1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
|
/*
* Copyright (c) Meta Platforms, Inc. and affiliates.
*
* This source code is licensed under the MIT license found in the
* LICENSE file in the root directory of this source tree.
*/
#include <gtest/gtest.h>
#include <cstddef>
#include <cstdint>
#include <random>
#include <vector>
#include <faiss/utils/distances.h>
// reference implementations
void fvec_inner_products_ny_ref(
float* ip,
const float* x,
const float* y,
size_t d,
size_t ny) {
for (size_t i = 0; i < ny; i++) {
ip[i] = faiss::fvec_inner_product(x, y, d);
y += d;
}
}
void fvec_L2sqr_ny_ref(
float* dis,
const float* x,
const float* y,
size_t d,
size_t ny) {
for (size_t i = 0; i < ny; i++) {
dis[i] = faiss::fvec_L2sqr(x, y, d);
y += d;
}
}
// test templated versions of fvec_L2sqr_ny
TEST(TestFvecL2sqrNy, D2) {
// we're using int values in order to get 100% accurate
// results with floats.
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> u(0, 32);
for (const auto dim : {2, 4, 8, 12}) {
std::vector<float> x(dim, 0);
for (size_t i = 0; i < x.size(); i++) {
x[i] = u(rng);
}
for (const auto nrows : {1, 2, 5, 10, 15, 20, 25}) {
std::vector<float> y(nrows * dim);
for (size_t i = 0; i < y.size(); i++) {
y[i] = u(rng);
}
std::vector<float> distances(nrows, 0);
faiss::fvec_L2sqr_ny(
distances.data(), x.data(), y.data(), dim, nrows);
std::vector<float> distances_ref(nrows, 0);
fvec_L2sqr_ny_ref(
distances_ref.data(), x.data(), y.data(), dim, nrows);
ASSERT_EQ(distances, distances_ref)
<< "Mismatching results for dim = " << dim
<< ", nrows = " << nrows;
}
}
}
// fvec_inner_products_ny
TEST(TestFvecInnerProductsNy, D2) {
// we're using int values in order to get 100% accurate
// results with floats.
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> u(0, 32);
for (const auto dim : {2, 4, 8, 12}) {
std::vector<float> x(dim, 0);
for (size_t i = 0; i < x.size(); i++) {
x[i] = u(rng);
}
for (const auto nrows : {1, 2, 5, 10, 15, 20, 25}) {
std::vector<float> y(nrows * dim);
for (size_t i = 0; i < y.size(); i++) {
y[i] = u(rng);
}
std::vector<float> distances(nrows, 0);
faiss::fvec_inner_products_ny(
distances.data(), x.data(), y.data(), dim, nrows);
std::vector<float> distances_ref(nrows, 0);
fvec_inner_products_ny_ref(
distances_ref.data(), x.data(), y.data(), dim, nrows);
ASSERT_EQ(distances, distances_ref)
<< "Mismatching results for dim = " << dim
<< ", nrows = " << nrows;
}
}
}
TEST(TestFvecL2sqr, distances_L2_squared_y_transposed) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
// modulo 8 results - 16 is to repeat the loop in the function
int ny = 11; // this value will hit all the codepaths
for (const auto d : {1, 2, 3, 4, 5, 6, 7, 8, 16}) {
// initialize inputs
std::vector<float> x(d);
float x_sqlen = 0;
for (size_t i = 0; i < x.size(); i++) {
x[i] = uniform(rng);
x_sqlen += x[i] * x[i];
}
std::vector<float> y(d * ny);
std::vector<float> y_sqlens(ny, 0);
for (size_t i = 0; i < ny; i++) {
for (size_t j = 0; j < y.size(); j++) {
y[j] = uniform(rng);
y_sqlens[i] += y[j] * y[j];
}
}
// perform function
std::vector<float> true_distances(ny, 0);
for (size_t i = 0; i < ny; i++) {
float dp = 0;
for (size_t j = 0; j < d; j++) {
dp += x[j] * y[i + j * ny];
}
true_distances[i] = x_sqlen + y_sqlens[i] - 2 * dp;
}
std::vector<float> distances(ny);
faiss::fvec_L2sqr_ny_transposed(
distances.data(),
x.data(),
y.data(),
y_sqlens.data(),
d,
ny, // no need for special offset to test all lines of code
ny);
ASSERT_EQ(distances, true_distances)
<< "Mismatching fvec_L2sqr_ny_transposed results for d = " << d;
}
}
TEST(TestFvecL2sqr, nearest_L2_squared_y_transposed) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
// modulo 8 results - 16 is to repeat the loop in the function
int ny = 11; // this value will hit all the codepaths
for (const auto d : {1, 2, 3, 4, 5, 6, 7, 8, 16}) {
// initialize inputs
std::vector<float> x(d);
float x_sqlen = 0;
for (size_t i = 0; i < x.size(); i++) {
x[i] = uniform(rng);
x_sqlen += x[i] * x[i];
}
std::vector<float> y(d * ny);
std::vector<float> y_sqlens(ny, 0);
for (size_t i = 0; i < ny; i++) {
for (size_t j = 0; j < y.size(); j++) {
y[j] = uniform(rng);
y_sqlens[i] += y[j] * y[j];
}
}
// get distances
std::vector<float> distances(ny, 0);
for (size_t i = 0; i < ny; i++) {
float dp = 0;
for (size_t j = 0; j < d; j++) {
dp += x[j] * y[i + j * ny];
}
distances[i] = x_sqlen + y_sqlens[i] - 2 * dp;
}
// find nearest
size_t true_nearest_idx = 0;
float min_dis = HUGE_VALF;
for (size_t i = 0; i < ny; i++) {
if (distances[i] < min_dis) {
min_dis = distances[i];
true_nearest_idx = i;
}
}
std::vector<float> buffer(ny);
size_t nearest_idx = faiss::fvec_L2sqr_ny_nearest_y_transposed(
buffer.data(),
x.data(),
y.data(),
y_sqlens.data(),
d,
ny, // no need for special offset to test all lines of code
ny);
ASSERT_EQ(nearest_idx, true_nearest_idx)
<< "Mismatching fvec_L2sqr_ny_nearest_y_transposed results for d = "
<< d;
}
}
TEST(TestFvecL1, manhattan_distance) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
// modulo 8 results - 16 is to repeat the while loop in the function
for (const auto nrows : {8, 9, 10, 11, 12, 13, 14, 15, 16}) {
std::vector<float> x(nrows);
std::vector<float> y(nrows);
float true_distance = 0;
for (size_t i = 0; i < x.size(); i++) {
x[i] = uniform(rng);
y[i] = uniform(rng);
true_distance += std::abs(x[i] - y[i]);
}
auto distance = faiss::fvec_L1(x.data(), y.data(), x.size());
ASSERT_EQ(distance, true_distance)
<< "Mismatching fvec_Linf results for nrows = " << nrows;
}
}
TEST(TestFvecLinf, chebyshev_distance) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
// modulo 8 results - 16 is to repeat the while loop in the function
for (const auto nrows : {8, 9, 10, 11, 12, 13, 14, 15, 16}) {
std::vector<float> x(nrows);
std::vector<float> y(nrows);
float true_distance = 0;
for (size_t i = 0; i < x.size(); i++) {
x[i] = uniform(rng);
y[i] = uniform(rng);
true_distance = std::max(true_distance, std::abs(x[i] - y[i]));
}
auto distance = faiss::fvec_Linf(x.data(), y.data(), x.size());
ASSERT_EQ(distance, true_distance)
<< "Mismatching fvec_Linf results for nrows = " << nrows;
}
}
TEST(TestFvecMadd, multiple_add) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
// modulo 8 results - 16 is to repeat the while loop in the function
for (const auto nrows : {8, 9, 10, 11, 12, 13, 14, 15, 16}) {
std::vector<float> a(nrows);
std::vector<float> b(nrows);
const float bf = uniform(rng);
std::vector<float> true_distances(nrows);
for (size_t i = 0; i < a.size(); i++) {
a[i] = uniform(rng);
b[i] = uniform(rng);
true_distances[i] = a[i] + bf * b[i];
}
std::vector<float> distances(nrows);
faiss::fvec_madd(a.size(), a.data(), bf, b.data(), distances.data());
ASSERT_EQ(distances, true_distances)
<< "Mismatching fvec_madd results for nrows = " << nrows;
}
}
TEST(TestFvecAdd, add_array) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
for (const auto nrows : {1, 2, 5, 10, 15, 20, 25}) {
std::vector<float> a(nrows);
std::vector<float> b(nrows);
std::vector<float> true_distances(nrows);
for (size_t i = 0; i < a.size(); i++) {
a[i] = uniform(rng);
b[i] = uniform(rng);
true_distances[i] = a[i] + b[i];
}
std::vector<float> distances(nrows);
faiss::fvec_add(a.size(), a.data(), b.data(), distances.data());
ASSERT_EQ(distances, true_distances)
<< "Mismatching array-array fvec_add results for nrows = "
<< nrows;
}
}
TEST(TestFvecAdd, add_value) {
// ints instead of floats for 100% accuracy
std::default_random_engine rng(123);
std::uniform_int_distribution<int32_t> uniform(0, 32);
for (const auto nrows : {1, 2, 5, 10, 15, 20, 25}) {
std::vector<float> a(nrows);
const float b = uniform(rng); // value to add
std::vector<float> true_distances(nrows);
for (size_t i = 0; i < a.size(); i++) {
a[i] = uniform(rng);
true_distances[i] = a[i] + b;
}
std::vector<float> distances(nrows);
faiss::fvec_add(a.size(), a.data(), b, distances.data());
ASSERT_EQ(distances, true_distances)
<< "Mismatching array-value fvec_add results for nrows = "
<< nrows;
}
}
|
