Faster than C?

Presentation given at the Prague JS Usergroup.

Transcript

1. felixge
   Faster than C?
   Parsing binary data in JavaScript
   Felix Geisendörfer
   

   View Slide

2. @felixge
   felixge
   Background
   2005 - 2008 2008 - now 2009 - now
   

   View Slide

3. @felixge
   felixge
   The NodeCopter
   nodecopter.com
   

   View Slide

4. felixge
   This Talk
   

   View Slide

5. felixge
   JavaScript vs C
   

   View Slide

6. felixge
   Good vs Evil
   

   View Slide

7. felixge
   Good Parts vs Evil
   

   View Slide

8. felixge
   Bad Parts vs Evil
   

   View Slide

9. felixge
   Faster than C?
   

   View Slide

10. felixge
    No
    Sorry for the “title bait”
    

    View Slide

11. felixge
    As fast as C bindings
    For parsing binary protocols / database drivers
    

    View Slide

12. felixge
    The Story
    

    View Slide

13. felixge
    early 2010
    

    View Slide

14. felixge
    No MySQL module
    for node.js
    early 2010
    

    View Slide

15. felixge
    

    View Slide

16. felixge
    Pure JS / No C/C++
    

    View Slide

17. felixge
    Before Buffers
    became usable
    

    View Slide

18. felixge
    The Parser was using
    JavaScript Strings
    

    View Slide

19. felixge
    Node.js Trivia
    

    View Slide

20. felixge
    “ Buffers” used to be
    called “Blobs”
    

    View Slide

21. felixge
    For 3 min and 15 sec
    

    View Slide

22. felixge
    RIP Blobs
    ✞
    

    View Slide

23. felixge
    Sun Dec 13 08:39:20 2009
    -
    Sun Dec 13 08:42:45 2009
    

    View Slide

24. felixge
    Anyway
    

    View Slide

25. felixge
    mysql can be done
    without libmysql
    

    View Slide

26. felixge
    

    View Slide

27. felixge
    No good deed goes
    unpunished
    

    View Slide

28. felixge
    Sir Isaac Newton
    

    View Slide

29. felixge
    Third Law of Motion
    

    View Slide

30. “When a first body exerts a
    force F1 on a second body, the
    second body simultaneously
    exerts a force F2 = −F1 on the
    first body. This means that F1
    and F2 are equal in magnitude
    and opposite in direction.”
    

    View Slide

31. felixge
    Third Law of Github
    

    View Slide

32. “When a first person pushes
    a library L1 into a remote
    repository, a second person
    simultaneously starts working
    on a second library L2 which
    will be equally awesome, but
    in a different way.”
    

    View Slide

33. felixge
    <3 Github!
    

    View Slide

34. felixge
    

    View Slide

35. @felixge
    felixge
    Benchmark
    • Parse ~180 MB / 100.000 rows of MySQL result data
    • 5 Columns: id, title, text, created, updated
    • -> create 100k objects with 500k keys + 500k values
    

    View Slide

36. felixge
    0
    500
    1,000
    1,500
    mysql−0.9.6 mysql−libmysqlclient−1.5.1
    benchmark
    mbit
    benchmark
    mysql−0.9.6
    mysql−libmysqlclient−1.5.1
    

    View Slide

37. felixge
    Of course.
    

    View Slide

38. felixge
    libmysql = C
    

    View Slide

39. felixge
    my library = JavaScript
    

    View Slide

40. felixge
    C > JS, right?
    

    View Slide

41. felixge
    But V8!
    

    View Slide

42. felixge
    And Crankshaft!!
    

    View Slide

43. felixge
    Node.js !!!1!
    

    View Slide

44. felixge
    Was I
    living a lie?
    

    View Slide

45. felixge
    Kind of
    

    View Slide

46. felixge
    V8 / Node
    = Tools
    

    View Slide

47. felixge
    Performance
    is not a tool
    

    View Slide

48. felixge
    Performance is
    hard work &
    data analysis
    

    View Slide

49. felixge
    0
    500
    1,000
    1,500
    mysql−0.9.6 mysql−libmysqlclient−1.5.1mysql−2.0.0−alpha3
    benchmark
    mbit
    benchmark
    mysql−0.9.6
    mysql−libmysqlclient−1.5.1
    mysql−2.0.0−alpha3
    

    View Slide

50. felixge
    Third Law of Github
    

    View Slide

51. felixge
    

    View Slide

52. felixge
    0
    1,000
    2,000
    3,000
    4,000
    mysql−0.9.6
    mysql−libmysqlclient−1.5.1
    mysql−2.0.0−alpha3 mariadb−0.1.7
    benchmark
    mbit
    benchmark
    mysql−0.9.6
    mysql−libmysqlclient−1.5.1
    mysql−2.0.0−alpha3
    mariadb−0.1.7
    

    View Slide

53. felixge
    Time to give up?
    

    View Slide

54. felixge
    NEVER!
    

    View Slide

55. felixge
    New Parser
    

    View Slide

56. @felixge
    felixge
    0
    2,000
    4,000
    6,000
    mysql2 new−parser
    benchmark
    mbit
    benchmark
    mysql2
    new−parser
    

    View Slide

57. felixge
    Third law of Github?
    

    View Slide

58. felixge
    Endgame
    

    View Slide

59. felixge
    Last bottleneck:
    V8 / Creating JS Objects
    

    View Slide

60. felixge
    Also: MySQL Server saturated
    

    View Slide

61. felixge
    Anyway
    

    View Slide

62. felixge
    How to write fast JS
    

    View Slide

63. felixge
    Does not work
    

    View Slide

64. @felixge
    felixge
    Profiling
    • Good at telling you which functions are slow
    • Bad at telling you how to fix it
    

    View Slide

65. felixge
    Taking performance advice
    from strangers
    

    View Slide

66. @felixge
    felixge
    Taking performance advice
    from strangers
    • Good for ideas & inspiration
    • But useless when applied cargo-cult style
    

    View Slide

67. @felixge
    felixge
    for (var i = 0; i < array.length; i++) {
    // do some work with array[i]
    }
    

    View Slide

68. @felixge
    felixge
    for (var i = 0, length = array.length; i < length; i++) {
    // do some work with array[i]
    }
    

    View Slide

69. felixge
    What Does Work?
    

    View Slide

70. felixge
    Benchmark Driven
    Development
    

    View Slide

71. @felixge
    felixge
    Benchmark Driven Development
    • Similar to test driven development
    • Use it when performance is an explicit design goal
    • Benchmark first > benchmark after !
    

    View Slide

72. felixge
    1 function benchmark() {
    2 // intentionally empty
    3 }
    

    View Slide

73. felixge
    1 while (true) {
    2 var start = Date.now();
    3 benchmark();
    4 var duration = Date.now() - start;
    5 console.log(duration);
    6 }
    

    View Slide

74. @felixge
    felixge
    Benchmark Driven Development
    • Next step: Implement a tiny part of your function
    • Example: Parse headers of MySQL packets
    • Look at impact, tweak code, repeat
    

    View Slide

75. @felixge
    felixge
    Example Results
    • try...catch is ok
    • big switch statement = bad
    • function calls = very cheap
    • buffering is ok
    

    View Slide

76. @felixge
    felixge
    Favorite Optimization:
    Loop unrolling with eval()’s
    good twin new Function()
    

    View Slide

77. @felixge
    felixge
    1 function parseRow(columns, parser) {
    2 var row = {};
    3 for (var i = 0; i < columns.length; i++) {
    4 row[columns[i].name] = parser.readColumnValue();
    5 }
    6 return row;
    7 }
    

    View Slide

78. @felixge
    felixge
    1 function parseRow(columns, parser) {
    2 return {
    3 id : parser.readColumnValue(),
    4 title : parser.readColumnValue(),
    5 body : parser.readColumnValue(),
    6 created : parser.readColumnValue(),
    7 updated : parser.readColumnValue(),
    8 };
    9 }
    

    View Slide

79. @felixge
    felixge
    How can we unroll this loop at
    runtime?
    

    View Slide

80. @felixge
    felixge
    1 var code = 'return {\n';
    2
    3 columns.forEach(function(column) {
    4 code += '"' + column.name + '":' + 'parser.readColumnValue(),\n';
    5 });
    6
    7 code += '};\n';
    8
    9 var parseRow = new Function('columns', 'parser', code);
    

    View Slide

81. PLEASE DO NOT
    REMEMBER THIS
    -
    DO YOUR OWN
    BENCHMARKS!!!
    

    View Slide

82. @felixge
    felixge
    Data Analysis
    

    View Slide

83. @felixge
    felixge
    Data analysis
    • Produce data points as tab separated values
    • Add as many VM/OS metrics as you can get to every line
    • Do not mix data and analysis !!
    

    View Slide

84. @felixge
    felixge
    Recommended Tools
    • node benchmark.js | tee results.tsv
    • R Programming language (with ggplot2) !
    • Makefiles, Image Magick, Skitch
    

    View Slide

85. @felixge
    felixge
    Why?
    

    View Slide

86. @felixge
    felixge
    0
    2,000
    4,000
    6,000
    mysql2 new−parser
    benchmark
    mbit
    benchmark
    mysql2
    new−parser
    

    View Slide

87. @felixge
    felixge
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ● ● ●
    ● ● ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ● ● ●●
    ● ● ●
    ● ● ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ●
    ●
    ● ● ●
    ● ●
    ● ● ● ●
    ●
    ● ●
    ● ● ●
    ● ● ●
    ● ●
    ●
    ● ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ● ●
    ●
    ● ● ●
    ●
    ● ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ● ●●
    ● ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●● ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●● ● ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●
    ● ● ●
    ● ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ● ● ●
    ● ●●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ● ●
    ● ● ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ●
    ● ● ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ●
    ● ●
    ●
    ● ●●
    ● ● ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    3,000
    4,000
    5,000
    6,000
    mysql2 new−parser
    benchmark
    mbit
    benchmark
    ●
    ●
    mysql2
    new−parser
    

    View Slide

88. @felixge
    felixge
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ● ● ●
    ● ● ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ● ● ●●
    ● ● ●
    ● ● ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ● ●
    ●
    ●
    ● ● ●
    ● ●
    ● ● ● ●
    ●
    ● ●
    ● ● ●
    ● ● ●
    ● ●
    ●
    ● ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ● ●
    ●
    ● ● ●
    ●
    ● ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ● ●●
    ● ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●● ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●● ● ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●
    ● ● ●
    ● ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ● ●
    ● ● ●
    ● ●
    ●
    ●
    ●
    ● ● ●
    ● ●●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ● ●
    ● ● ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ● ● ●
    ● ● ●
    ●
    ● ●
    ● ●
    ● ● ●
    ●
    ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ●
    ●
    ● ●
    ●
    ● ●●
    ● ● ●
    ● ●
    ● ●
    ●
    ● ●
    ●
    ● ●
    ●
    ●
    ● ● ● ●
    ●
    ● ●
    ● ●
    ●
    ●
    ● ●
    ● ●
    ● ●
    ● ●
    ●
    ● ● ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ●
    ● ●
    ●
    ●
    ●
    ● ●
    3,000
    4,000
    5,000
    6,000
    mysql2 new−parser
    benchmark
    mbit
    benchmark
    ●
    ●
    mysql2
    new−parser
    Dafuq?
    Dafuq?
    

    View Slide

89. @felixge
    felixge
    3,000
    4,000
    5,000
    6,000
    0 100 200 300
    number
    mbit
    benchmark
    mysql2
    new−parser
    

    View Slide

90. @felixge
    felixge
    10
    20
    30
    0 100 200 300
    number
    Heap Total (MB)
    benchmark
    mysql2
    new−parser
    

    View Slide

91. @felixge
    felixge
    5
    10
    15
    0 100 200 300
    number
    Heap Used (MB)
    benchmark
    mysql2
    new−parser
    

    View Slide

92. @felixge
    felixge
    tl;dr
    

    View Slide

93. @felixge
    felixge
    1. Write a benchmark
    2. Write/change a little code
    3. Collect data
    4. Find problems
    5. Goto 2
    

    View Slide

94. @felixge
    felixge
    Thank you
    Felix Geisendörfer
    

    View Slide

95. @felixge
    felixge
    github.com/felixge/faster-than-c
    All benchmarks, results and analysis scripts
    Felix Geisendörfer
    

    View Slide