percona-discussion team mailing list archive
-
percona-discussion team
-
Mailing list archive
-
Message #00217
[Bug 326464] Re: performance regression in tpc-c 1 thread
10 threads
"not tuned"
10, 2048(0):0.2, 2045(0):0.2, 205(0):0.2, 204(0):0.2, 205(0):0.4
20, 1915(0):0.2, 1915(0):0.2, 190(0):0.2, 191(0):0.2, 191(0):0.2
30, 1794(0):0.2, 1797(0):0.2, 181(0):0.2, 181(0):0.2, 179(0):0.2
40, 1693(0):0.2, 1697(0):0.2, 170(0):0.2, 170(0):0.2, 170(0):0.2
50, 1689(0):0.2, 1687(0):0.2, 168(0):0.2, 169(0):0.2, 169(0):0.2
60, 1638(0):0.2, 1637(0):0.2, 164(0):0.2, 164(0):0.2, 164(0):0.2
70, 1632(0):0.2, 1635(0):0.2, 163(0):0.2, 163(0):0.2, 163(0):0.2
80, 1627(0):0.2, 1624(0):0.2, 164(0):0.2, 162(0):0.2, 163(0):0.2
90, 1596(0):0.2, 1597(0):0.2, 159(0):0.2, 160(0):0.2, 159(0):0.2
100, 1461(0):0.2, 1460(0):0.2, 146(0):0.2, 146(0):0.2, 147(0):0.2
110, 955(0):0.2, 957(0):0.2, 96(0):0.2, 96(0):0.2, 97(0):0.2
120, 255(0):0.2, 255(0):0.2, 26(0):0.2, 27(0):0.2, 26(0):0.2
130, 22(0):0.2, 20(0):0.2, 1(0):0.2, 1(0):0.2, 2(0):0.2
140, 0(0):0.0, 0(0):0.0, 0(0):0.0, 0(0):0.0, 0(0):0.0
150, 1300(5):0.2, 1301(5):0.2, 130(0):0.2, 130(0):0.2, 129(0):0.2
160, 1081(0):0.2, 1083(0):0.2, 109(0):0.2, 107(0):0.2, 108(0):0.2
170, 968(3):0.2, 959(0):0.2, 97(0):0.2, 98(0):0.2, 98(0):0.2
180, 294(1):0.2, 303(8):0.2, 29(0):0.2, 29(0):0.2, 28(0):0.2
190, 1714(0):0.2, 1713(0):0.2, 171(0):0.2, 172(0):0.2, 172(0):0.2
200, 1720(0):0.2, 1721(0):0.2, 173(0):0.2, 172(0):0.2, 173(0):0.2
210, 1752(0):0.2, 1750(0):0.2, 175(0):0.2, 175(0):0.2, 175(0):0.2
220, 1720(0):0.2, 1723(0):0.2, 172(0):0.2, 172(0):0.2, 172(0):0.2
230, 1764(0):0.2, 1766(0):0.2, 177(0):0.2, 177(0):0.2, 177(0):0.2
240, 1742(0):0.2, 1739(0):0.2, 174(0):0.2, 174(0):0.2, 172(0):0.2
250, 1688(0):0.2, 1688(0):0.2, 168(0):0.2, 168(0):0.2, 169(0):0.2
260, 1625(0):0.2, 1620(0):0.2, 163(0):0.2, 163(0):0.2, 163(0):0.2
270, 1683(0):0.2, 1687(0):0.2, 169(0):0.2, 167(0):0.2, 169(0):0.2
280, 1706(0):0.2, 1710(0):0.2, 171(0):0.2, 173(0):0.2, 171(0):0.2
290, 1678(0):0.2, 1675(0):0.2, 168(0):0.2, 167(0):0.2, 167(0):0.2
300, 1631(0):0.2, 1626(0):0.2, 163(0):0.2, 163(0):0.2, 162(0):0.2
310, 1482(0):0.2, 1489(0):0.2, 148(0):0.2, 148(0):0.2, 151(0):0.2
320, 1288(0):0.2, 1286(0):0.2, 129(0):0.2, 130(0):0.2, 127(0):0.2
330, 1049(0):0.2, 1049(0):0.2, 105(0):0.2, 106(0):0.2, 106(0):0.2
340, 21(0):0.2, 19(0):0.2, 2(0):0.2, 2(0):0.2, 3(0):0.6
350, 244(3):0.2, 254(7):0.2, 25(0):0.2, 23(0):0.2, 25(0):0.2
360, 1299(0):0.2, 1295(0):0.2, 130(0):0.2, 131(0):0.2, 130(0):0.2
370, 1334(0):0.2, 1329(0):0.2, 133(0):0.2, 134(0):0.2, 133(0):0.2
380, 554(0):0.2, 550(0):0.2, 56(0):0.2, 56(0):0.2, 55(0):0.2
390, 1545(2):0.2, 1555(5):0.2, 154(0):0.2, 153(0):0.2, 155(0):0.2
400, 1629(0):0.2, 1634(0):0.2, 164(0):0.2, 163(0):0.2, 164(0):0.2
410, 1686(0):0.2, 1689(0):0.2, 169(0):0.2, 170(0):0.2, 168(0):0.2
420, 1687(0):0.2, 1688(0):0.2, 168(0):0.2, 168(0):0.2, 169(0):0.2
430, 1693(0):0.2, 1693(0):0.2, 170(0):0.2, 168(0):0.2, 169(0):0.2
440, 1560(0):0.2, 1559(0):0.2, 156(0):0.2, 158(0):0.2, 156(0):0.2
450, 1622(0):0.2, 1622(0):0.2, 162(0):0.2, 161(0):0.2, 162(0):0.2
460, 1575(0):0.2, 1570(0):0.2, 156(0):0.2, 158(0):0.2, 157(0):0.2
470, 1598(0):0.2, 1597(0):0.2, 161(0):0.2, 160(0):0.2, 160(0):0.2
480, 1613(0):0.2, 1612(0):0.2, 161(0):0.2, 161(0):0.2, 161(0):0.2
490, 1608(0):0.2, 1614(0):0.2, 162(0):0.2, 162(0):0.2, 161(0):0.2
500, 1535(0):0.2, 1528(0):0.2, 152(0):0.2, 153(0):0.2, 154(0):0.2
510, 1530(0):0.2, 1534(0):0.2, 154(0):0.2, 153(0):0.2, 153(0):0.2
520, 1394(0):0.2, 1392(0):0.2, 139(0):0.2, 140(0):0.2, 139(0):0.2
530, 1496(0):0.2, 1499(0):0.2, 151(0):0.2, 149(0):0.2, 149(0):0.4
540, 958(0):0.2, 953(0):0.2, 96(0):0.2, 97(0):0.2, 98(0):0.2
550, 330(0):0.2, 331(0):0.2, 32(0):0.2, 33(0):0.4, 33(0):0.2
560, 540(3):0.2, 541(6):0.2, 54(0):0.2, 53(0):0.2, 53(0):0.2
570, 1057(1):0.2, 1059(3):0.2, 106(0):0.2, 106(0):0.2, 104(0):0.2
580, 1398(0):0.2, 1398(0):0.2, 140(0):0.2, 140(0):0.2, 143(0):0.2
590, 1041(3):0.2, 1041(1):0.2, 103(0):0.2, 103(0):0.2, 103(0):0.2
600, 1634(0):0.2, 1634(0):0.2, 164(0):0.2, 164(0):0.2, 164(0):0.2
"tuned"
10, 1405(0):0.2, 1402(0):0.2, 139(0):0.2, 141(0):0.2, 141(0):0.4
20, 1619(0):0.2, 1630(0):0.2, 163(0):0.2, 162(0):0.2, 163(0):0.2
30, 1402(0):0.2, 1398(0):0.2, 140(0):0.2, 140(0):0.2, 139(0):0.2
40, 1424(0):0.2, 1427(0):0.2, 143(0):0.2, 143(0):0.2, 144(0):0.2
50, 1360(0):0.2, 1365(0):0.2, 136(0):0.2, 134(0):0.2, 136(0):0.2
60, 1485(0):0.2, 1483(0):0.2, 150(0):0.2, 151(0):0.2, 150(0):0.4
70, 1296(0):0.2, 1301(0):0.2, 129(0):0.2, 130(0):0.2, 129(0):0.2
80, 1401(0):0.2, 1401(0):0.2, 141(0):0.2, 140(0):0.2, 141(0):0.2
90, 1138(0):0.2, 1140(0):0.2, 113(0):0.2, 114(0):0.4, 112(0):0.4
100, 1086(0):0.2, 1082(0):0.2, 109(0):0.2, 109(0):0.4, 110(0):0.4
110, 1495(0):0.2, 1499(0):0.2, 149(0):0.2, 149(0):0.2, 149(0):0.2
120, 1156(2):0.2, 1152(0):0.2, 116(0):0.2, 115(0):0.4, 116(0):0.4
130, 1171(1):0.2, 1173(1):0.2, 117(0):0.2, 118(0):0.2, 118(0):0.2
140, 1431(0):0.2, 1433(0):0.2, 144(0):0.2, 144(0):0.2, 143(0):0.2
150, 1498(0):0.2, 1497(0):0.2, 150(0):0.2, 149(0):0.2, 150(0):0.2
160, 1536(0):0.2, 1538(0):0.2, 153(0):0.2, 154(0):0.2, 154(0):0.2
170, 1498(0):0.2, 1494(0):0.2, 151(0):0.2, 150(0):0.2, 150(0):0.2
180, 1275(0):0.2, 1273(0):0.2, 127(0):0.2, 127(0):0.4, 127(0):0.4
190, 1489(0):0.2, 1494(0):0.2, 149(0):0.2, 149(0):0.2, 148(0):0.2
200, 1471(0):0.2, 1475(0):0.2, 147(0):0.2, 148(0):0.2, 149(0):0.2
210, 1471(0):0.2, 1477(0):0.2, 148(0):0.2, 148(0):0.2, 147(0):0.2
220, 1525(0):0.2, 1522(0):0.2, 153(0):0.2, 153(0):0.2, 152(0):0.2
230, 1462(0):0.2, 1464(0):0.2, 146(0):0.2, 146(0):0.2, 146(0):0.2
240, 1606(0):0.2, 1600(0):0.2, 160(0):0.2, 159(0):0.2, 161(0):0.2
250, 1454(0):0.2, 1456(0):0.2, 146(0):0.2, 147(0):0.2, 146(0):0.2
260, 1459(0):0.2, 1458(0):0.2, 145(0):0.2, 146(0):0.2, 145(0):0.2
270, 1461(0):0.2, 1468(0):0.2, 148(0):0.2, 147(0):0.2, 148(0):0.2
280, 1492(0):0.2, 1494(0):0.2, 149(0):0.2, 149(0):0.2, 148(0):0.2
290, 1565(0):0.2, 1561(0):0.2, 156(0):0.2, 156(0):0.2, 158(0):0.2
300, 1483(0):0.2, 1484(0):0.2, 149(0):0.2, 148(0):0.2, 147(0):0.2
310, 1570(0):0.2, 1570(0):0.2, 158(0):0.2, 158(0):0.2, 160(0):0.2
320, 1560(0):0.2, 1561(0):0.2, 155(0):0.2, 156(0):0.2, 154(0):0.2
330, 1615(0):0.2, 1613(0):0.2, 162(0):0.2, 162(0):0.2, 162(0):0.2
340, 1616(0):0.2, 1621(0):0.2, 162(0):0.2, 161(0):0.2, 162(0):0.2
350, 1636(0):0.2, 1638(0):0.2, 163(0):0.2, 164(0):0.2, 163(0):0.2
360, 1577(0):0.2, 1580(0):0.2, 159(0):0.2, 159(0):0.2, 160(0):0.2
370, 1621(0):0.2, 1616(0):0.2, 161(0):0.2, 161(0):0.2, 161(0):0.2
380, 1578(0):0.2, 1582(0):0.2, 158(0):0.2, 158(0):0.2, 158(0):0.2
390, 1507(0):0.2, 1506(0):0.2, 151(0):0.2, 151(0):0.2, 150(0):0.2
400, 1601(0):0.2, 1606(0):0.2, 160(0):0.2, 161(0):0.2, 161(0):0.2
410, 1492(0):0.2, 1493(0):0.2, 150(0):0.2, 149(0):0.2, 149(0):0.2
420, 1567(0):0.2, 1573(0):0.2, 157(0):0.2, 156(0):0.2, 157(0):0.2
430, 1449(0):0.2, 1450(0):0.2, 146(0):0.2, 146(0):0.2, 145(0):0.2
440, 1603(0):0.2, 1599(0):0.2, 160(0):0.2, 160(0):0.2, 160(0):0.2
450, 1361(0):0.2, 1363(0):0.2, 136(0):0.2, 138(0):0.2, 138(0):0.2
460, 1664(0):0.2, 1663(0):0.2, 166(0):0.2, 165(0):0.2, 166(0):0.2
470, 1525(0):0.2, 1527(0):0.2, 153(0):0.2, 152(0):0.2, 152(0):0.2
480, 1563(0):0.2, 1564(0):0.2, 156(0):0.2, 157(0):0.2, 158(0):0.2
490, 1582(0):0.2, 1579(0):0.2, 159(0):0.2, 159(0):0.2, 157(0):0.2
500, 1783(0):0.2, 1784(0):0.2, 178(0):0.2, 178(0):0.2, 179(0):0.2
510, 1688(0):0.2, 1683(0):0.2, 168(0):0.2, 169(0):0.2, 167(0):0.2
520, 1749(0):0.2, 1750(0):0.2, 175(0):0.2, 175(0):0.2, 174(0):0.2
530, 1848(0):0.2, 1851(0):0.2, 186(0):0.2, 185(0):0.2, 187(0):0.2
540, 1604(0):0.2, 1599(0):0.2, 160(0):0.2, 160(0):0.2, 160(0):0.2
550, 1776(0):0.2, 1782(0):0.2, 179(0):0.2, 178(0):0.2, 178(0):0.2
560, 1570(0):0.2, 1570(0):0.2, 156(0):0.2, 157(0):0.2, 156(0):0.2
570, 1660(0):0.2, 1663(0):0.2, 166(0):0.2, 166(0):0.2, 167(0):0.2
580, 1699(0):0.2, 1703(0):0.2, 171(0):0.2, 172(0):0.2, 171(0):0.2
590, 1731(0):0.2, 1729(0):0.2, 173(0):0.2, 172(0):0.2, 174(0):0.2
600, 1710(0):0.2, 1718(0):0.2, 171(0):0.2, 171(0):0.2, 171(0):0.2
** Description changed:
I observe performance regression if run XtraDB (the same for percona
patches in 5.0.75) with IO tuned parameters in 1 thread tpc-c
benchmarks. With 4 threads = results are OK.
tuned parameters:
innodb_write_io_threads=4
innodb_read_io_threads=4
innodb_io_capacity=800
innodb_adaptive_checkpoint=1
full my.cnf:
====
[mysqld]
#mysqld options in alphabetical order
user=root
default_table_type=MYISAM
innodb_buffer_pool_size=3G
innodb_data_file_path=ibdata1:10M:autoextend
innodb_file_per_table=1
innodb_flush_log_at_trx_commit=1
innodb_log_buffer_size=8M
innodb_log_files_in_group=2
innodb_log_file_size=128M
innodb_thread_concurrency=0
innodb_flush_method = O_DIRECT
innodb_write_io_threads=4
innodb_read_io_threads=4
innodb_io_capacity=800
innodb_adaptive_checkpoint=1
#innodb_extra_rsegments=16
#innodb_thread_concurrency=16
max_connections=3000
query_cache_size=0
skip-name-resolve
table_cache=2048
===
I use tpc-c 100W.
for 1 thread not tuned I have 3483.8 TpmC
tuned - 3156.6
- for 4 threads:
+ for 10 threads:
not tuned: 8039.1
tuned: 9111.9
So we have something wrong in 1-thread case.
--
performance regression in tpc-c 1 thread
https://bugs.launchpad.net/bugs/326464
You received this bug notification because you are a member of Percona
developers, which is the registrant for Percona patches.
Status in Patches for MySQL by Percona: Confirmed
Status in Percona XtraDB Storage Engine for MySQL: Confirmed
Bug description:
I observe performance regression if run XtraDB (the same for percona patches in 5.0.75) with IO tuned parameters in 1 thread tpc-c benchmarks. With 4 threads = results are OK.
tuned parameters:
innodb_write_io_threads=4
innodb_read_io_threads=4
innodb_io_capacity=800
innodb_adaptive_checkpoint=1
full my.cnf:
====
[mysqld]
#mysqld options in alphabetical order
user=root
default_table_type=MYISAM
innodb_buffer_pool_size=3G
innodb_data_file_path=ibdata1:10M:autoextend
innodb_file_per_table=1
innodb_flush_log_at_trx_commit=1
innodb_log_buffer_size=8M
innodb_log_files_in_group=2
innodb_log_file_size=128M
innodb_thread_concurrency=0
innodb_flush_method = O_DIRECT
innodb_write_io_threads=4
innodb_read_io_threads=4
innodb_io_capacity=800
innodb_adaptive_checkpoint=1
#innodb_extra_rsegments=16
#innodb_thread_concurrency=16
max_connections=3000
query_cache_size=0
skip-name-resolve
table_cache=2048
===
I use tpc-c 100W.
for 1 thread not tuned I have 3483.8 TpmC
tuned - 3156.6
for 10 threads:
not tuned: 8039.1
tuned: 9111.9
So we have something wrong in 1-thread case.
References