storage: Update usearch and simsimd (pingcap#270)

Signed-off-by: Wish <[email protected]>
Lloyd-Pottiger · Sep 20, 2024 · 34348e0 · 34348e0
1 parent c9ed0fd
commit 34348e0
Show file tree

Hide file tree

Showing 8 changed files with 40 additions and 21 deletions.
diff --git a/contrib/simsimd b/contrib/simsimd
diff --git a/contrib/usearch b/contrib/usearch
diff --git a/dbms/src/Storages/DeltaMerge/Index/VectorIndexHNSW/Index.cpp b/dbms/src/Storages/DeltaMerge/Index/VectorIndexHNSW/Index.cpp
@@ -84,10 +84,7 @@ void VectorIndexHNSWBuilder::addBlock(
 
     const auto * del_mark_data = (!del_mark) ? nullptr : &(del_mark->getData());
 
-    if (!index.reserve(unum::usearch::ceil2(index.size() + column.size())))
-    {
-        throw Exception(ErrorCodes::CANNOT_ALLOCATE_MEMORY, "Could not reserve memory for HNSW index");
-    }
+    index.reserve(unum::usearch::ceil2(index.size() + column.size()));
 
     Stopwatch w;
     SCOPE_EXIT({ total_duration += w.elapsedSeconds(); });
@@ -170,6 +167,7 @@ VectorIndexViewerPtr VectorIndexHNSWViewer::view(const dtpb::VectorIndexFileProp
     SCOPE_EXIT({ GET_METRIC(tiflash_vector_index_duration, type_view).Observe(w.elapsedSeconds()); });
 
     auto vi = std::make_shared<VectorIndexHNSWViewer>(file_props);
+
     vi->index = USearchImplType::make(
         unum::usearch::metric_punned_t( //
             file_props.dimensions(),
@@ -179,6 +177,10 @@ VectorIndexViewerPtr VectorIndexHNSWViewer::view(const dtpb::VectorIndexFileProp
             unum::usearch::default_expansion_add(),
             16 /* default is 64 */));
 
+    // Currently may have a lot of threads querying concurrently
+    auto limit = unum::usearch::index_limits_t(0, /* threads */ std::thread::hardware_concurrency() * 10);
+    vi->index.reserve(limit);
+
     auto result = vi->index.view(unum::usearch::memory_mapped_file_t(path.data()));
     RUNTIME_CHECK_MSG(result, "Failed to load vector index: {}", result.error.what());
 

diff --git a/dbms/src/Storages/DeltaMerge/tests/gtest_dm_delta_merge_store_vector_index.cpp b/dbms/src/Storages/DeltaMerge/tests/gtest_dm_delta_merge_store_vector_index.cpp
@@ -262,7 +262,7 @@ try
 
         auto filter = std::make_shared<PushDownFilter>(wrapWithANNQueryInfo(nullptr, ann_query_info));
 
-        read(left_segment_range, filter, createVecFloat32Column<Array>({})); // FIXME: should be 63.0
+        read(left_segment_range, filter, createVecFloat32Column<Array>({{63.0}}));
     }
 
     // merge segment

diff --git a/dbms/src/TiDB/Decode/Vector.cpp b/dbms/src/TiDB/Decode/Vector.cpp
@@ -16,7 +16,7 @@
 #include <IO/Buffer/WriteBufferFromString.h>
 #include <IO/WriteHelpers.h>
 #include <TiDB/Decode/Vector.h>
-#include <VectorSearch/simdsimd-internals.h>
+#include <VectorSearch/SimSIMD.h>
 
 #include <compare>
 

diff --git a/dbms/src/VectorSearch/DistanceSIMDFeatures.cpp b/dbms/src/VectorSearch/DistanceSIMDFeatures.cpp
@@ -16,7 +16,7 @@
 // polluting all compile units.
 
 #include <VectorSearch/DistanceSIMDFeatures.h>
-#include <VectorSearch/simdsimd-internals.h>
+#include <VectorSearch/SimSIMD.h>
 
 namespace simsimd_details
 {
@@ -52,21 +52,39 @@ std::vector<std::string> VectorDistanceSIMDFeatures::get()
     simsimd_capability_t cap_l2 = simsimd_details::actual_capability(simsimd_datatype_f32_k, simsimd_metric_l2sq_k);
     simsimd_capability_t cap_cos = simsimd_details::actual_capability(simsimd_datatype_f32_k, simsimd_metric_cos_k);
 
-    auto cap_to_string = [](simsimd_capability_t cap) -> std::string {
-        switch (cap)
+    auto cap_to_string = [](simsimd_capability_t isa_kind) -> std::string {
+        switch (isa_kind)
         {
-        case simsimd_cap_sve2_k:
-            return "sve2";
-        case simsimd_cap_sve_k:
-            return "sve";
+        case simsimd_cap_serial_k:
+            return "serial";
         case simsimd_cap_neon_k:
             return "neon";
-        case simsimd_cap_skylake_k:
-            return "skylake";
+        case simsimd_cap_neon_i8_k:
+            return "neon_i8";
+        case simsimd_cap_neon_f16_k:
+            return "neon_f16";
+        case simsimd_cap_neon_bf16_k:
+            return "neon_bf16";
+        case simsimd_cap_sve_k:
+            return "sve";
+        case simsimd_cap_sve_i8_k:
+            return "sve_i8";
+        case simsimd_cap_sve_f16_k:
+            return "sve_f16";
+        case simsimd_cap_sve_bf16_k:
+            return "sve_bf16";
         case simsimd_cap_haswell_k:
             return "haswell";
+        case simsimd_cap_skylake_k:
+            return "skylake";
+        case simsimd_cap_ice_k:
+            return "ice";
+        case simsimd_cap_genoa_k:
+            return "genoa";
+        case simsimd_cap_sapphire_k:
+            return "sapphire";
         default:
-            return "serial";
+            return "unknown";
         }
     };
 

diff --git a/dbms/src/VectorSearch/simdsimd-internals.h → dbms/src/VectorSearch/SimSIMD.h b/dbms/src/VectorSearch/simdsimd-internals.h → dbms/src/VectorSearch/SimSIMD.h
@@ -29,12 +29,11 @@
 #define SIMSIMD_TARGET_SVE 1
 #define SIMSIMD_TARGET_HASWELL 1
 #define SIMSIMD_TARGET_SKYLAKE 1
-#define SIMSIMD_TARGET_ICE 0
+#define SIMSIMD_TARGET_ICE 1
 #define SIMSIMD_TARGET_GENOA 0
 #define SIMSIMD_TARGET_SAPPHIRE 0
 #include <simsimd/simsimd.h>
 
-
 namespace simsimd_details
 {
 

diff --git a/dbms/src/VectorSearch/USearch.h b/dbms/src/VectorSearch/USearch.h
@@ -26,7 +26,7 @@
 #define SIMSIMD_TARGET_SVE 1
 #define SIMSIMD_TARGET_HASWELL 1
 #define SIMSIMD_TARGET_SKYLAKE 1
-#define SIMSIMD_TARGET_ICE 0
+#define SIMSIMD_TARGET_ICE 1
 #define SIMSIMD_TARGET_GENOA 0
 #define SIMSIMD_TARGET_SAPPHIRE 0
+0 −107		.github/workflows/package.json
+34 −3		.github/workflows/prerelease.yml
+38 −31		.github/workflows/release.yml
+0 −9		.github/workflows/update_version.sh
+15 −0		.vscode/launch.json
+1 −1		CMakeLists.txt
+24 −4		CONTRIBUTING.md
+1 −1		Cargo.lock
+1 −1		Cargo.toml
+8 −8		README.md
+1 −1		VERSION
+6 −0		c/lib.c
+6 −2		cpp/bench.cxx
+6 −0		cpp/test.c
+7 −5		include/simsimd/binary.h
+42 −26		include/simsimd/dot.h
+295 −65		include/simsimd/simsimd.h
+199 −30		include/simsimd/spatial.h
+43 −35		include/simsimd/types.h
+1 −1		package.json
+2 −2		python/bench.py
+56 −29		python/lib.c
+131 −0		python/simsimd/__init__.pyi
+0 −0		python/simsimd/py.typed
+6 −1		python/test.py
+45 −15		rust/lib.rs
+7 −2		setup.py
+39 −7		.github/workflows/prerelease.yml
+37 −19		.github/workflows/release.yml
+0 −19		.github/workflows/update_version.sh
+6 −4		.gitignore
+13 −0		.vscode/launch.json
+11 −1		.vscode/settings.json
+2 −2		.vscode/tasks.json
+14 −14		BENCHMARKS.md
+1 −1		CITATION.cff
+8 −8		CMakeLists.txt
+59 −27		CONTRIBUTING.md
+15 −15		Cargo.lock
+1 −1		Cargo.toml
+1 −0		MANIFEST.in
+98 −28		README.md
+1 −1		VERSION
+11 −12		binding.gyp
+5 −2		build.gradle
+12 −1		build.rs
+3 −0		c/README.md
+50 −13		c/lib.cpp
+60 −55		c/test.c
+23 −4		c/usearch.h
+1 −1		conanfile.py
+34 −13		cpp/README.md
+99 −47		cpp/bench.cpp
+649 −176		cpp/test.cpp
+1 −1		csharp/nuget/nuget-package.props
+10 −10		csharp/src/Cloud.Unum.USearch.Tests/USearchIndexTests.cs
+6 −44		csharp/src/Cloud.Unum.USearch/NativeMethods.cs
+295 −60		csharp/src/Cloud.Unum.USearch/USearchIndex.cs
+117 −8		csharp/src/Cloud.Unum.USearch/USearchTypes.cs
+1 −1		docs/index.rst
+0 −5		docs/java/index.rst
+0 −5		docs/javascript/index.rst
+35 −14		golang/README.md
+32 −24		golang/lib.go
+857 −264		include/usearch/index.hpp
+412 −156		include/usearch/index_dense.hpp
+84 −39		include/usearch/index_plugins.hpp
+56 −5		java/README.md
+126 −36		java/cloud/unum/usearch/Index.java
+3 −0		java/cloud/unum/usearch/NativeUtils.java
+38 −2		java/cloud/unum/usearch/cloud_unum_usearch_Index.cpp
+16 −0		java/cloud/unum/usearch/cloud_unum_usearch_Index.h
+57 −3		java/test/IndexTest.java
+3 −2		javascript/tsconfig-esm.json
+8 −7		javascript/usearch.ts
+6 −2		objc/USearchObjective.mm
+0 −10		package-ci.json
+2 −2		package.json
+12 −7		python/lib.cpp
+8 −0		python/scripts/test_index.py
+119 −0		python/scripts/test_jit.py
+0 −0		python/usearch/py.typed
+42 −29		rust/lib.cpp
+1 −1		rust/lib.hpp
+53 −5		rust/lib.rs
+10 −0		setup.py
+1 −1		simsimd
+30 −0		swift/Test.swift
+1 −1		wasmer.toml
+44 −4		wolfram/CMakeLists.txt
+30 −19		wolfram/lib.cpp