Loading CHANGELOG.md +2 −1 Original line number Diff line number Diff line Loading @@ -6,12 +6,13 @@ Please mark all change in change log and use the ticket from JIRA. ## Bug - \#134 - JFrog cache error - \#161 - Search IVFSQHybrid crash on gpu ## Feature - \#90 - The server start error messages could be improved to enhance user experience - \#104 - test_scheduler core dump - \#115 - Using new structure for tasktable - \#139 - New config opion use_gpu_threshold - \#139 - New config option use_gpu_threshold - \#146 - Add only GPU and only CPU version for IVF_SQ8 and IVF_FLAT ## Improvement Loading ci/jenkins/pod/milvus-build-env-pod.yaml +1 −1 Original line number Diff line number Diff line Loading @@ -8,7 +8,7 @@ metadata: spec: containers: - name: milvus-build-env image: registry.zilliz.com/milvus/milvus-build-env:v0.5.0-ubuntu18.04 image: registry.zilliz.com/milvus/milvus-build-env:v0.5.1-ubuntu18.04 env: - name: POD_IP valueFrom: Loading core/src/db/engine/ExecutionEngineImpl.cpp +2 −0 Original line number Diff line number Diff line Loading @@ -256,6 +256,7 @@ ExecutionEngineImpl::Load(bool to_cache) { Status ExecutionEngineImpl::CopyToGpu(uint64_t device_id, bool hybrid) { #if 0 if (hybrid) { const std::string key = location_ + ".quantizer"; std::vector<uint64_t> gpus{device_id}; Loading Loading @@ -307,6 +308,7 @@ ExecutionEngineImpl::CopyToGpu(uint64_t device_id, bool hybrid) { } return Status::OK(); } #endif auto index = std::static_pointer_cast<VecIndex>(cache::GpuCacheMgr::GetInstance(device_id)->GetIndex(location_)); bool already_in_cache = (index != nullptr); Loading core/src/index/knowhere/knowhere/index/vector_index/IndexIVFSQHybrid.cpp +6 −14 Original line number Diff line number Diff line Loading @@ -84,22 +84,14 @@ IVFSQHybrid::CopyGpuToCpu(const Config& config) { VectorIndexPtr IVFSQHybrid::CopyCpuToGpu(const int64_t& device_id, const Config& config) { if (gpu_mode != 0) { KNOWHERE_THROW_MSG("Not a GpuIndex Type"); } if (auto res = FaissGpuResourceMgr::GetInstance().GetRes(device_id)) { ResScope rs(res, device_id, false); faiss::gpu::GpuClonerOptions option; option.allInGpu = true; faiss::IndexComposition index_composition; index_composition.index = index_.get(); index_composition.quantizer = nullptr; index_composition.mode = 0; // copy all auto gpu_index = faiss::gpu::index_cpu_to_gpu(res->faiss_res.get(), device_id, &index_composition, &option); auto idx = dynamic_cast<faiss::IndexIVF*>(index_.get()); idx->restore_quantizer(); auto gpu_index = faiss::gpu::index_cpu_to_gpu(res->faiss_res.get(), device_id, index_.get(), &option); std::shared_ptr<faiss::Index> device_index = std::shared_ptr<faiss::Index>(gpu_index); auto new_idx = std::make_shared<IVFSQHybrid>(device_index, device_id, res); return new_idx; Loading core/src/index/unittest/test_ivf.cpp +4 −0 Original line number Diff line number Diff line Loading @@ -213,6 +213,10 @@ TEST_P(IVFTest, clone_test) { } } if (index_type == "IVFSQHybrid") { return; } { // copy to gpu std::vector<std::string> support_idx_vec{"IVF", "GPUIVF", "IVFSQ", "GPUIVFSQ"}; Loading Loading
CHANGELOG.md +2 −1 Original line number Diff line number Diff line Loading @@ -6,12 +6,13 @@ Please mark all change in change log and use the ticket from JIRA. ## Bug - \#134 - JFrog cache error - \#161 - Search IVFSQHybrid crash on gpu ## Feature - \#90 - The server start error messages could be improved to enhance user experience - \#104 - test_scheduler core dump - \#115 - Using new structure for tasktable - \#139 - New config opion use_gpu_threshold - \#139 - New config option use_gpu_threshold - \#146 - Add only GPU and only CPU version for IVF_SQ8 and IVF_FLAT ## Improvement Loading
ci/jenkins/pod/milvus-build-env-pod.yaml +1 −1 Original line number Diff line number Diff line Loading @@ -8,7 +8,7 @@ metadata: spec: containers: - name: milvus-build-env image: registry.zilliz.com/milvus/milvus-build-env:v0.5.0-ubuntu18.04 image: registry.zilliz.com/milvus/milvus-build-env:v0.5.1-ubuntu18.04 env: - name: POD_IP valueFrom: Loading
core/src/db/engine/ExecutionEngineImpl.cpp +2 −0 Original line number Diff line number Diff line Loading @@ -256,6 +256,7 @@ ExecutionEngineImpl::Load(bool to_cache) { Status ExecutionEngineImpl::CopyToGpu(uint64_t device_id, bool hybrid) { #if 0 if (hybrid) { const std::string key = location_ + ".quantizer"; std::vector<uint64_t> gpus{device_id}; Loading Loading @@ -307,6 +308,7 @@ ExecutionEngineImpl::CopyToGpu(uint64_t device_id, bool hybrid) { } return Status::OK(); } #endif auto index = std::static_pointer_cast<VecIndex>(cache::GpuCacheMgr::GetInstance(device_id)->GetIndex(location_)); bool already_in_cache = (index != nullptr); Loading
core/src/index/knowhere/knowhere/index/vector_index/IndexIVFSQHybrid.cpp +6 −14 Original line number Diff line number Diff line Loading @@ -84,22 +84,14 @@ IVFSQHybrid::CopyGpuToCpu(const Config& config) { VectorIndexPtr IVFSQHybrid::CopyCpuToGpu(const int64_t& device_id, const Config& config) { if (gpu_mode != 0) { KNOWHERE_THROW_MSG("Not a GpuIndex Type"); } if (auto res = FaissGpuResourceMgr::GetInstance().GetRes(device_id)) { ResScope rs(res, device_id, false); faiss::gpu::GpuClonerOptions option; option.allInGpu = true; faiss::IndexComposition index_composition; index_composition.index = index_.get(); index_composition.quantizer = nullptr; index_composition.mode = 0; // copy all auto gpu_index = faiss::gpu::index_cpu_to_gpu(res->faiss_res.get(), device_id, &index_composition, &option); auto idx = dynamic_cast<faiss::IndexIVF*>(index_.get()); idx->restore_quantizer(); auto gpu_index = faiss::gpu::index_cpu_to_gpu(res->faiss_res.get(), device_id, index_.get(), &option); std::shared_ptr<faiss::Index> device_index = std::shared_ptr<faiss::Index>(gpu_index); auto new_idx = std::make_shared<IVFSQHybrid>(device_index, device_id, res); return new_idx; Loading
core/src/index/unittest/test_ivf.cpp +4 −0 Original line number Diff line number Diff line Loading @@ -213,6 +213,10 @@ TEST_P(IVFTest, clone_test) { } } if (index_type == "IVFSQHybrid") { return; } { // copy to gpu std::vector<std::string> support_idx_vec{"IVF", "GPUIVF", "IVFSQ", "GPUIVFSQ"}; Loading