target: test basic search function, all the search params are correct, test all index params, and build
@ -2781,8 +2781,8 @@ class TestSearchBase(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_ip_after_index(self, index, params): def test_search_ip_after_index(self, index, params):
""" """
target: test basic search function, all the search params are correct, test all index params, and build target: test basic search function, all the search params are correct, test all index params, and build
@ -2838,8 +2838,8 @@ class TestSearchBase(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_ip_index_empty_partition(self, index, params): def test_search_ip_index_empty_partition(self, index, params):
""" """
target: test basic search function, all the search params are correct, test all index params, and build target: test basic search function, all the search params are correct, test all index params, and build
@ -2881,8 +2881,8 @@ class TestSearchBase(TestcaseBase):
@pytest.mark.tags(CaseLabel.L2) @pytest.mark.tags(CaseLabel.L2)
@pytest.mark.parametrize("index, params", @pytest.mark.parametrize("index, params",
zip(ct.all_index_types[:9], zip(ct.all_index_types[:6],
ct.default_index_params[:9])) ct.default_index_params[:6]))
def test_search_ip_index_partitions(self, index, params): def test_search_ip_index_partitions(self, index, params):
""" """
target: test basic search function, all the search params are correct, test all index params, and build target: test basic search function, all the search params are correct, test all index params, and build

74
tests/python_client/utils/util_pymilvus.py
View File

@ -32,19 +32,13 @@ default_tag = "1970_01_01"
row_count = "row_count" row_count = "row_count"
# TODO: # TODO:
# TODO: disable RHNSW_SQ/PQ in 0.11.0
all_index_types = [ all_index_types = [
"FLAT", "FLAT",
"IVF_FLAT", "IVF_FLAT",
"IVF_SQ8", "IVF_SQ8",
# "IVF_SQ8_HYBRID",
"IVF_PQ", "IVF_PQ",
"HNSW", "HNSW",
# "NSG",
"ANNOY", "ANNOY",
"RHNSW_FLAT",
"RHNSW_PQ",
"RHNSW_SQ",
"BIN_FLAT", "BIN_FLAT",
"BIN_IVF_FLAT" "BIN_IVF_FLAT"
] ]
@ -53,14 +47,9 @@ default_index_params = [
{"nlist": 128}, {"nlist": 128},
{"nlist": 128}, {"nlist": 128},
{"nlist": 128}, {"nlist": 128},
# {"nlist": 128},
{"nlist": 128, "m": 16, "nbits": 8}, {"nlist": 128, "m": 16, "nbits": 8},
{"M": 48, "efConstruction": 500}, {"M": 48, "efConstruction": 500},
# {"search_length": 50, "out_degree": 40, "candidate_pool_size": 100, "knng": 50},
{"n_trees": 50}, {"n_trees": 50},
{"M": 48, "efConstruction": 500},
{"M": 48, "efConstruction": 500, "PQM": 8},
{"M": 48, "efConstruction": 500},
{"nlist": 128}, {"nlist": 128},
{"nlist": 128} {"nlist": 128}
] ]
@ -70,24 +59,20 @@ def create_target_index(index, field_name):
index["field_name"] = field_name index["field_name"] = field_name
def index_cpu_not_support():
return ["IVF_SQ8_HYBRID"]
def binary_support(): def binary_support():
return ["BIN_FLAT", "BIN_IVF_FLAT"] return ["BIN_FLAT", "BIN_IVF_FLAT"]
def delete_support(): def delete_support():
return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8_HYBRID", "IVF_PQ"] return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
def ivf(): def ivf():
return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_SQ8_HYBRID", "IVF_PQ"] return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
def skip_pq(): def skip_pq():
return ["IVF_PQ", "RHNSW_PQ", "RHNSW_SQ"] return ["IVF_PQ"]
def binary_metrics(): def binary_metrics():
@ -748,15 +733,10 @@ def gen_invaild_search_params():
for nprobe in gen_invalid_params(): for nprobe in gen_invalid_params():
ivf_search_params = {"index_type": index_type, "search_params": {"nprobe": nprobe}} ivf_search_params = {"index_type": index_type, "search_params": {"nprobe": nprobe}}
search_params.append(ivf_search_params) search_params.append(ivf_search_params)
elif index_type in ["HNSW", "RHNSW_PQ", "RHNSW_SQ"]: elif index_type in ["HNSW"]:
for ef in gen_invalid_params(): for ef in gen_invalid_params():
hnsw_search_param = {"index_type": index_type, "search_params": {"ef": ef}} hnsw_search_param = {"index_type": index_type, "search_params": {"ef": ef}}
search_params.append(hnsw_search_param) search_params.append(hnsw_search_param)
elif index_type == "NSG":
for search_length in gen_invalid_params():
nsg_search_param = {"index_type": index_type, "search_params": {"search_length": search_length}}
search_params.append(nsg_search_param)
search_params.append({"index_type": index_type, "search_params": {"invalid_key": 100}})
elif index_type == "ANNOY": elif index_type == "ANNOY":
for search_k in gen_invalid_params(): for search_k in gen_invalid_params():
if isinstance(search_k, int): if isinstance(search_k, int):
@ -776,36 +756,12 @@ def gen_invalid_index():
index_params.append(index_param) index_params.append(index_param)
for M in gen_invalid_params(): for M in gen_invalid_params():
index_param = {"index_type": "HNSW", "params": {"M": M, "efConstruction": 100}} index_param = {"index_type": "HNSW", "params": {"M": M, "efConstruction": 100}}
index_param = {"index_type": "RHNSW_PQ", "params": {"M": M, "efConstruction": 100}}
index_param = {"index_type": "RHNSW_SQ", "params": {"M": M, "efConstruction": 100}}
index_params.append(index_param) index_params.append(index_param)
for efConstruction in gen_invalid_params(): for efConstruction in gen_invalid_params():
index_param = {"index_type": "HNSW", "params": {"M": 16, "efConstruction": efConstruction}} index_param = {"index_type": "HNSW", "params": {"M": 16, "efConstruction": efConstruction}}
index_param = {"index_type": "RHNSW_PQ", "params": {"M": 16, "efConstruction": efConstruction}}
index_param = {"index_type": "RHNSW_SQ", "params": {"M": 16, "efConstruction": efConstruction}}
index_params.append(index_param)
for search_length in gen_invalid_params():
index_param = {"index_type": "NSG",
"params": {"search_length": search_length, "out_degree": 40, "candidate_pool_size": 50,
"knng": 100}}
index_params.append(index_param)
for out_degree in gen_invalid_params():
index_param = {"index_type": "NSG",
"params": {"search_length": 100, "out_degree": out_degree, "candidate_pool_size": 50,
"knng": 100}}
index_params.append(index_param)
for candidate_pool_size in gen_invalid_params():
index_param = {"index_type": "NSG", "params": {"search_length": 100, "out_degree": 40,
"candidate_pool_size": candidate_pool_size,
"knng": 100}}
index_params.append(index_param) index_params.append(index_param)
index_params.append({"index_type": "IVF_FLAT", "params": {"invalid_key": 1024}}) index_params.append({"index_type": "IVF_FLAT", "params": {"invalid_key": 1024}})
index_params.append({"index_type": "HNSW", "params": {"invalid_key": 16, "efConstruction": 100}}) index_params.append({"index_type": "HNSW", "params": {"invalid_key": 16, "efConstruction": 100}})
index_params.append({"index_type": "RHNSW_PQ", "params": {"invalid_key": 16, "efConstruction": 100}})
index_params.append({"index_type": "RHNSW_SQ", "params": {"invalid_key": 16, "efConstruction": 100}})
index_params.append({"index_type": "NSG",
"params": {"invalid_key": 100, "out_degree": 40, "candidate_pool_size": 300,
"knng": 100}})
for invalid_n_trees in gen_invalid_params(): for invalid_n_trees in gen_invalid_params():
index_params.append({"index_type": "ANNOY", "params": {"n_trees": invalid_n_trees}}) index_params.append({"index_type": "ANNOY", "params": {"n_trees": invalid_n_trees}})
@ -817,16 +773,12 @@ def gen_index():
pq_ms = [128, 64, 32, 16, 8, 4] pq_ms = [128, 64, 32, 16, 8, 4]
Ms = [5, 24, 48] Ms = [5, 24, 48]
efConstructions = [100, 300, 500] efConstructions = [100, 300, 500]
search_lengths = [10, 100, 300]
out_degrees = [5, 40, 300]
candidate_pool_sizes = [50, 100, 300]
knngs = [5, 100, 300]
index_params = [] index_params = []
for index_type in all_index_types: for index_type in all_index_types:
if index_type in ["FLAT", "BIN_FLAT", "BIN_IVF_FLAT"]: if index_type in ["FLAT", "BIN_FLAT", "BIN_IVF_FLAT"]:
index_params.append({"index_type": index_type, "index_param": {"nlist": 1024}}) index_params.append({"index_type": index_type, "index_param": {"nlist": 1024}})
elif index_type in ["IVF_FLAT", "IVF_SQ8", "IVF_SQ8_HYBRID"]: elif index_type in ["IVF_FLAT", "IVF_SQ8"]:
ivf_params = [{"index_type": index_type, "index_param": {"nlist": nlist}} \ ivf_params = [{"index_type": index_type, "index_param": {"nlist": nlist}} \
for nlist in nlists] for nlist in nlists]
index_params.extend(ivf_params) index_params.extend(ivf_params)
@ -835,20 +787,11 @@ def gen_index():
for nlist in nlists \ for nlist in nlists \
for m in pq_ms] for m in pq_ms]
index_params.extend(IVFPQ_params) index_params.extend(IVFPQ_params)
elif index_type in ["HNSW", "RHNSW_SQ", "RHNSW_PQ"]: elif index_type in ["HNSW"]:
hnsw_params = [{"index_type": index_type, "index_param": {"M": M, "efConstruction": efConstruction}} \ hnsw_params = [{"index_type": index_type, "index_param": {"M": M, "efConstruction": efConstruction}} \
for M in Ms \ for M in Ms \
for efConstruction in efConstructions] for efConstruction in efConstructions]
index_params.extend(hnsw_params) index_params.extend(hnsw_params)
elif index_type == "NSG":
nsg_params = [{"index_type": index_type,
"index_param": {"search_length": search_length, "out_degree": out_degree,
"candidate_pool_size": candidate_pool_size, "knng": knng}} \
for search_length in search_lengths \
for out_degree in out_degrees \
for candidate_pool_size in candidate_pool_sizes \
for knng in knngs]
index_params.extend(nsg_params)
return index_params return index_params
@ -888,12 +831,9 @@ def get_search_param(index_type, metric_type="L2"):
if index_type in ivf() or index_type in binary_support(): if index_type in ivf() or index_type in binary_support():
nprobe64 = {"nprobe": 64} nprobe64 = {"nprobe": 64}
search_params.update({"params": nprobe64}) search_params.update({"params": nprobe64})
elif index_type in ["HNSW", "RHNSW_FLAT", "RHNSW_SQ", "RHNSW_PQ"]: elif index_type in ["HNSW"]:
ef64 = {"ef": 64} ef64 = {"ef": 64}
search_params.update({"params": ef64}) search_params.update({"params": ef64})
elif index_type == "NSG":
length100 = {"search_length": 100}
search_params.update({"params": length100})
elif index_type == "ANNOY": elif index_type == "ANNOY":
search_k = {"search_k": 1000} search_k = {"search_k": 1000}
search_params.update({"params": search_k}) search_params.update({"params": search_k})