Code Duplication    Length = 23-23 lines in 2 locations

core/src/index/thirdparty/faiss/benchs/bench_gpu_1bn.py 1 location

@@ 199-221 (lines=23) @@
196
197
print("Preparing dataset", dbname)
198
199
if dbname.startswith('SIFT'):
200
    # SIFT1M to SIFT1000M
201
    dbsize = int(dbname[4:-1])
202
    xb = mmap_bvecs('bigann/bigann_base.bvecs')
203
    xq = mmap_bvecs('bigann/bigann_query.bvecs')
204
    xt = mmap_bvecs('bigann/bigann_learn.bvecs')
205
206
    # trim xb to correct size
207
    xb = xb[:dbsize * 1000 * 1000]
208
209
    gt_I = ivecs_read('bigann/gnd/idx_%dM.ivecs' % dbsize)
210
211
elif dbname == 'Deep1B':
212
    xb = mmap_fvecs('deep1b/base.fvecs')
213
    xq = mmap_fvecs('deep1b/deep1B_queries.fvecs')
214
    xt = mmap_fvecs('deep1b/learn.fvecs')
215
    # deep1B's train is is outrageously big
216
    xt = xt[:10 * 1000 * 1000]
217
    gt_I = ivecs_read('deep1b/deep1B_groundtruth.ivecs')
218
219
else:
220
    print('unknown dataset', dbname, file=sys.stderr)
221
    sys.exit(1)
222
223
224
if knngraph:

core/src/index/thirdparty/faiss/benchs/bench_polysemous_1bn.py 1 location

@@ 59-81 (lines=23) @@
56
57
print("Preparing dataset", dbname)
58
59
if dbname.startswith('SIFT'):
60
    # SIFT1M to SIFT1000M
61
    dbsize = int(dbname[4:-1])
62
    xb = mmap_bvecs('bigann/bigann_base.bvecs')
63
    xq = mmap_bvecs('bigann/bigann_query.bvecs')
64
    xt = mmap_bvecs('bigann/bigann_learn.bvecs')
65
66
    # trim xb to correct size
67
    xb = xb[:dbsize * 1000 * 1000]
68
69
    gt = ivecs_read('bigann/gnd/idx_%dM.ivecs' % dbsize)
70
71
elif dbname == 'Deep1B':
72
    xb = mmap_fvecs('deep1b/base.fvecs')
73
    xq = mmap_fvecs('deep1b/deep1B_queries.fvecs')
74
    xt = mmap_fvecs('deep1b/learn.fvecs')
75
    # deep1B's train is is outrageously big
76
    xt = xt[:10 * 1000 * 1000]
77
    gt = ivecs_read('deep1b/deep1B_groundtruth.ivecs')
78
79
else:
80
    print('unknown dataset', dbname, file=sys.stderr)
81
    sys.exit(1)
82
83
84
print("sizes: B %s Q %s T %s gt %s" % (