[B! benchmark] potato777ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

potato777 id:potato777

benchmarkã«é–¢ã™ã‚‹potato777ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (13)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

ãƒ‡ã‚·ãƒªã‚¢ãƒ©ã‚¤ã‚ºé€Ÿåº¦ã®æ¯”è¼ƒ ByteBuffer vs DirectBuffer vs Unsafe vs C - Blog by Sadayuki Furuhashi
OpenJDK ã‚„ Hotspot VM ã«ã¯ sun.misc.Unsafe ã¨ã„ã†å†…éƒ¨APIãŒã‚ã‚Š*1ã€ã“ã‚Œã‚’ä½¿ã†ã¨ ByteBuffer.getInt ã‚„ ByteBuffer.getLong ã‚ˆã‚Šã‚‚é«˜é€Ÿã«ãƒã‚¤ãƒˆåˆ—ã‹ã‚‰æ•´æ•°å€¤ã‚’ãƒ‡ã‚³ãƒ¼ãƒ‰ã§ãã‚‹ã¨ã„ã†ã€‚ã“ã‚Œã‚’é§†ä½¿ã™ã‚‹ã“ã¨ã§ã€Cã§å®Ÿè£…ã•ã‚ŒãŸæ‹¡å¼µãƒ©ã‚¤ãƒ–ãƒ©ãƒªã«åŒ¹æ•µã™ã‚‹é€Ÿåº¦ã‚’å‡ºã›ã‚‹ã‚‰ã—ã„ã€‚ ãã‚ŒãŒæœ¬å½“ãªã‚‰ã€ãƒ‡ãƒ¼ã‚¿åœ§ç¸®ã‚„ãƒãƒƒã‚·ãƒ¥é–¢æ•°ã€ã‚·ãƒªã‚¢ãƒ©ã‚¤ã‚¶/ãƒ‡ã‚·ãƒªã‚¢ãƒ©ã‚¤ã‚¶ãªã©ã®å®Ÿè£…ã‚’é«˜é€ŸåŒ–ã§ãã‚‹ã€‚ä¾‹ãˆã°ã€lz4 ã‚„ xxhash ã®Javaå®Ÿè£…ãŒ Unsafe API ã‚’ä½¿ç”¨ã—ã¦ã„ã‚‹*2ï¼šjpountz/lz4-java Prestoã‚‚ã€ä¸é–“ãƒ‡ãƒ¼ã‚¿ã®ã‚·ãƒªã‚¢ãƒ©ã‚¤ã‚º/ãƒ‡ã‚·ãƒªã‚¢ãƒ©ã‚¤ã‚ºã«ã¯ã™ã¹ã¦ Unsafe API ã‚’ä½¿ã£ã¦ã„ã‚‹*3ã€‚ ãã“ã§ã€å®Ÿéš›ã«ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã—ã¦ã¿ãŸã€‚ ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯å†…å®¹ 10MBã®ãƒ©ãƒ³ãƒ€ãƒ ãªãƒã‚¤ãƒˆåˆ—ã‚’ç”Ÿæˆã™ã‚‹ å…ˆé ã‹ã‚‰1ãƒã‚¤ãƒˆèªã¿å‡ºã™ ãã®1ãƒ
potato777 2014/03/14
serialize

benchmark
ãƒªãƒ³ã‚¯
ã‚¹ã‚±ãƒ¼ãƒ«å¯èƒ½ãªTwitterã®è² è·ãƒ†ã‚¹ãƒˆã®ä»•çµ„ã¿ [GTAC 2013] - ãƒ¯ã‚¶ãƒŽãƒ | wazanova.jp
[Video]Â http://www.youtube.com/watch?v=99RABfKNfcY [Slide]Â http://goo.gl/9VY2b Twitterã®James WaldropãŒã€ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã®ãƒãƒ¼ãƒ‰ã‚¸ã‚§ãƒãƒ¬ãƒ¼ã‚¿ãƒ©ã‚¤ãƒ–ãƒ©ãƒªlagoã¨ã€ã‚¹ã‚±ãƒ¼ãƒ«ãŒå¯èƒ½ãªè² è·ãƒ†ã‚¹ãƒˆã®ä»•çµ„ã¿ã«ã¤ã„ã¦GTAC (Google Test Automation Conference) 2013ã§èªžã£ã¦ã¾ã™ã€‚Twitterã«ã¯æœ¬ç•ªãƒˆãƒ©ãƒ•ã‚£ãƒƒã‚¯ã®ä¸€éƒ¨ã‚’åˆ©ç”¨ã—ãŸãƒ†ã‚¹ãƒˆã‚‚ã‚ã‚Šã¾ã™ãŒã€æœ¬è¬›æ¼”ã§ã¯ãƒ†ã‚¹ãƒˆç’°å¢ƒã§ã®ãƒãƒ¼ãƒ‰ã‚¸ã‚§ãƒãƒ¬ãƒ¼ã‚¿ã‚’è©±é¡Œã«ã—ã¦ã¾ã™ã€‚ Facebookã‹ã‚‰Twitterã«è»¢è·ã—ã€lagoã‚’ã¤ãã£ãŸç†ç”±ã¯ã€ ã‚ã‚Šã‚ã‚ã›ã®ãƒ„ãƒ¼ãƒ«ã‚’ã‚‚ã£ã¦ãã¦ä½¿ã£ã¦ã‚‚ã ã‚ã§ã€ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã§ã‚ã‚Œã°ä½•ã‹é–‹ç™ºã—ãªã„ã¨ä»–ã®ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã‹ã‚‰å°Šæ•¬ã•ã‚Œãªã„ã€‚ Scalaã§æ›¸ã„ã¦ã¿ãŸã‹ã£ãŸã€‚ æ—¢å˜ã®ãƒãƒ¼ãƒ‰ã‚¸ã‚§ãƒãƒ¬ãƒ¼ã‚¿ã¯ã©ã‚Œã‚‚å½¹ã«ç«‹ãŸãªã„
potato777 2013/10/03
twitter

benchmark
ãƒªãƒ³ã‚¯
è¦‹ã›ã¦ã‚‚ã‚‰ãŠã†ã‹ã€Intelã®AES-NIã®æ€§èƒ½ã¨ã‚„ã‚‰ã‚’ ã€œOpenSSLã§ãƒ™ãƒ³ãƒã—ã¦ã¿ãŸã€œ - ..ãŸã‚Œã‚ã..
AES-NI ã®ä¹—ã£ã¦ã‚‹ãƒžã‚·ãƒ³ãŒä½¿ãˆãŸã®ã§ OpenSSL ã§ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã—ã¦ã¿ãŸã€‚ çµæžœï¼šAES-NI ãªã— 80ã€œ90MB/s AES-NI ã‚ã‚Š 500ã€œ530MB/s ã€Œé€šå¸¸ã®ï¼•å€ã®ã‚¹ãƒ”ãƒ¼ãƒ‰ã§å‡¦ç†ã—ã¦ã¾ã™ï¼ã€ã€Œåœ§å€’çš„ã˜ã‚ƒãªã„ã‹ã€æˆ‘ãŒè»ã¯ã€ ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚·ãƒ§ãƒ³å‘½ä»¤ã®åŠ¹æžœã£ã¦ã™ã”ã„ãï½— ç’°å¢ƒ ãƒ‘ãƒƒã‚±ãƒ¼ã‚¸ openssl-1.0.0-20.el6_2.3.x86_64 OpenSSL 1.0.0-fips 29 Mar 2010 built on: Wed Mar 28 01:11:34 BST 2012 options:bn(64,64) md2(int) rc4(16x,int) des(idx,cisc,16,int) aes(partial) blowfish(idx) compiler: gcc -fPIC -DOPENSSL_PIC -DZLIB -DOPENSSL_THRE
potato777 2013/08/13
cpu

AES

benchmark
ãƒªãƒ³ã‚¯
ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã®çµ‚äº†æ—¥ã‚’2020å¹´1æœˆ31æ—¥(é‡‘)ã«æ±ºå®šã—ã¾ã—ãŸ - ã¯ã¦ãªã®å‘ŠçŸ¥
ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã®çµ‚äº†æ—¥ã‚’2020å¹´1æœˆ31æ—¥(é‡‘)ã«æ±ºå®šã—ã¾ã—ãŸ ä»¥ä¸‹ã®ã‚¨ãƒ³ãƒˆãƒªã®é€šã‚Šã€ä»Šå¹´æœ«ã‚’ç›®å‡¦ã«ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã‚’çµ‚äº†äºˆå®šã§ã‚ã‚‹æ—¨ã‚’ãŠçŸ¥ã‚‰ã›ã—ã¦ãŠã‚Šã¾ã—ãŸã€‚ 2019å¹´æœ«ã‚’ç›®å‡¦ã«ã€ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã®æä¾›ã‚’çµ‚äº†ã™ã‚‹äºˆå®šã§ã™ - ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—æ—¥è¨˜ ã“ã®ãŸã³ã€æ£å¼ã«çµ‚äº†æ—¥ã‚’æ±ºå®šã„ãŸã—ã¾ã—ãŸã®ã§ã€ä»¥ä¸‹ã®é€šã‚Šã”ç¢ºèªãã ã•ã„ã€‚ çµ‚äº†æ—¥: 2020å¹´1æœˆ31æ—¥(é‡‘) ã‚¨ã‚¯ã‚¹ãƒãƒ¼ãƒˆå¸Œæœ›ç”³è«‹æœŸé™:2020å¹´1æœˆ31æ—¥(é‡‘) çµ‚äº†æ—¥ä»¥é™ã¯ã€ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã®é–²è¦§ãŠã‚ˆã³æŠ•ç¨¿ã¯è¡Œãˆã¾ã›ã‚“ã€‚æ—¥è¨˜ã®ã‚¨ã‚¯ã‚¹ãƒãƒ¼ãƒˆãŒå¿…è¦ãªæ–¹ã¯ä»¥ä¸‹ã®è¨˜äº‹ã«ã—ãŸãŒã£ã¦æ‰‹ç¶šãã‚’ã—ã¦ãã ã•ã„ã€‚ ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—ã«æŠ•ç¨¿ã•ã‚ŒãŸæ—¥è¨˜ãƒ‡ãƒ¼ã‚¿ã®ã‚¨ã‚¯ã‚¹ãƒãƒ¼ãƒˆã«ã¤ã„ã¦ - ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—æ—¥è¨˜ ã”åˆ©ç”¨ã®ã¿ãªã•ã¾ã«ã¯ã”è¿·æƒ‘ã‚’ãŠã‹ã‘ã„ãŸã—ã¾ã™ãŒã€ã©ã†ãžã‚ˆã‚ã—ããŠé¡˜ã„ã„ãŸã—ã¾ã™ã€‚ 2020-06-25 è¿½è¨˜ ã¯ã¦ãªã‚°ãƒ«ãƒ¼ãƒ—æ—¥è¨˜ã®ã‚¨ã‚¯ã‚¹ãƒãƒ¼ãƒˆãƒ‡ãƒ¼ã‚¿ã¯2020å¹´2æœˆ28
potato777 2011/09/29
Ruby

benchmark
ãƒªãƒ³ã‚¯
Railsã®å„ç¨®ã‚µãƒ¼ãƒãƒ¼ã®é€Ÿåº¦ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯: Poundã€Varnishã€Nginxã€Unicornã€Mongrel
Railsã®å„ç¨®ã‚µãƒ¼ãƒãƒ¼ã®é€Ÿåº¦ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯: Poundã€Varnishã€Nginxã€Unicornã€Mongrel Railsã®ã‚¢ãƒ—ãƒªã‚’Poundã€Varnishã€Nginxã€Unicornã€Mongrelã‚’ä½¿ç”¨ã—ã¦å‹•ã‹ã—ãŸéš›ã®é€Ÿåº¦ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã€‚ä»¥ä¸‹ã®æ§‹æˆã§ã®åŒä¸€Railsã‚¢ãƒ—ãƒªã‚’httperfã§ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã—ãŸçµæžœã‚’ã¾ã¨ã‚ã¾ã™ã€‚ Pound -> Varnish -> Nginx -> Unicorn * 2 Varnish -> Nginx -> Unicorn * 2 Nginx -> Unicorn * 2 Mongrel (Rails.env == 'development') node.js (Express + Socket.IO) å…¨ä½“ã®æ§‹æˆPoundã‚’80ç•ªãƒãƒ¼ãƒˆã‚’å…¬é–‹ã—ã€Varnishã‚’81ç•ªãƒãƒ¼ãƒˆã€nginxã‚’8080ç•ªãƒãƒ¼ãƒˆã€Unicornã®2ãƒ—ãƒã‚»ã‚¹ã‚’socke
potato777 2011/09/07
rails

nginx

unicorn

varnish

Ruby

node.js

benchmark
ãƒªãƒ³ã‚¯
Linux and H/W optimizations for MySQL
potato777 2011/04/18
mysql

ssd

benchmark
ãƒªãƒ³ã‚¯
Effect from innodb log block size 4096 bytes
All of Perconaâ€™s open-source software products, in one place, to download as much or as little as you need.
potato777 2011/01/05
innodb-log-block-size=4096 ã¨ 512(ãƒ‡ãƒ•ã‚©ãƒ«ãƒˆ)ã®æ¯”è¼ƒã€‚

mysql

benchmark
ãƒªãƒ³ã‚¯
MyISAM mmap feature (5.1)
As you know MyISAM does not cache data, only indexes. MyISAM assumes OS cache is good enough and uses pread/pwrite system calls for reading/writing datafiles. However OS is not always good in this task, my benchmarks show Linux/Solaris arenâ€™t scala ble on intensive pread calls (I believe the same for Windows, but I did not test it). In 5.1 I implemented a new feature: memory mapping for the datafil
potato777 2010/12/14
myisam_use_mmap=1ã‚’ä½¿ç”¨ã—ãŸå ´åˆã¨ã—ã¦ãªã„å ´åˆã®æ¯”è¼ƒ

mysql

benchmark
ãƒªãƒ³ã‚¯
FlashCache: first experiments
I wrote about FlashCache there, and since that I run couple benchmarks, to see what performance benefits we can expect. For initial tries I took sysbench oltp tests ( read-only and read-write) and case when data fully fits into L2 cache. I made binaries for FlashCache for CentOS 5.4, kernel 2.6.18-164.15, you can download it from our testing stage. It took some efforts to make binary, you may get
potato777 2010/05/13
flashcach

mysql

SSD

benchmark
ãƒªãƒ³ã‚¯
MySQL5.5.3-m3ã®DATETIMEåž‹ã®ãƒã‚°ã€‚ã‚ã¨MySQLã® DATETIMEåž‹ã¯æœ¬å½“ã«é…ã„ã®ã‹æ¤œè¨¼ã—ã¦ã¿ãŸ - 2010-04-30 - å°é‡Žãƒžãƒˆãƒšã®æ¥å‹™æ—¥èªŒ(ã‚¢ãƒ‹ãƒ¡åˆ¶ä½œã—ã¦ãªã„ç¯‡)
ãƒã‚°ã®è©± è¿‘ã€…ãµãã¼ã£ãŸãƒ¼DBã®InnoDBåŒ–ã‚’ä¼ã¦ã¦ã„ã‚‹ã®ã§ã€ãã‚Œã«ä¼´ã„MySQL5.0.67(Tritonn)ã‹ã‚‰ã€å…ˆæ—¥ãƒªãƒªãƒ¼ã‚¹ã•ã‚ŒãŸã°ã‹ã‚Šã®MySQL5.5.3-m3ã«ä¹—ã‚Šæ›ãˆã¦ã¿ãŸã€‚RC(ãƒªãƒªãƒ¼ã‚¹å€™è£œ)ç‰ˆã¨ã„ã†ã“ã¨ã§ã€GAç‰ˆã¨ã»ã¼å¤‰ã‚ã‚‰ãªã„å“è³ªã¨èžã„ãŸã®ã§ã€å‰²ã¨è»½ã„æ°—æŒã¡ã§ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã—ãŸã‚“ã ã‘ã©ã€ã„ããªã‚Šãƒã‚°ã«ãƒãƒžã£ãŸã€‚ ãƒã‚°ã¨ã¯ã€DATETIME, TIMESTAMP, DATE, TIMEåž‹ã¨æ–‡å—åˆ—å®šæ•°ã¨ã®çµåˆã§ã‚¤ãƒ³ãƒ‡ãƒƒã‚¯ã‚¹ãŒä½¿ã‚ã‚Œãªã„ã€ã¨ã„ã†ã‚‚ã®ã€‚ ä»¥ä¸‹ã®ã‚ˆã†ãªã€date(DATEåž‹)ã®çµåˆã—ã‹ã—ã¦ã„ãªã„ã‚¯ã‚¨ãƒªã§ã‚‚ã€dateã‚¤ãƒ³ãƒ‡ãƒƒã‚¯ã‚¹ãŒä½¿ã‚ã‚Œãšæ˜‡é †ãƒ•ãƒ«ãƒ†ãƒ¼ãƒ–ãƒ«ã‚¹ã‚ãƒ£ãƒ³ã•ã‚Œã€20ç§’ãã‚‰ã„æŽ›ã‹ã£ãŸã€‚ select date from STATUS force index(date) where date='2010-01-19' limit 10; ã“ã®ç¾è±¡ã¯ã€5.5.3,5
potato777 2010/05/07
mysql

performance

Benchmark
ãƒªãƒ³ã‚¯
PHPãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯: Zend Framework vs Symfony vs CakePHP vs CodeIgniter vs PHP on TRAX - å¾’ç„¶ãªã‚‹ã¾ã¾ã«Blog
potato777 2009/10/22
PHP

benchmark

ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯
ãƒªãƒ³ã‚¯
ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹æƒ…å ±ãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹ã‚·ã‚¹ãƒ†ãƒ (OSS iPedia) ã®ã‚³ãƒ³ãƒ†ãƒ³ãƒ„ã«ã¤ã„ã¦
ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹æƒ…å ±ãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹ã‚·ã‚¹ãƒ†ãƒ (OSS iPedia) ã¯ã€2013å¹´5æœˆ17æ—¥(é‡‘) ã‚’ã‚‚ã¡ã¾ã—ã¦é‹ç”¨ã‚’çµ‚äº†ã„ãŸã—ã¾ã—ãŸã€‚ é•·ã„é–“ã”åˆ©ç”¨ã‚’ã„ãŸã ãã¾ã—ã¦ã‚ã‚ŠãŒã¨ã†ã”ã–ã„ã¾ã—ãŸã€‚ OSS iPediaã§æä¾›ã—ã¦ãŠã‚Šã¾ã—ãŸã€IPA ãƒ•ã‚©ãƒ³ãƒˆã€æ–‡å—æƒ…å ±åŸºç›¤ã€ãã®ä»–å ±å‘Šæ›¸ç‰ã«ã¤ã„ã¦ã¯ã€ä¸‹è¨˜ãƒªãƒ³ã‚¯ã‚’ã”å‚ç…§ãã ã•ã„ã€‚ çš†æ§˜ã«ã¯å¤§å¤‰ã”ä¸ä¾¿ã‚’ãŠã‹ã‘ã„ãŸã—ã¾ã™ãŒã€ä½•å’ã”ç†è§£ã®ç¨‹ã‚’ã‚ˆã‚ã—ããŠé¡˜ã„ç”³ã—ä¸Šã’ã¾ã™ã€‚
potato777 2009/09/11
Count(*) ã¨ SQL_CALC_FOUND_ROWS ã®æ¯”è¼ƒ

mysql

benchmark

innodb
ãƒªãƒ³ã‚¯
RAID vs SSD vs FusionIO
In benchmarks passion (see my two previous posts) I managed to setup all three devices (RAID was on board; Intel X25-E SSD connected to HighPoint controller, FusionIO card) on our working horse Dell PowerEdge R900 (btw, to do that I had to switch from CentOS 5.2 to Ubuntu 8.10, as CentOS was not able to start with attached SSD card to HighPoint controller) and along with other tests I ran tpcc-lik
potato777 2009/06/22
HDD

mysql

benchmark
ãƒªãƒ³ã‚¯
1