Skip to content

Commit a82ac73

Browse files
committed
finished
1 parent 85ce607 commit a82ac73

3 files changed

Lines changed: 40 additions & 36 deletions

File tree

README.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,9 @@
11
# zk-IVF-PQ
22

3+
## Technical Note
4+
5+
Implementation details are provided in our [technical note](Technical_Details_of_V3DB.pdf).
6+
37
## Build
48

59
Build and install the Python extension:

Technical_Details_of_V3DB.pdf

471 KB
Binary file not shown.

scripts/acc_bench.sh

Lines changed: 36 additions & 36 deletions
Original file line numberDiff line numberDiff line change
@@ -6,31 +6,31 @@ set -euo pipefail
66

77
# Experiment 1 / Classic ANN (Table "Experiment 1 datasets and IVF-PQ layouts")
88
# SIFT1M high-acc: (n_list, n_probe, n) = (8192, 64, 256)
9-
# python -m tests.acc_bench \
10-
# --data-name sift \
11-
# --name gpu5_exp1_sift_high_acc_multik \
12-
# --num-runs 1 \
13-
# --M 8 \
14-
# --K 256 \
15-
# --top-k 100 \
16-
# --report-ks 1,10,50,100,500 \
17-
# --scale-n 65536 \
18-
# --n-list 8192 \
19-
# --n-probe 64 \
20-
# --cluster-bound 256 \
21-
# --layout none
22-
#
23-
# sleep 60
9+
python -m tests.acc_bench \
10+
--data-name sift \
11+
--name exp1_sift_high_acc_multik \
12+
--num-runs 1 \
13+
--M 8 \
14+
--K 256 \
15+
--top-k 100 \
16+
--report-ks 10,50 \
17+
--scale-n 65536 \
18+
--n-list 8192 \
19+
--n-probe 64 \
20+
--cluster-bound 256 \
21+
--layout none
22+
23+
sleep 60
2424

2525
# GIST1M high-acc: (n_list, n_probe, n) = (8192, 64, 256)
26-
CUDA_VISIBLE_DEVICES=1,2,3,4,5,6,7 python -m tests.acc_bench \
26+
python -m tests.acc_bench \
2727
--data-name gist \
28-
--name gpu7_exp1_gist_high_acc_multik \
28+
--name exp1_gist_high_acc_multik \
2929
--num-runs 1 \
3030
--M 8 \
3131
--K 256 \
3232
--top-k 100 \
33-
--report-ks 1,10,50,100,500 \
33+
--report-ks 10,50 \
3434
--scale-n 65536 \
3535
--n-list 8192 \
3636
--n-probe 64 \
@@ -40,31 +40,31 @@ CUDA_VISIBLE_DEVICES=1,2,3,4,5,6,7 python -m tests.acc_bench \
4040
sleep 60
4141

4242
# SIFT1M zk-opt: (n_list, n_probe, n) = (1024, 8, 2048)
43-
# python -m tests.acc_bench \
44-
# --data-name sift \
45-
# --name gpu5_exp1_sift_zk_opt_multik \
46-
# --num-runs 1 \
47-
# --M 8 \
48-
# --K 256 \
49-
# --top-k 100 \
50-
# --report-ks 1,10,50,100,500 \
51-
# --scale-n 65536 \
52-
# --n-list 1024 \
53-
# --n-probe 8 \
54-
# --cluster-bound 2048 \
55-
# --layout none
56-
#
57-
# sleep 60
43+
python -m tests.acc_bench \
44+
--data-name sift \
45+
--name exp1_sift_zk_opt_multik \
46+
--num-runs 1 \
47+
--M 8 \
48+
--K 256 \
49+
--top-k 100 \
50+
--report-ks 10,50 \
51+
--scale-n 65536 \
52+
--n-list 1024 \
53+
--n-probe 8 \
54+
--cluster-bound 2048 \
55+
--layout none
56+
57+
sleep 60
5858

5959
# GIST1M zk-opt: (n_list, n_probe, n) = (512, 4, 4096)
60-
CUDA_VISIBLE_DEVICES=1,2,3,4,5,6,7 python -m tests.acc_bench \
60+
python -m tests.acc_bench \
6161
--data-name gist \
62-
--name gpu7_exp1_gist_zk_opt_multik \
62+
--name exp1_gist_zk_opt_multik \
6363
--num-runs 1 \
6464
--M 8 \
6565
--K 256 \
6666
--top-k 100 \
67-
--report-ks 1,10,50,100,500 \
67+
--report-ks 10,50 \
6868
--scale-n 65536 \
6969
--n-list 512 \
7070
--n-probe 4 \

0 commit comments

Comments
 (0)