percona-discussion team mailing list archive

Thread
Date
[Bug 326464] Re: performance regression in tpc-c 1 thread

To: percona-discussion@xxxxxxxxxxxxxxxxxxx
From: Vadim Tkachenko <vadim@xxxxxxxxxxx>
Date: Mon, 09 Feb 2009 01:09:06 -0000
Reply-to: Bug 326464 <326464@xxxxxxxxxxxxxxxxxx>
Sender: bounces@xxxxxxxxxxxxx
10 sec output
"not tuned"
  10, 630(0):0.2, 630(0):0.2, 63(0):0.2, 64(0):0.2, 63(0):0.2
  20, 620(0):0.2, 616(0):0.2, 62(0):0.2, 61(0):0.2, 62(0):0.2
  30, 597(0):0.2, 599(0):0.2, 60(0):0.2, 59(0):0.2, 59(0):0.2
  40, 614(0):0.2, 617(0):0.2, 61(0):0.2, 62(0):0.2, 62(0):0.2
  50, 627(0):0.2, 624(0):0.2, 63(0):0.2, 63(0):0.2, 62(0):0.2
  60, 616(0):0.2, 615(0):0.2, 61(0):0.2, 61(0):0.2, 62(0):0.2
  70, 627(0):0.2, 631(0):0.2, 63(0):0.2, 63(0):0.2, 63(0):0.2
  80, 601(0):0.2, 600(0):0.2, 60(0):0.2, 60(0):0.2, 60(0):0.2
  90, 599(0):0.2, 603(0):0.2, 60(0):0.2, 60(0):0.2, 60(0):0.2
 100, 637(0):0.2, 633(0):0.2, 64(0):0.2, 64(0):0.2, 64(0):0.2
 110, 618(0):0.2, 619(0):0.2, 61(0):0.2, 62(0):0.2, 62(0):0.2
 120, 620(0):0.2, 619(0):0.2, 63(0):0.2, 62(0):0.2, 62(0):0.2
 130, 477(0):0.2, 474(0):0.2, 47(0):0.2, 47(0):0.2, 47(0):0.2
 140, 230(0):0.2, 235(0):0.2, 23(0):0.2, 24(0):0.2, 23(0):0.2
 150, 322(0):0.2, 319(0):0.2, 33(0):0.2, 31(0):0.2, 32(0):0.2
 160, 323(0):0.2, 326(0):0.2, 32(0):0.2, 32(0):0.2, 33(0):0.2
 170, 374(0):0.2, 375(0):0.2, 38(0):0.2, 38(0):0.2, 37(0):0.2
 180, 394(0):0.2, 388(0):0.2, 39(0):0.2, 39(0):0.2, 40(0):0.2
 190, 452(0):0.2, 455(0):0.2, 45(0):0.2, 46(0):0.2, 45(0):0.2
 200, 518(0):0.2, 519(0):0.2, 51(0):0.2, 51(0):0.2, 52(0):0.2
 210, 523(0):0.2, 522(0):0.2, 53(0):0.2, 53(0):0.2, 52(0):0.2
 220, 534(0):0.2, 535(0):0.2, 54(0):0.2, 53(0):0.2, 53(0):0.2
 230, 587(0):0.2, 587(0):0.2, 58(0):0.2, 59(0):0.2, 59(0):0.2
 240, 578(0):0.2, 575(0):0.2, 57(0):0.2, 58(0):0.2, 57(0):0.2
 250, 606(0):0.2, 612(0):0.2, 61(0):0.2, 61(0):0.2, 62(0):0.2
 260, 622(0):0.2, 616(0):0.2, 63(0):0.2, 61(0):0.2, 61(0):0.2
 270, 649(0):0.2, 652(0):0.2, 64(0):0.2, 66(0):0.2, 65(0):0.2
 280, 645(0):0.2, 644(0):0.2, 65(0):0.2, 65(0):0.2, 65(0):0.2
 290, 640(0):0.2, 640(0):0.2, 63(0):0.2, 62(0):0.2, 64(0):0.2
 300, 677(0):0.2, 679(0):0.2, 69(0):0.2, 68(0):0.2, 68(0):0.2
 310, 679(0):0.2, 678(0):0.2, 68(0):0.2, 68(0):0.2, 68(0):0.2
 320, 669(0):0.2, 668(0):0.2, 67(0):0.2, 67(0):0.2, 67(0):0.2
 330, 699(0):0.2, 700(0):0.2, 70(0):0.2, 71(0):0.2, 70(0):0.2
 340, 667(0):0.2, 666(0):0.2, 66(0):0.2, 67(0):0.2, 66(0):0.2
 350, 671(0):0.2, 669(0):0.2, 67(0):0.2, 66(0):0.2, 67(0):0.2
 360, 667(0):0.2, 670(0):0.2, 67(0):0.2, 67(0):0.2, 67(0):0.2
 370, 643(0):0.2, 643(0):0.2, 64(0):0.2, 64(0):0.2, 64(0):0.2
 380, 631(0):0.2, 628(0):0.2, 64(0):0.2, 63(0):0.2, 63(0):0.2
 390, 637(0):0.2, 637(0):0.2, 63(0):0.2, 63(0):0.2, 64(0):0.2
 400, 600(0):0.2, 602(0):0.2, 60(0):0.2, 61(0):0.2, 60(0):0.2
 410, 591(0):0.2, 592(0):0.2, 60(0):0.2, 59(0):0.2, 60(0):0.2
 420, 600(0):0.2, 598(0):0.2, 59(0):0.2, 60(0):0.2, 59(0):0.2
 430, 569(0):0.2, 571(0):0.2, 57(0):0.2, 57(0):0.2, 58(0):0.2
 440, 577(0):0.2, 575(0):0.2, 58(0):0.2, 58(0):0.2, 57(0):0.2
 450, 581(0):0.2, 582(0):0.2, 58(0):0.2, 58(0):0.2, 57(0):0.2
 460, 561(0):0.2, 562(0):0.2, 56(0):0.2, 56(0):0.2, 57(0):0.2
 470, 610(0):0.2, 610(0):0.2, 61(0):0.2, 61(0):0.2, 61(0):0.2
 480, 607(0):0.2, 608(0):0.2, 61(0):0.2, 61(0):0.2, 61(0):0.2
 490, 608(0):0.2, 605(0):0.2, 60(0):0.2, 61(0):0.2, 60(0):0.2
 500, 627(0):0.2, 629(0):0.2, 63(0):0.2, 62(0):0.2, 63(0):0.2
 510, 593(0):0.2, 595(0):0.2, 60(0):0.2, 60(0):0.2, 60(0):0.2
 520, 611(0):0.2, 610(0):0.2, 61(0):0.2, 61(0):0.2, 61(0):0.2
 530, 607(0):0.2, 603(0):0.2, 60(0):0.2, 60(0):0.2, 60(0):0.2
 540, 597(0):0.2, 599(0):0.2, 60(0):0.2, 59(0):0.2, 60(0):0.2
 550, 591(0):0.2, 591(0):0.2, 59(0):0.2, 60(0):0.2, 58(0):0.2
 560, 591(0):0.2, 590(0):0.2, 59(0):0.2, 60(0):0.2, 60(0):0.2
 570, 596(0):0.2, 598(0):0.2, 60(0):0.2, 59(0):0.2, 60(0):0.2
 580, 616(0):0.2, 613(0):0.2, 61(0):0.2, 61(0):0.2, 61(0):0.2
 590, 588(0):0.2, 589(0):0.2, 59(0):0.2, 59(0):0.2, 59(0):0.2
 600, 397(0):0.2, 398(0):0.2, 40(0):0.2, 39(0):0.2, 40(0):0.2

"tuned"

  10, 362(0):0.2, 368(0):0.2, 37(0):0.2, 37(0):0.2, 37(0):0.2
  20, 401(0):0.2, 399(0):0.2, 40(0):0.2, 39(0):0.2, 40(0):0.2
  30, 392(0):0.2, 391(0):0.2, 40(0):0.2, 40(0):0.2, 40(0):0.2
  40, 365(0):0.2, 362(0):0.2, 36(0):0.2, 36(0):0.2, 36(0):0.2
  50, 370(0):0.2, 371(0):0.2, 37(0):0.2, 37(0):0.2, 37(0):0.2
  60, 412(0):0.2, 414(0):0.2, 41(0):0.2, 41(0):0.2, 42(0):0.2
  70, 419(0):0.2, 417(0):0.2, 42(0):0.2, 42(0):0.2, 41(0):0.2
  80, 416(0):0.2, 416(0):0.2, 41(0):0.2, 41(0):0.2, 41(0):0.2
  90, 449(0):0.2, 447(0):0.2, 46(0):0.2, 45(0):0.2, 46(0):0.2
 100, 440(0):0.2, 444(0):0.2, 44(0):0.2, 44(0):0.2, 43(0):0.2
 110, 633(0):0.2, 628(0):0.2, 62(0):0.2, 64(0):0.2, 64(0):0.2
 120, 636(0):0.2, 637(0):0.2, 64(0):0.2, 64(0):0.2, 63(0):0.2
 130, 618(0):0.2, 620(0):0.2, 62(0):0.2, 62(0):0.2, 62(0):0.2
 140, 676(0):0.2, 675(0):0.2, 68(0):0.2, 67(0):0.2, 68(0):0.2
 150, 681(0):0.2, 683(0):0.2, 67(0):0.2, 68(0):0.2, 68(0):0.2
 160, 685(0):0.2, 684(0):0.2, 70(0):0.2, 69(0):0.2, 68(0):0.2
 170, 599(0):0.2, 598(0):0.2, 60(0):0.2, 59(0):0.2, 60(0):0.2
 180, 714(0):0.2, 715(0):0.2, 71(0):0.2, 72(0):0.2, 72(0):0.2
 190, 723(0):0.2, 725(0):0.2, 72(0):0.2, 72(0):0.2, 72(0):0.2
 200, 650(0):0.2, 653(0):0.2, 66(0):0.2, 66(0):0.2, 65(0):0.2
 210, 665(0):0.2, 658(0):0.2, 66(0):0.2, 65(0):0.2, 67(0):0.2
 220, 620(0):0.2, 623(0):0.2, 62(0):0.2, 63(0):0.2, 61(0):0.2
 230, 533(0):0.2, 531(0):0.2, 53(0):0.2, 52(0):0.2, 54(0):0.2
 240, 481(0):0.2, 482(0):0.2, 48(0):0.2, 48(0):0.2, 48(0):0.2
 250, 407(0):0.2, 407(0):0.2, 41(0):0.2, 42(0):0.2, 41(0):0.2
 260, 466(0):0.2, 468(0):0.2, 46(0):0.2, 46(0):0.2, 46(0):0.2
 270, 448(0):0.2, 445(0):0.2, 44(0):0.2, 45(0):0.2, 44(0):0.2
 280, 451(0):0.2, 451(0):0.2, 46(0):0.2, 46(0):0.2, 46(0):0.2
 290, 433(0):0.2, 433(0):0.2, 44(0):0.2, 42(0):0.2, 44(0):0.2
 300, 489(0):0.2, 488(0):0.2, 49(0):0.2, 50(0):0.2, 48(0):0.2
 310, 432(0):0.2, 432(0):0.2, 42(0):0.2, 43(0):0.2, 44(0):0.2
 320, 479(0):0.2, 484(0):0.2, 49(0):0.2, 48(0):0.2, 47(0):0.2
 330, 467(0):0.2, 463(0):0.2, 46(0):0.2, 47(0):0.2, 47(0):0.2
 340, 429(0):0.2, 430(0):0.2, 43(0):0.2, 41(0):0.2, 43(0):0.2
 350, 463(0):0.2, 464(0):0.2, 46(0):0.2, 47(0):0.2, 47(0):0.2
 360, 442(0):0.2, 440(0):0.2, 44(0):0.2, 45(0):0.2, 43(0):0.2
 370, 470(0):0.2, 471(0):0.2, 47(0):0.2, 47(0):0.2, 47(0):0.2
 380, 479(0):0.2, 479(0):0.2, 48(0):0.2, 47(0):0.2, 49(0):0.2
 390, 495(0):0.2, 494(0):0.2, 50(0):0.2, 50(0):0.2, 49(0):0.2
 400, 535(0):0.2, 538(0):0.2, 54(0):0.2, 54(0):0.2, 53(0):0.2
 410, 578(0):0.2, 578(0):0.2, 58(0):0.2, 57(0):0.2, 59(0):0.2
 420, 656(0):0.2, 654(0):0.2, 65(0):0.2, 66(0):0.2, 64(0):0.2
 430, 669(0):0.2, 672(0):0.2, 67(0):0.2, 67(0):0.2, 68(0):0.2
 440, 602(0):0.2, 599(0):0.2, 60(0):0.2, 60(0):0.2, 59(0):0.2
 450, 639(0):0.2, 641(0):0.2, 64(0):0.2, 63(0):0.2, 65(0):0.2
 460, 667(0):0.2, 666(0):0.2, 67(0):0.2, 68(0):0.2, 67(0):0.2
 470, 597(0):0.2, 595(0):0.2, 59(0):0.2, 59(0):0.2, 59(0):0.2
 480, 638(0):0.2, 639(0):0.2, 64(0):0.2, 64(0):0.2, 64(0):0.2
 490, 632(0):0.2, 633(0):0.2, 64(0):0.2, 63(0):0.2, 63(0):0.2
 500, 639(0):0.2, 637(0):0.2, 63(0):0.2, 64(0):0.2, 64(0):0.2
 510, 561(0):0.2, 561(0):0.2, 56(0):0.2, 56(0):0.2, 56(0):0.2
 520, 489(0):0.2, 489(0):0.2, 49(0):0.2, 49(0):0.2, 49(0):0.2
 530, 478(0):0.2, 480(0):0.2, 48(0):0.2, 48(0):0.2, 48(0):0.2
 540, 478(0):0.2, 479(0):0.2, 48(0):0.2, 48(0):0.2, 48(0):0.2
 550, 478(0):0.2, 476(0):0.2, 48(0):0.2, 47(0):0.2, 48(0):0.2
 560, 494(0):0.2, 495(0):0.2, 48(0):0.2, 49(0):0.2, 49(0):0.2
 570, 501(0):0.2, 499(0):0.2, 51(0):0.2, 50(0):0.2, 50(0):0.2
 580, 480(0):0.2, 482(0):0.2, 48(0):0.2, 49(0):0.2, 48(0):0.2
 590, 459(0):0.2, 455(0):0.2, 46(0):0.2, 45(0):0.2, 46(0):0.2
 600, 506(0):0.2, 513(0):0.2, 51(0):0.2, 51(0):0.2, 50(0):0.2

-- 
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-XtraDB.

Status in Percona XtraDB Storage Engine for MySQL: New

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 4 threads:
not tuned: 8039.1
tuned: 9111.9

So we have something wrong in 1-thread case.
References

[Bug 326464] [NEW] performance regression in tpc-c 1 thread
From: Vadim Tkachenko, 2009-02-07