faiss 0.1.5 → 0.2.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +24 -0
- data/README.md +12 -0
- data/ext/faiss/ext.cpp +1 -1
- data/ext/faiss/extconf.rb +6 -2
- data/ext/faiss/index.cpp +114 -43
- data/ext/faiss/index_binary.cpp +24 -30
- data/ext/faiss/kmeans.cpp +20 -16
- data/ext/faiss/numo.hpp +867 -0
- data/ext/faiss/pca_matrix.cpp +13 -14
- data/ext/faiss/product_quantizer.cpp +23 -24
- data/ext/faiss/utils.cpp +10 -37
- data/ext/faiss/utils.h +2 -13
- data/lib/faiss.rb +0 -5
- data/lib/faiss/version.rb +1 -1
- data/vendor/faiss/faiss/AutoTune.cpp +292 -291
- data/vendor/faiss/faiss/AutoTune.h +55 -56
- data/vendor/faiss/faiss/Clustering.cpp +334 -195
- data/vendor/faiss/faiss/Clustering.h +88 -35
- data/vendor/faiss/faiss/IVFlib.cpp +171 -195
- data/vendor/faiss/faiss/IVFlib.h +48 -51
- data/vendor/faiss/faiss/Index.cpp +85 -103
- data/vendor/faiss/faiss/Index.h +54 -48
- data/vendor/faiss/faiss/Index2Layer.cpp +139 -164
- data/vendor/faiss/faiss/Index2Layer.h +22 -22
- data/vendor/faiss/faiss/IndexBinary.cpp +45 -37
- data/vendor/faiss/faiss/IndexBinary.h +140 -132
- data/vendor/faiss/faiss/IndexBinaryFlat.cpp +73 -53
- data/vendor/faiss/faiss/IndexBinaryFlat.h +29 -24
- data/vendor/faiss/faiss/IndexBinaryFromFloat.cpp +46 -43
- data/vendor/faiss/faiss/IndexBinaryFromFloat.h +16 -15
- data/vendor/faiss/faiss/IndexBinaryHNSW.cpp +215 -232
- data/vendor/faiss/faiss/IndexBinaryHNSW.h +25 -24
- data/vendor/faiss/faiss/IndexBinaryHash.cpp +182 -177
- data/vendor/faiss/faiss/IndexBinaryHash.h +41 -34
- data/vendor/faiss/faiss/IndexBinaryIVF.cpp +489 -461
- data/vendor/faiss/faiss/IndexBinaryIVF.h +97 -68
- data/vendor/faiss/faiss/IndexFlat.cpp +116 -147
- data/vendor/faiss/faiss/IndexFlat.h +35 -46
- data/vendor/faiss/faiss/IndexHNSW.cpp +372 -348
- data/vendor/faiss/faiss/IndexHNSW.h +57 -41
- data/vendor/faiss/faiss/IndexIVF.cpp +474 -454
- data/vendor/faiss/faiss/IndexIVF.h +146 -113
- data/vendor/faiss/faiss/IndexIVFFlat.cpp +248 -250
- data/vendor/faiss/faiss/IndexIVFFlat.h +48 -51
- data/vendor/faiss/faiss/IndexIVFPQ.cpp +457 -516
- data/vendor/faiss/faiss/IndexIVFPQ.h +74 -66
- data/vendor/faiss/faiss/IndexIVFPQFastScan.cpp +406 -372
- data/vendor/faiss/faiss/IndexIVFPQFastScan.h +82 -57
- data/vendor/faiss/faiss/IndexIVFPQR.cpp +104 -102
- data/vendor/faiss/faiss/IndexIVFPQR.h +33 -28
- data/vendor/faiss/faiss/IndexIVFSpectralHash.cpp +125 -133
- data/vendor/faiss/faiss/IndexIVFSpectralHash.h +19 -21
- data/vendor/faiss/faiss/IndexLSH.cpp +75 -96
- data/vendor/faiss/faiss/IndexLSH.h +21 -26
- data/vendor/faiss/faiss/IndexLattice.cpp +42 -56
- data/vendor/faiss/faiss/IndexLattice.h +11 -16
- data/vendor/faiss/faiss/IndexNNDescent.cpp +231 -0
- data/vendor/faiss/faiss/IndexNNDescent.h +72 -0
- data/vendor/faiss/faiss/IndexNSG.cpp +303 -0
- data/vendor/faiss/faiss/IndexNSG.h +85 -0
- data/vendor/faiss/faiss/IndexPQ.cpp +405 -464
- data/vendor/faiss/faiss/IndexPQ.h +64 -67
- data/vendor/faiss/faiss/IndexPQFastScan.cpp +143 -170
- data/vendor/faiss/faiss/IndexPQFastScan.h +46 -32
- data/vendor/faiss/faiss/IndexPreTransform.cpp +120 -150
- data/vendor/faiss/faiss/IndexPreTransform.h +33 -36
- data/vendor/faiss/faiss/IndexRefine.cpp +115 -131
- data/vendor/faiss/faiss/IndexRefine.h +22 -23
- data/vendor/faiss/faiss/IndexReplicas.cpp +147 -153
- data/vendor/faiss/faiss/IndexReplicas.h +62 -56
- data/vendor/faiss/faiss/IndexResidual.cpp +291 -0
- data/vendor/faiss/faiss/IndexResidual.h +152 -0
- data/vendor/faiss/faiss/IndexScalarQuantizer.cpp +120 -155
- data/vendor/faiss/faiss/IndexScalarQuantizer.h +41 -45
- data/vendor/faiss/faiss/IndexShards.cpp +256 -240
- data/vendor/faiss/faiss/IndexShards.h +85 -73
- data/vendor/faiss/faiss/MatrixStats.cpp +112 -97
- data/vendor/faiss/faiss/MatrixStats.h +7 -10
- data/vendor/faiss/faiss/MetaIndexes.cpp +135 -157
- data/vendor/faiss/faiss/MetaIndexes.h +40 -34
- data/vendor/faiss/faiss/MetricType.h +7 -7
- data/vendor/faiss/faiss/VectorTransform.cpp +652 -474
- data/vendor/faiss/faiss/VectorTransform.h +61 -89
- data/vendor/faiss/faiss/clone_index.cpp +77 -73
- data/vendor/faiss/faiss/clone_index.h +4 -9
- data/vendor/faiss/faiss/gpu/GpuAutoTune.cpp +33 -38
- data/vendor/faiss/faiss/gpu/GpuAutoTune.h +11 -9
- data/vendor/faiss/faiss/gpu/GpuCloner.cpp +197 -170
- data/vendor/faiss/faiss/gpu/GpuCloner.h +53 -35
- data/vendor/faiss/faiss/gpu/GpuClonerOptions.cpp +12 -14
- data/vendor/faiss/faiss/gpu/GpuClonerOptions.h +27 -25
- data/vendor/faiss/faiss/gpu/GpuDistance.h +116 -112
- data/vendor/faiss/faiss/gpu/GpuFaissAssert.h +1 -2
- data/vendor/faiss/faiss/gpu/GpuIndex.h +134 -137
- data/vendor/faiss/faiss/gpu/GpuIndexBinaryFlat.h +76 -73
- data/vendor/faiss/faiss/gpu/GpuIndexFlat.h +173 -162
- data/vendor/faiss/faiss/gpu/GpuIndexIVF.h +67 -64
- data/vendor/faiss/faiss/gpu/GpuIndexIVFFlat.h +89 -86
- data/vendor/faiss/faiss/gpu/GpuIndexIVFPQ.h +150 -141
- data/vendor/faiss/faiss/gpu/GpuIndexIVFScalarQuantizer.h +101 -103
- data/vendor/faiss/faiss/gpu/GpuIndicesOptions.h +17 -16
- data/vendor/faiss/faiss/gpu/GpuResources.cpp +116 -128
- data/vendor/faiss/faiss/gpu/GpuResources.h +182 -186
- data/vendor/faiss/faiss/gpu/StandardGpuResources.cpp +433 -422
- data/vendor/faiss/faiss/gpu/StandardGpuResources.h +131 -130
- data/vendor/faiss/faiss/gpu/impl/InterleavedCodes.cpp +468 -456
- data/vendor/faiss/faiss/gpu/impl/InterleavedCodes.h +25 -19
- data/vendor/faiss/faiss/gpu/impl/RemapIndices.cpp +22 -20
- data/vendor/faiss/faiss/gpu/impl/RemapIndices.h +9 -8
- data/vendor/faiss/faiss/gpu/perf/IndexWrapper-inl.h +39 -44
- data/vendor/faiss/faiss/gpu/perf/IndexWrapper.h +16 -14
- data/vendor/faiss/faiss/gpu/perf/PerfClustering.cpp +77 -71
- data/vendor/faiss/faiss/gpu/perf/PerfIVFPQAdd.cpp +109 -88
- data/vendor/faiss/faiss/gpu/perf/WriteIndex.cpp +75 -64
- data/vendor/faiss/faiss/gpu/test/TestCodePacking.cpp +230 -215
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexBinaryFlat.cpp +80 -86
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexFlat.cpp +284 -277
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFFlat.cpp +416 -416
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFPQ.cpp +611 -517
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFScalarQuantizer.cpp +166 -164
- data/vendor/faiss/faiss/gpu/test/TestGpuMemoryException.cpp +61 -53
- data/vendor/faiss/faiss/gpu/test/TestUtils.cpp +274 -238
- data/vendor/faiss/faiss/gpu/test/TestUtils.h +73 -57
- data/vendor/faiss/faiss/gpu/test/demo_ivfpq_indexing_gpu.cpp +47 -50
- data/vendor/faiss/faiss/gpu/utils/DeviceUtils.h +79 -72
- data/vendor/faiss/faiss/gpu/utils/StackDeviceMemory.cpp +140 -146
- data/vendor/faiss/faiss/gpu/utils/StackDeviceMemory.h +69 -71
- data/vendor/faiss/faiss/gpu/utils/StaticUtils.h +21 -16
- data/vendor/faiss/faiss/gpu/utils/Timer.cpp +25 -29
- data/vendor/faiss/faiss/gpu/utils/Timer.h +30 -29
- data/vendor/faiss/faiss/impl/AdditiveQuantizer.cpp +270 -0
- data/vendor/faiss/faiss/impl/AdditiveQuantizer.h +115 -0
- data/vendor/faiss/faiss/impl/AuxIndexStructures.cpp +90 -120
- data/vendor/faiss/faiss/impl/AuxIndexStructures.h +81 -65
- data/vendor/faiss/faiss/impl/FaissAssert.h +73 -58
- data/vendor/faiss/faiss/impl/FaissException.cpp +56 -48
- data/vendor/faiss/faiss/impl/FaissException.h +41 -29
- data/vendor/faiss/faiss/impl/HNSW.cpp +595 -611
- data/vendor/faiss/faiss/impl/HNSW.h +179 -200
- data/vendor/faiss/faiss/impl/LocalSearchQuantizer.cpp +672 -0
- data/vendor/faiss/faiss/impl/LocalSearchQuantizer.h +172 -0
- data/vendor/faiss/faiss/impl/NNDescent.cpp +487 -0
- data/vendor/faiss/faiss/impl/NNDescent.h +154 -0
- data/vendor/faiss/faiss/impl/NSG.cpp +682 -0
- data/vendor/faiss/faiss/impl/NSG.h +199 -0
- data/vendor/faiss/faiss/impl/PolysemousTraining.cpp +484 -454
- data/vendor/faiss/faiss/impl/PolysemousTraining.h +52 -55
- data/vendor/faiss/faiss/impl/ProductQuantizer-inl.h +26 -47
- data/vendor/faiss/faiss/impl/ProductQuantizer.cpp +469 -459
- data/vendor/faiss/faiss/impl/ProductQuantizer.h +76 -87
- data/vendor/faiss/faiss/impl/ResidualQuantizer.cpp +448 -0
- data/vendor/faiss/faiss/impl/ResidualQuantizer.h +130 -0
- data/vendor/faiss/faiss/impl/ResultHandler.h +96 -132
- data/vendor/faiss/faiss/impl/ScalarQuantizer.cpp +648 -701
- data/vendor/faiss/faiss/impl/ScalarQuantizer.h +48 -46
- data/vendor/faiss/faiss/impl/ThreadedIndex-inl.h +129 -131
- data/vendor/faiss/faiss/impl/ThreadedIndex.h +61 -55
- data/vendor/faiss/faiss/impl/index_read.cpp +547 -479
- data/vendor/faiss/faiss/impl/index_write.cpp +497 -407
- data/vendor/faiss/faiss/impl/io.cpp +75 -94
- data/vendor/faiss/faiss/impl/io.h +31 -41
- data/vendor/faiss/faiss/impl/io_macros.h +40 -29
- data/vendor/faiss/faiss/impl/lattice_Zn.cpp +137 -186
- data/vendor/faiss/faiss/impl/lattice_Zn.h +40 -51
- data/vendor/faiss/faiss/impl/platform_macros.h +29 -8
- data/vendor/faiss/faiss/impl/pq4_fast_scan.cpp +77 -124
- data/vendor/faiss/faiss/impl/pq4_fast_scan.h +39 -48
- data/vendor/faiss/faiss/impl/pq4_fast_scan_search_1.cpp +41 -52
- data/vendor/faiss/faiss/impl/pq4_fast_scan_search_qbs.cpp +80 -117
- data/vendor/faiss/faiss/impl/simd_result_handlers.h +109 -137
- data/vendor/faiss/faiss/index_factory.cpp +269 -218
- data/vendor/faiss/faiss/index_factory.h +6 -7
- data/vendor/faiss/faiss/index_io.h +23 -26
- data/vendor/faiss/faiss/invlists/BlockInvertedLists.cpp +67 -75
- data/vendor/faiss/faiss/invlists/BlockInvertedLists.h +22 -24
- data/vendor/faiss/faiss/invlists/DirectMap.cpp +96 -112
- data/vendor/faiss/faiss/invlists/DirectMap.h +29 -33
- data/vendor/faiss/faiss/invlists/InvertedLists.cpp +307 -364
- data/vendor/faiss/faiss/invlists/InvertedLists.h +151 -151
- data/vendor/faiss/faiss/invlists/InvertedListsIOHook.cpp +29 -34
- data/vendor/faiss/faiss/invlists/InvertedListsIOHook.h +17 -18
- data/vendor/faiss/faiss/invlists/OnDiskInvertedLists.cpp +257 -293
- data/vendor/faiss/faiss/invlists/OnDiskInvertedLists.h +50 -45
- data/vendor/faiss/faiss/python/python_callbacks.cpp +23 -26
- data/vendor/faiss/faiss/python/python_callbacks.h +9 -16
- data/vendor/faiss/faiss/utils/AlignedTable.h +79 -44
- data/vendor/faiss/faiss/utils/Heap.cpp +40 -48
- data/vendor/faiss/faiss/utils/Heap.h +186 -209
- data/vendor/faiss/faiss/utils/WorkerThread.cpp +67 -76
- data/vendor/faiss/faiss/utils/WorkerThread.h +32 -33
- data/vendor/faiss/faiss/utils/distances.cpp +301 -310
- data/vendor/faiss/faiss/utils/distances.h +133 -118
- data/vendor/faiss/faiss/utils/distances_simd.cpp +456 -516
- data/vendor/faiss/faiss/utils/extra_distances-inl.h +117 -0
- data/vendor/faiss/faiss/utils/extra_distances.cpp +113 -232
- data/vendor/faiss/faiss/utils/extra_distances.h +30 -29
- data/vendor/faiss/faiss/utils/hamming-inl.h +260 -209
- data/vendor/faiss/faiss/utils/hamming.cpp +375 -469
- data/vendor/faiss/faiss/utils/hamming.h +62 -85
- data/vendor/faiss/faiss/utils/ordered_key_value.h +16 -18
- data/vendor/faiss/faiss/utils/partitioning.cpp +393 -318
- data/vendor/faiss/faiss/utils/partitioning.h +26 -21
- data/vendor/faiss/faiss/utils/quantize_lut.cpp +78 -66
- data/vendor/faiss/faiss/utils/quantize_lut.h +22 -20
- data/vendor/faiss/faiss/utils/random.cpp +39 -63
- data/vendor/faiss/faiss/utils/random.h +13 -16
- data/vendor/faiss/faiss/utils/simdlib.h +4 -2
- data/vendor/faiss/faiss/utils/simdlib_avx2.h +88 -85
- data/vendor/faiss/faiss/utils/simdlib_emulated.h +226 -165
- data/vendor/faiss/faiss/utils/simdlib_neon.h +832 -0
- data/vendor/faiss/faiss/utils/utils.cpp +304 -287
- data/vendor/faiss/faiss/utils/utils.h +53 -48
- metadata +24 -10
- data/lib/faiss/index.rb +0 -20
- data/lib/faiss/index_binary.rb +0 -20
- data/lib/faiss/kmeans.rb +0 -15
- data/lib/faiss/pca_matrix.rb +0 -15
- data/lib/faiss/product_quantizer.rb +0 -22
@@ -15,15 +15,15 @@
|
|
15
15
|
|
16
16
|
#include <stdint.h>
|
17
17
|
|
18
|
-
#include <
|
19
|
-
#include <unordered_set>
|
18
|
+
#include <cstring>
|
20
19
|
#include <memory>
|
21
20
|
#include <mutex>
|
21
|
+
#include <unordered_set>
|
22
|
+
#include <vector>
|
22
23
|
|
23
24
|
#include <faiss/Index.h>
|
24
25
|
#include <faiss/impl/platform_macros.h>
|
25
26
|
|
26
|
-
|
27
27
|
namespace faiss {
|
28
28
|
|
29
29
|
/** The objective is to have a simple result structure while
|
@@ -31,42 +31,39 @@ namespace faiss {
|
|
31
31
|
* do_allocation can be overloaded to allocate the result tables in
|
32
32
|
* the matrix type of a scripting language like Lua or Python. */
|
33
33
|
struct RangeSearchResult {
|
34
|
-
size_t nq;
|
35
|
-
size_t
|
34
|
+
size_t nq; ///< nb of queries
|
35
|
+
size_t* lims; ///< size (nq + 1)
|
36
36
|
|
37
37
|
typedef Index::idx_t idx_t;
|
38
38
|
|
39
|
-
idx_t
|
40
|
-
float
|
39
|
+
idx_t* labels; ///< result for query i is labels[lims[i]:lims[i+1]]
|
40
|
+
float* distances; ///< corresponding distances (not sorted)
|
41
41
|
|
42
42
|
size_t buffer_size; ///< size of the result buffers used
|
43
43
|
|
44
44
|
/// lims must be allocated on input to range_search.
|
45
|
-
explicit RangeSearchResult
|
45
|
+
explicit RangeSearchResult(idx_t nq, bool alloc_lims = true);
|
46
46
|
|
47
47
|
/// called when lims contains the nb of elements result entries
|
48
48
|
/// for each query
|
49
49
|
|
50
|
-
virtual void do_allocation
|
50
|
+
virtual void do_allocation();
|
51
51
|
|
52
|
-
virtual ~RangeSearchResult
|
52
|
+
virtual ~RangeSearchResult();
|
53
53
|
};
|
54
54
|
|
55
|
-
|
56
55
|
/** Encapsulates a set of ids to remove. */
|
57
56
|
struct IDSelector {
|
58
57
|
typedef Index::idx_t idx_t;
|
59
|
-
virtual bool is_member
|
58
|
+
virtual bool is_member(idx_t id) const = 0;
|
60
59
|
virtual ~IDSelector() {}
|
61
60
|
};
|
62
61
|
|
63
|
-
|
64
|
-
|
65
62
|
/** remove ids between [imni, imax) */
|
66
|
-
struct IDSelectorRange: IDSelector {
|
63
|
+
struct IDSelectorRange : IDSelector {
|
67
64
|
idx_t imin, imax;
|
68
65
|
|
69
|
-
IDSelectorRange
|
66
|
+
IDSelectorRange(idx_t imin, idx_t imax);
|
70
67
|
bool is_member(idx_t id) const override;
|
71
68
|
~IDSelectorRange() override {}
|
72
69
|
};
|
@@ -76,11 +73,11 @@ struct IDSelectorRange: IDSelector {
|
|
76
73
|
* this is inefficient in most cases, except for IndexIVF with
|
77
74
|
* maintain_direct_map
|
78
75
|
*/
|
79
|
-
struct IDSelectorArray: IDSelector {
|
76
|
+
struct IDSelectorArray : IDSelector {
|
80
77
|
size_t n;
|
81
|
-
const idx_t
|
78
|
+
const idx_t* ids;
|
82
79
|
|
83
|
-
IDSelectorArray
|
80
|
+
IDSelectorArray(size_t n, const idx_t* ids);
|
84
81
|
bool is_member(idx_t id) const override;
|
85
82
|
~IDSelectorArray() override {}
|
86
83
|
};
|
@@ -91,8 +88,7 @@ struct IDSelectorArray: IDSelector {
|
|
91
88
|
* unordered_set are just the least significant bits of the id. This
|
92
89
|
* works fine for random ids or ids in sequences but will produce many
|
93
90
|
* hash collisions if lsb's are always the same */
|
94
|
-
struct IDSelectorBatch: IDSelector {
|
95
|
-
|
91
|
+
struct IDSelectorBatch : IDSelector {
|
96
92
|
std::unordered_set<idx_t> set;
|
97
93
|
|
98
94
|
typedef unsigned char uint8_t;
|
@@ -100,7 +96,7 @@ struct IDSelectorBatch: IDSelector {
|
|
100
96
|
int nbits;
|
101
97
|
idx_t mask;
|
102
98
|
|
103
|
-
IDSelectorBatch
|
99
|
+
IDSelectorBatch(size_t n, const idx_t* indices);
|
104
100
|
bool is_member(idx_t id) const override;
|
105
101
|
~IDSelectorBatch() override {}
|
106
102
|
};
|
@@ -124,28 +120,26 @@ struct BufferList {
|
|
124
120
|
size_t buffer_size;
|
125
121
|
|
126
122
|
struct Buffer {
|
127
|
-
idx_t
|
128
|
-
float
|
123
|
+
idx_t* ids;
|
124
|
+
float* dis;
|
129
125
|
};
|
130
126
|
|
131
127
|
std::vector<Buffer> buffers;
|
132
128
|
size_t wp; ///< write pointer in the last buffer.
|
133
129
|
|
134
|
-
explicit BufferList
|
130
|
+
explicit BufferList(size_t buffer_size);
|
135
131
|
|
136
|
-
~BufferList
|
132
|
+
~BufferList();
|
137
133
|
|
138
134
|
/// create a new buffer
|
139
|
-
void append_buffer
|
135
|
+
void append_buffer();
|
140
136
|
|
141
137
|
/// add one result, possibly appending a new buffer if needed
|
142
|
-
void add
|
138
|
+
void add(idx_t id, float dis);
|
143
139
|
|
144
140
|
/// copy elemnts ofs:ofs+n-1 seen as linear data in the buffers to
|
145
141
|
/// tables dest_ids, dest_dis
|
146
|
-
void copy_range
|
147
|
-
idx_t * dest_ids, float *dest_dis);
|
148
|
-
|
142
|
+
void copy_range(size_t ofs, size_t n, idx_t* dest_ids, float* dest_dis);
|
149
143
|
};
|
150
144
|
|
151
145
|
struct RangeSearchPartialResult;
|
@@ -153,46 +147,45 @@ struct RangeSearchPartialResult;
|
|
153
147
|
/// result structure for a single query
|
154
148
|
struct RangeQueryResult {
|
155
149
|
using idx_t = Index::idx_t;
|
156
|
-
idx_t qno;
|
157
|
-
size_t nres;
|
158
|
-
RangeSearchPartialResult
|
150
|
+
idx_t qno; //< id of the query
|
151
|
+
size_t nres; //< nb of results for this query
|
152
|
+
RangeSearchPartialResult* pres;
|
159
153
|
|
160
154
|
/// called by search function to report a new result
|
161
|
-
void add
|
155
|
+
void add(float dis, idx_t id);
|
162
156
|
};
|
163
157
|
|
164
158
|
/// the entries in the buffers are split per query
|
165
|
-
struct RangeSearchPartialResult: BufferList {
|
166
|
-
RangeSearchResult
|
159
|
+
struct RangeSearchPartialResult : BufferList {
|
160
|
+
RangeSearchResult* res;
|
167
161
|
|
168
162
|
/// eventually the result will be stored in res_in
|
169
|
-
explicit RangeSearchPartialResult
|
163
|
+
explicit RangeSearchPartialResult(RangeSearchResult* res_in);
|
170
164
|
|
171
165
|
/// query ids + nb of results per query.
|
172
166
|
std::vector<RangeQueryResult> queries;
|
173
167
|
|
174
168
|
/// begin a new result
|
175
|
-
RangeQueryResult
|
169
|
+
RangeQueryResult& new_result(idx_t qno);
|
176
170
|
|
177
171
|
/*****************************************
|
178
172
|
* functions used at the end of the search to merge the result
|
179
173
|
* lists */
|
180
|
-
void finalize
|
174
|
+
void finalize();
|
181
175
|
|
182
176
|
/// called by range_search before do_allocation
|
183
|
-
void set_lims
|
177
|
+
void set_lims();
|
184
178
|
|
185
179
|
/// called by range_search after do_allocation
|
186
|
-
void copy_result
|
180
|
+
void copy_result(bool incremental = false);
|
187
181
|
|
188
182
|
/// merge a set of PartialResult's into one RangeSearchResult
|
189
183
|
/// on ouptut the partialresults are empty!
|
190
|
-
static void merge
|
191
|
-
|
192
|
-
|
184
|
+
static void merge(
|
185
|
+
std::vector<RangeSearchPartialResult*>& partial_results,
|
186
|
+
bool do_delete = true);
|
193
187
|
};
|
194
188
|
|
195
|
-
|
196
189
|
/***********************************************************
|
197
190
|
* The distance computer maintains a current query and computes
|
198
191
|
* distances to elements in an index that supports random access.
|
@@ -202,19 +195,19 @@ struct RangeSearchPartialResult: BufferList {
|
|
202
195
|
* instantiate one from each thread if needed.
|
203
196
|
***********************************************************/
|
204
197
|
struct DistanceComputer {
|
205
|
-
|
198
|
+
using idx_t = Index::idx_t;
|
206
199
|
|
207
|
-
|
208
|
-
|
209
|
-
|
200
|
+
/// called before computing distances. Pointer x should remain valid
|
201
|
+
/// while operator () is called
|
202
|
+
virtual void set_query(const float* x) = 0;
|
210
203
|
|
211
|
-
|
212
|
-
|
204
|
+
/// compute distance of vector i to current query
|
205
|
+
virtual float operator()(idx_t i) = 0;
|
213
206
|
|
214
|
-
|
215
|
-
|
207
|
+
/// compute distance between two stored vectors
|
208
|
+
virtual float symmetric_dis(idx_t i, idx_t j) = 0;
|
216
209
|
|
217
|
-
|
210
|
+
virtual ~DistanceComputer() {}
|
218
211
|
};
|
219
212
|
|
220
213
|
/***********************************************************
|
@@ -222,7 +215,7 @@ struct DistanceComputer {
|
|
222
215
|
***********************************************************/
|
223
216
|
|
224
217
|
struct FAISS_API InterruptCallback {
|
225
|
-
virtual bool want_interrupt
|
218
|
+
virtual bool want_interrupt() = 0;
|
226
219
|
virtual ~InterruptCallback() {}
|
227
220
|
|
228
221
|
// lock that protects concurrent calls to is_interrupted
|
@@ -230,7 +223,7 @@ struct FAISS_API InterruptCallback {
|
|
230
223
|
|
231
224
|
static std::unique_ptr<InterruptCallback> instance;
|
232
225
|
|
233
|
-
static void clear_instance
|
226
|
+
static void clear_instance();
|
234
227
|
|
235
228
|
/** check if:
|
236
229
|
* - an interrupt callback is set
|
@@ -238,23 +231,46 @@ struct FAISS_API InterruptCallback {
|
|
238
231
|
* if this is the case, then throw an exception. Should not be called
|
239
232
|
* from multiple threads.
|
240
233
|
*/
|
241
|
-
static void check
|
234
|
+
static void check();
|
242
235
|
|
243
236
|
/// same as check() but return true if is interrupted instead of
|
244
237
|
/// throwing. Can be called from multiple threads.
|
245
|
-
static bool is_interrupted
|
238
|
+
static bool is_interrupted();
|
246
239
|
|
247
240
|
/** assuming each iteration takes a certain number of flops, what
|
248
241
|
* is a reasonable interval to check for interrupts?
|
249
242
|
*/
|
250
|
-
static size_t get_period_hint
|
251
|
-
|
243
|
+
static size_t get_period_hint(size_t flops);
|
252
244
|
};
|
253
245
|
|
246
|
+
/// set implementation optimized for fast access.
|
247
|
+
struct VisitedTable {
|
248
|
+
std::vector<uint8_t> visited;
|
249
|
+
int visno;
|
250
|
+
|
251
|
+
explicit VisitedTable(int size) : visited(size), visno(1) {}
|
252
|
+
|
253
|
+
/// set flag #no to true
|
254
|
+
void set(int no) {
|
255
|
+
visited[no] = visno;
|
256
|
+
}
|
257
|
+
|
258
|
+
/// get flag #no
|
259
|
+
bool get(int no) const {
|
260
|
+
return visited[no] == visno;
|
261
|
+
}
|
262
|
+
|
263
|
+
/// reset all flags to false
|
264
|
+
void advance() {
|
265
|
+
visno++;
|
266
|
+
if (visno == 250) {
|
267
|
+
// 250 rather than 255 because sometimes we use visno and visno+1
|
268
|
+
memset(visited.data(), 0, sizeof(visited[0]) * visited.size());
|
269
|
+
visno = 1;
|
270
|
+
}
|
271
|
+
}
|
272
|
+
};
|
254
273
|
|
255
|
-
|
256
|
-
}; // namespace faiss
|
257
|
-
|
258
|
-
|
274
|
+
} // namespace faiss
|
259
275
|
|
260
276
|
#endif
|
@@ -12,85 +12,100 @@
|
|
12
12
|
|
13
13
|
#include <faiss/impl/FaissException.h>
|
14
14
|
#include <faiss/impl/platform_macros.h>
|
15
|
-
#include <cstdlib>
|
16
15
|
#include <cstdio>
|
16
|
+
#include <cstdlib>
|
17
17
|
#include <string>
|
18
18
|
|
19
19
|
///
|
20
20
|
/// Assertions
|
21
21
|
///
|
22
22
|
|
23
|
-
#define FAISS_ASSERT(X)
|
24
|
-
|
25
|
-
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
|
30
|
-
|
31
|
-
|
23
|
+
#define FAISS_ASSERT(X) \
|
24
|
+
do { \
|
25
|
+
if (!(X)) { \
|
26
|
+
fprintf(stderr, \
|
27
|
+
"Faiss assertion '%s' failed in %s " \
|
28
|
+
"at %s:%d\n", \
|
29
|
+
#X, \
|
30
|
+
__PRETTY_FUNCTION__, \
|
31
|
+
__FILE__, \
|
32
|
+
__LINE__); \
|
33
|
+
abort(); \
|
34
|
+
} \
|
35
|
+
} while (false)
|
32
36
|
|
33
|
-
#define FAISS_ASSERT_MSG(X, MSG)
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
|
40
|
-
|
41
|
-
|
37
|
+
#define FAISS_ASSERT_MSG(X, MSG) \
|
38
|
+
do { \
|
39
|
+
if (!(X)) { \
|
40
|
+
fprintf(stderr, \
|
41
|
+
"Faiss assertion '%s' failed in %s " \
|
42
|
+
"at %s:%d; details: " MSG "\n", \
|
43
|
+
#X, \
|
44
|
+
__PRETTY_FUNCTION__, \
|
45
|
+
__FILE__, \
|
46
|
+
__LINE__); \
|
47
|
+
abort(); \
|
48
|
+
} \
|
49
|
+
} while (false)
|
42
50
|
|
43
|
-
#define FAISS_ASSERT_FMT(X, FMT, ...)
|
44
|
-
|
45
|
-
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
|
51
|
-
|
51
|
+
#define FAISS_ASSERT_FMT(X, FMT, ...) \
|
52
|
+
do { \
|
53
|
+
if (!(X)) { \
|
54
|
+
fprintf(stderr, \
|
55
|
+
"Faiss assertion '%s' failed in %s " \
|
56
|
+
"at %s:%d; details: " FMT "\n", \
|
57
|
+
#X, \
|
58
|
+
__PRETTY_FUNCTION__, \
|
59
|
+
__FILE__, \
|
60
|
+
__LINE__, \
|
61
|
+
__VA_ARGS__); \
|
62
|
+
abort(); \
|
63
|
+
} \
|
64
|
+
} while (false)
|
52
65
|
|
53
66
|
///
|
54
67
|
/// Exceptions for returning user errors
|
55
68
|
///
|
56
69
|
|
57
|
-
#define FAISS_THROW_MSG(MSG)
|
58
|
-
|
59
|
-
|
60
|
-
|
70
|
+
#define FAISS_THROW_MSG(MSG) \
|
71
|
+
do { \
|
72
|
+
throw faiss::FaissException( \
|
73
|
+
MSG, __PRETTY_FUNCTION__, __FILE__, __LINE__); \
|
74
|
+
} while (false)
|
61
75
|
|
62
|
-
#define FAISS_THROW_FMT(FMT, ...)
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
76
|
+
#define FAISS_THROW_FMT(FMT, ...) \
|
77
|
+
do { \
|
78
|
+
std::string __s; \
|
79
|
+
int __size = snprintf(nullptr, 0, FMT, __VA_ARGS__); \
|
80
|
+
__s.resize(__size + 1); \
|
81
|
+
snprintf(&__s[0], __s.size(), FMT, __VA_ARGS__); \
|
82
|
+
throw faiss::FaissException( \
|
83
|
+
__s, __PRETTY_FUNCTION__, __FILE__, __LINE__); \
|
84
|
+
} while (false)
|
70
85
|
|
71
86
|
///
|
72
87
|
/// Exceptions thrown upon a conditional failure
|
73
88
|
///
|
74
89
|
|
75
|
-
#define FAISS_THROW_IF_NOT(X)
|
76
|
-
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
90
|
+
#define FAISS_THROW_IF_NOT(X) \
|
91
|
+
do { \
|
92
|
+
if (!(X)) { \
|
93
|
+
FAISS_THROW_FMT("Error: '%s' failed", #X); \
|
94
|
+
} \
|
95
|
+
} while (false)
|
81
96
|
|
82
|
-
#define FAISS_THROW_IF_NOT_MSG(X, MSG)
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
|
97
|
+
#define FAISS_THROW_IF_NOT_MSG(X, MSG) \
|
98
|
+
do { \
|
99
|
+
if (!(X)) { \
|
100
|
+
FAISS_THROW_FMT("Error: '%s' failed: " MSG, #X); \
|
101
|
+
} \
|
102
|
+
} while (false)
|
88
103
|
|
89
|
-
#define FAISS_THROW_IF_NOT_FMT(X, FMT, ...)
|
90
|
-
|
91
|
-
|
92
|
-
|
93
|
-
|
94
|
-
|
104
|
+
#define FAISS_THROW_IF_NOT_FMT(X, FMT, ...) \
|
105
|
+
do { \
|
106
|
+
if (!(X)) { \
|
107
|
+
FAISS_THROW_FMT("Error: '%s' failed: " FMT, #X, __VA_ARGS__); \
|
108
|
+
} \
|
109
|
+
} while (false)
|
95
110
|
|
96
111
|
#endif
|