Skip to content

M
MariaDB
Commit 518f9cb7 authored by Sergei Golubchik's avatar Sergei Golubchik
Browse files
Options

if we require Eigen, we can as well use it everywhere 

it's measurably faster even in items
parent 78bd693a
Show whitespace changes
Inline Side-by-side
Showing with 180 additions and 179 deletions
mysql-test/main/vector.result
View file @ 518f9cb7
...@@ -108,14 +108,14 @@ id hex(v) ...@@ -108,14 +108,14 @@ id hex(v)
flush tables; flush tables;
select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
9 0.4719976290006591 9 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
select id,vec_distance_euclidean(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3; select id,vec_distance_euclidean(x'b047263C9f87233fcfd27e3eae493e3f0329f43e', v) d from t1 order by d limit 3;
id d id d
9 0.4719976290006591 9 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
select id>0,vec_distance_euclidean(v, NULL) d from t1 order by d limit 3; select id>0,vec_distance_euclidean(v, NULL) d from t1 order by d limit 3;
id>0 d id>0 d
1 NULL 1 NULL
...@@ -138,149 +138,149 @@ id1 id2 vec_distance_euclidean(t1.v, t2.v) ...@@ -138,149 +138,149 @@ id1 id2 vec_distance_euclidean(t1.v, t2.v)
8 8 0 8 8 0
9 9 0 9 9 0
10 10 0 10 10 0
7 10 0.35209010323904116 7 10 0.35209009051322937
10 7 0.35209010323904116 10 7 0.35209009051322937
1 7 0.557267332724855 1 7 0.5572673082351685
7 1 0.557267332724855 7 1 0.5572673082351685
2 3 0.6065128837978769 2 3 0.6065129041671753
3 2 0.6065128837978769 3 2 0.6065129041671753
1 3 0.6128238020507096 1 3 0.6128237843513489
3 1 0.6128238020507096 3 1 0.6128237843513489
5 8 0.6219995745138945 5 8 0.6219995617866516
8 5 0.6219995745138945 8 5 0.6219995617866516
3 10 0.6523185662547816 3 10 0.65231853723526
10 3 0.6523185662547816 10 3 0.65231853723526
9 10 0.6732681362788765 9 10 0.6732681393623352
10 9 0.6732681362788765 10 9 0.6732681393623352
3 7 0.6799892416547949 3 7 0.679989218711853
7 3 0.6799892416547949 7 3 0.679989218711853
3 9 0.6820752294088018 3 9 0.6820752024650574
9 3 0.6820752294088018 9 3 0.6820752024650574
2 10 0.6916305331777215 2 10 0.6916305422782898
10 2 0.6916305331777215 10 2 0.6916305422782898
2 9 0.6966650510789955 2 9 0.6966650485992432
9 2 0.6966650510789955 9 2 0.6966650485992432
3 6 0.7102823580937639 3 6 0.7102823853492737
6 3 0.7102823580937639 6 3 0.7102823853492737
2 7 0.7120217580666971 2 7 0.7120217680931091
7 2 0.7120217580666971 7 2 0.7120217680931091
2 6 0.7351618106552689 2 6 0.7351617813110352
6 2 0.7351618106552689 6 2 0.7351617813110352
1 10 0.7386864491588024 1 10 0.7386864423751831
10 1 0.7386864491588024 10 1 0.7386864423751831
4 6 0.7784357824370262 4 6 0.7784357666969299
6 4 0.7784357824370262 6 4 0.7784357666969299
4 8 0.7795837407361241 4 8 0.779583752155304
8 4 0.7795837407361241 8 4 0.779583752155304
4 5 0.8132007346697969 4 5 0.8132007122039795
5 4 0.8132007346697969 5 4 0.8132007122039795
2 4 0.8260925223296488 2 4 0.826092541217804
4 2 0.8260925223296488 4 2 0.826092541217804
5 10 0.8286488932765299 5 10 0.8286488652229309
10 5 0.8286488932765299 10 5 0.8286488652229309
5 9 0.8769351333060768 5 9 0.8769351243972778
9 5 0.8769351333060768 9 5 0.8769351243972778
1 6 0.8861410875047832 1 6 0.8861410617828369
6 1 0.8861410875047832 6 1 0.8861410617828369
3 5 0.9224201772876247 3 5 0.9224202036857605
5 3 0.9224201772876247 5 3 0.9224202036857605
4 7 0.9347916246876117 4 7 0.934791624546051
7 4 0.9347916246876117 7 4 0.934791624546051
7 9 0.9364253407685257 7 9 0.9364253282546997
9 7 0.9364253407685257 9 7 0.9364253282546997
3 4 0.9757105842688992 3 4 0.9757105708122253
4 3 0.9757105842688992 4 3 0.9757105708122253
1 2 0.9810272439433514 1 2 0.9810272455215454
2 1 0.9810272439433514 2 1 0.9810272455215454
1 4 0.9965475544626712 1 4 0.9965475797653198
4 1 0.9965475544626712 4 1 0.9965475797653198
5 7 0.9976863778073342 5 7 0.9976863861083984
7 5 0.9976863778073342 7 5 0.9976863861083984
4 10 1.0109345944029724 4 10 1.010934591293335
10 4 1.0109345944029724 10 4 1.010934591293335
1 5 1.0208359400987237 1 5 1.0208359956741333
5 1 1.0208359400987237 5 1 1.0208359956741333
6 7 1.0221332668982412 6 7 1.022133231163025
7 6 1.0221332668982412 7 6 1.022133231163025
2 5 1.050769316594881 2 5 1.050769329071045
5 2 1.050769316594881 5 2 1.050769329071045
6 8 1.103420381318026 6 8 1.103420376777649
8 6 1.103420381318026 8 6 1.103420376777649
3 8 1.1170300826294572 3 8 1.1170300245285034
8 3 1.1170300826294572 8 3 1.1170300245285034
6 10 1.1523451990991307 6 10 1.1523451805114746
10 6 1.1523451990991307 10 6 1.1523451805114746
1 9 1.1637750565139302 1 9 1.163775086402893
9 1 1.1637750565139302 9 1 1.163775086402893
2 8 1.1736571017573874 2 8 1.1736570596694946
8 2 1.1736571017573874 8 2 1.1736570596694946
4 9 1.1746893942711878 4 9 1.1746894121170044
9 4 1.1746893942711878 9 4 1.1746894121170044
1 8 1.1909959973982214 1 8 1.1909960508346558
8 1 1.1909959973982214 8 1 1.1909960508346558
8 10 1.209359617652948 8 10 1.2093596458435059
10 8 1.209359617652948 10 8 1.2093596458435059
6 9 1.214529873940304 6 9 1.2145298719406128
9 6 1.214529873940304 9 6 1.2145298719406128
5 6 1.227278506501395 5 6 1.2272785902023315
6 5 1.227278506501395 6 5 1.2272785902023315
8 9 1.2575258643523053 8 9 1.2575258016586304
9 8 1.2575258643523053 9 8 1.2575258016586304
7 8 1.288239696195716 7 8 1.288239598274231
8 7 1.288239696195716 8 7 1.288239598274231
flush session status; flush session status;
select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
9 0.4719976290006591 9 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 0 Handler_read_rnd_next 0
select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 use index () order by d limit 3; select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 use index () order by d limit 3;
id d id d
9 0.4719976290006591 9 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 11 Handler_read_rnd_next 11
flush session status; flush session status;
select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
10 0.05905546376032378 10 0.059055447578430176
9 0.06546887818344715 9 0.06546902656555176
3 0.10750282439505232 3 0.10750287771224976
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 11 Handler_read_rnd_next 11
delete from t1 where v = x'7b713f3e5258323f80d1113d673b2b3f66e3583f'; delete from t1 where v = x'7b713f3e5258323f80d1113d673b2b3f66e3583f';
select id,vec_distance_euclidean(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_euclidean(v, x'B047263C9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
7 0.7344464697214867 7 0.7344464659690857
insert t1 (v) values (x'7b713f3e5258323f80d1113d673b2b3f66e3583f'); insert t1 (v) values (x'7b713f3e5258323f80d1113d673b2b3f66e3583f');
select id,vec_distance_euclidean(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_euclidean(v, x'b047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
11 0.4719976290006591 11 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
select id,vec_distance_euclidean(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'B047263c9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
id d id d
11 0.4719976290006591 11 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
7 0.7344464697214867 7 0.7344464659690857
5 0.7671033529042712 5 0.7671033143997192
update t1 set v=x'76EDFC3E4B57243F10F8423FB158713F020BAA3E' where v=x'6CA1D43E9DF91B3FE580DA3E1C247D3F147CF33E'; update t1 set v=x'76EDFC3E4B57243F10F8423FB158713F020BAA3E' where v=x'6CA1D43E9DF91B3FE580DA3E1C247D3F147CF33E';
select id,vec_distance_euclidean(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'B047263C9F87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
id d id d
11 0.4719976290006591 11 0.47199761867523193
3 0.5865673124650332 3 0.5865673422813416
7 0.7344464697214867 7 0.7344464659690857
10 0.746836719209219 10 0.7468367218971252
5 0.7671033529042712 5 0.7671033143997192
delete from t1; delete from t1;
insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'), insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
(x'f511303f72224a3fdd05fe3eb22a133ffae86a3f'), (x'f511303f72224a3fdd05fe3eb22a133ffae86a3f'),
...@@ -294,11 +294,11 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'), ...@@ -294,11 +294,11 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
(x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e'); (x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e');
select id,vec_distance_euclidean(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'b047263c9f87233Fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 5;
id d id d
20 0.4719976290006591 20 0.47199761867523193
21 0.5069011044450041 21 0.5069010853767395
14 0.5865673124650332 14 0.5865673422813416
18 0.7344464697214867 18 0.7344464659690857
16 0.7671033529042712 16 0.7671033143997192
insert t1 (v) values (''); insert t1 (v) values ('');
ERROR 22007: Incorrect vector value: '...' for column `test`.`t1`.`v` at row 1 ERROR 22007: Incorrect vector value: '...' for column `test`.`t1`.`v` at row 1
insert t1 (v) values (x'1234'); insert t1 (v) values (x'1234');
...@@ -426,32 +426,32 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'), ...@@ -426,32 +426,32 @@ insert t1 (v) values (x'e360d63ebe554f3fcdbc523f4522193f5236083d'),
(x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e'); (x'6ca1d43e9df91b3fe580da3e1c247d3f147cf33e');
select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
10 0.05905546376032378 10 0.059055447578430176
9 0.06546887818344715 9 0.06546902656555176
3 0.10750282439505232 3 0.10750287771224976
flush session status; flush session status;
select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
10 0.05905546376032378 10 0.059055447578430176
9 0.06546887818344715 9 0.06546902656555176
3 0.10750282439505232 3 0.10750287771224976
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 0 Handler_read_rnd_next 0
select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 use index () order by d limit 3; select id,vec_distance_cosine(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 use index () order by d limit 3;
id d id d
10 0.05905546376032378 10 0.059055447578430176
9 0.06546887818344715 9 0.06546902656555176
3 0.10750282439505232 3 0.10750287771224976
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 11 Handler_read_rnd_next 11
flush session status; flush session status;
select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3; select id,vec_distance_euclidean(v, x'B047263c9f87233fcfd27e3eae493e3f0329f43e') d from t1 order by d limit 3;
id d id d
9 0.4719976290006591 9 0.47199761867523193
10 0.5069011044450041 10 0.5069010853767395
3 0.5865673124650332 3 0.5865673422813416
show status like 'handler_read_rnd_next'; show status like 'handler_read_rnd_next';
Variable_name Value Variable_name Value
Handler_read_rnd_next 11 Handler_read_rnd_next 11
......
mysql-test/main/vector_innodb.result
View file @ 518f9cb7
...@@ -40,18 +40,18 @@ delete from t1 where id=7; ...@@ -40,18 +40,18 @@ delete from t1 where id=7;
rollback; rollback;
select id,vec_distance_euclidean(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'c923e33dc0da313fe7c7983e526b3d3fde63963e6eaf3a3f27fa133fe27a583f') d from t1 order by d limit 5;
id d id d
2 0.8781474260354732 2 0.8781474828720093
10 0.8856208347761952 10 0.8856208324432373
30 1.0162643974895857 30 1.0162643194198608
7 1.026397313888122 7 1.0263972282409668
5 1.0308161006949719 5 1.0308160781860352
select id,vec_distance_euclidean(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
id d id d
2 0.9426904171992334 2 0.9426904320716858
33 0.9477554826856 33 0.9477554559707642
30 1.111405427702547 30 1.1114054918289185
10 1.118630286292343 10 1.118630290031433
8 1.1405733350751739 8 1.140573263168335
create table t2 (id int auto_increment primary key, v blob not null, vector index (v)) engine=innodb; create table t2 (id int auto_increment primary key, v blob not null, vector index (v)) engine=innodb;
insert t2 (v) values insert t2 (v) values
(x'45cf153f830a313f7a0a113fb1ff533f47a1533fcf9e6e3f'), (x'45cf153f830a313f7a0a113fb1ff533f47a1533fcf9e6e3f'),
...@@ -73,16 +73,16 @@ insert t2 values ...@@ -73,16 +73,16 @@ insert t2 values
commit; commit;
select id,vec_distance_euclidean(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5; select id,vec_distance_euclidean(v, x'1f4d053f7056493f937da03dd8c97a3f220cbb3c926c1c3facca213ec0618a3e') d from t1 order by d limit 5;
id d id d
6 0.9309383181777582 6 0.9309383034706116
5 0.9706304662574956 5 0.9706304669380188
30 0.98144492002831 30 0.9814448952674866
50 1.079862635421575 50 1.0798625946044922
2 1.0907138991979892 2 1.0907139778137207
select id,vec_distance_euclidean(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5; select id,vec_distance_euclidean(v, x'f618663f256be73e62cd453f8bcdbf3e16ae503c3858313f') d from t2 order by d limit 5;
id d id d
21 0.43559180321379337 21 0.43559178709983826
20 0.6435053022072372 20 0.643505334854126
6 0.6942000623336242 6 0.6942000389099121
2 0.7971622099055623 2 0.7971622347831726
9 0.8298589136476077 9 0.8298588991165161
drop table t1, t2; drop table t1, t2;
sql/item_vectorfunc.cc
View file @ 518f9cb7
...@@ -215,3 +215,23 @@ String *Item_func_vec_fromtext::val_str(String *buf) ...@@ -215,3 +215,23 @@ String *Item_func_vec_fromtext::val_str(String *buf)
null_value= true; null_value= true;
return nullptr; return nullptr;
} }
#include <eigen3/Eigen/Dense>
using namespace Eigen;
double Item_func_vec_distance_euclidean::
calc_distance(float *data1, float *data2, size_t d_len)
{
Map<VectorXf> v1(data1, d_len);
Map<VectorXf> v2(data2, d_len);
return (v1-v2).norm();
}
double Item_func_vec_distance_cosine::
calc_distance(float *data1, float *data2, size_t d_len)
{
Map<VectorXf> v1(data1, d_len);
Map<VectorXf> v2(data2, d_len);
return 1.0f - v1.dot(v2)/v1.norm()/v2.norm();
}
sql/item_vectorfunc.h
View file @ 518f9cb7
...@@ -61,16 +61,7 @@ class Item_func_vec_distance_common: public Item_real_func ...@@ -61,16 +61,7 @@ class Item_func_vec_distance_common: public Item_real_func
class Item_func_vec_distance_euclidean: public Item_func_vec_distance_common class Item_func_vec_distance_euclidean: public Item_func_vec_distance_common
{ {
double calc_distance(float *v1, float *v2, size_t v_len) override double calc_distance(float *v1, float *v2, size_t v_len) override;
{
double d= 0;
for (size_t i= 0; i < v_len; i++, v1++, v2++)
{
float dist= *v1 - *v2;
d+= dist * dist;
}
return sqrt(d);
}
public: public:
Item_func_vec_distance_euclidean(THD *thd, Item *a, Item *b) Item_func_vec_distance_euclidean(THD *thd, Item *a, Item *b)
...@@ -87,17 +78,7 @@ class Item_func_vec_distance_euclidean: public Item_func_vec_distance_common ...@@ -87,17 +78,7 @@ class Item_func_vec_distance_euclidean: public Item_func_vec_distance_common
class Item_func_vec_distance_cosine: public Item_func_vec_distance_common class Item_func_vec_distance_cosine: public Item_func_vec_distance_common
{ {
double calc_distance(float *v1, float *v2, size_t v_len) override double calc_distance(float *v1, float *v2, size_t v_len) override;
{
double dotp=0, abs1=0, abs2=0;
for (size_t i= 0; i < v_len; i++, v1++, v2++)
{
abs1+= *v1 * *v1;
abs2+= *v2 * *v2;
dotp+= *v1 * *v2;
}
return 1 - dotp/sqrt(abs1*abs2);
}
public: public:
Item_func_vec_distance_cosine(THD *thd, Item *a, Item *b) Item_func_vec_distance_cosine(THD *thd, Item *a, Item *b)
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7