Watt-Wise Web: Architecting a Responsive and Energy-Efficient Mobile Web

Guest lecture given at University of Utah on Web browsing

Transcript

1. Watt-Wise Web:
   Architecting a Responsive and
   Energy-Efficient Mobile Web
   Yuhao Zhu
   http://yuhaozhu.com
   

   View Slide

2. View Slide

3. View Slide

4. Snake
   circa 2000
   

   View Slide

5. Snake
   circa 2000
   Snake
   circa 2020
   

   View Slide

6. Performance
   

   View Slide

7. Performance
   Power
   

   View Slide

8. Measure Power (W)
   0
   2
   4
   6
   8
   Year
   2009 2010 2011 2012 2013 2014 2015
   Mobile CPU’s Rise to Power [HPCA 2016]
   3
   

   View Slide

9. Measure Power (W)
   0
   2
   4
   6
   8
   Year
   2009 2010 2011 2012 2013 2014 2015
   0
   2
   4
   6
   8
   2009 2010 2011 2012 2013 2014 2015
   Mobile CPU’s Rise to Power [HPCA 2016]
   3
   

   View Slide

10. Measure Power (W)
    0
    2
    4
    6
    8
    Year
    2009 2010 2011 2012 2013 2014 2015
    0
    2
    4
    6
    8
    2009 2010 2011 2012 2013 2014 2015
    Mobile CPU’s Rise to Power [HPCA 2016]
    3
    In pursuit of high
    performance
    

    View Slide

11. Measure Power (W)
    0
    2
    4
    6
    8
    Year
    2009 2010 2011 2012 2013 2014 2015
    0
    2
    4
    6
    8
    2009 2010 2011 2012 2013 2014 2015
    Mobile CPU’s Rise to Power [HPCA 2016]
    3
    Throttling
    In pursuit of high
    performance
    

    View Slide

12. Measure Power (W)
    0
    2
    4
    6
    8
    Year
    2009 2010 2011 2012 2013 2014 2015
    0
    2
    4
    6
    8
    2009 2010 2011 2012 2013 2014 2015
    Mobile CPU’s Rise to Power [HPCA 2016]
    3
    Throttling
    In pursuit of high
    performance
    SoC Thermal Design Power (TDP)
    

    View Slide

13. 4
    Mobile Processor Design “Strategy”
    Performance
    Power
    

    View Slide

14. 4
    Mobile Processor Design “Strategy”
    2007
    (In-order)
    Performance
    Power
    

    View Slide

15. 4
    Mobile Processor Design “Strategy”
    2007
    (In-order)
    2011
    (Out-of-order)
    2012
    (Multi-core)
    2013
    
    (Asymmetric
    Multi-core)
    Performance
    Power
    

    View Slide

16. 4
    Mobile Processor Design “Strategy”
    2007
    (In-order)
    2011
    (Out-of-order)
    2012
    (Multi-core)
    2013
    
    (Asymmetric
    Multi-core)
    Performance
    Power
    Squeeze 3 decades of desktop
    CPU techniques into a 6-year span
    

    View Slide

17. “Improving” Energy Capacity
    5
    600 smartphone from 2006 to 2014 on http://www.gsmarena.com/makers.php3
    

    View Slide

18. “Improving” Energy Capacity
    5
    Screen Size (inches)
    Battery Capacity (mAh)
    600 smartphone from 2006 to 2014 on http://www.gsmarena.com/makers.php3
    
    
    
    
    
    
    
    
    
    
    
    
    

    View Slide

19. “Improving” Energy Capacity
    5
    Screen Size (inches)
    Battery Capacity (mAh)
    600 smartphone from 2006 to 2014 on http://www.gsmarena.com/makers.php3
    
    
    
    
    
    
    
    
    
    
    
    
    

    View Slide

20. “Improving” Energy Capacity
    5
    Screen Size (inches)
    Battery Capacity (mAh)
    600 smartphone from 2006 to 2014 on http://www.gsmarena.com/makers.php3
    
    
    
    
    
    
    
    
    
    
    
    
    

    View Slide

21. “Improving” Energy Capacity
    6
    

    View Slide

22. “Improving” Energy Capacity
    6
    

    View Slide

23. “Improving” Energy Capacity
    6
    

    View Slide

24. “Improving” Energy Capacity
    6
    

    View Slide

25. “Improving” Energy Capacity
    6
    

    View Slide

26. Mobile Applications
    7
    

    View Slide

27. Mobile Applications
    7
    ^
    Web
    

    View Slide

28. Mobile Applications
    8
    ^
    Web
    Monthly Unique Mobile Users
    Non-Web Web
    3.3 M
    8.9 M
    comScore Mobile Metrix, U.S., June 2015
    

    View Slide

29. Mobile Applications
    8
    ^
    Web
    Monthly Unique Mobile Users
    2
    4
    6
    8
    10
    12
    Jun

    2014
    Sep

    2014
    Dec

    2014
    Mar

    2015
    Jun

    2015
    Sep

    2015
    Dec

    2015
    Mar

    2016
    Jun

    2016
    Million
    Web
    Non-Web
    comScore 2016 U.S. Mobile App Report
    

    View Slide

30. Is This (Just) a Network Issue? [IEEE MICRO 2015]
    9
    

    View Slide

31. 38
    32
    26
    20
    14
    8
    2
    Load time (s)
    10
    2 3 4 5 6 7 8
    100
    2 3 4 5 6 7 8
    1000
    2
    Network RTT (ms)
    10
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

32. 38
    32
    26
    20
    14
    8
    2
    Load time (s)
    10
    2 3 4 5 6 7 8
    100
    2 3 4 5 6 7 8
    1000
    2
    Network RTT (ms)
    10
    LTE 3G Adverse 3G
    2G
    Wi-Fi
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

33. 38
    32
    26
    20
    14
    8
    2
    Load time (s)
    10
    2 3 4 5 6 7 8
    100
    2 3 4 5 6 7 8
    1000
    2
    Network RTT (ms)
    10
    LTE 3G Adverse 3G
    2G
    Wi-Fi
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

34. 38
    32
    26
    20
    14
    8
    2
    Load time (s)
    10
    2 3 4 5 6 7 8
    100
    2 3 4 5 6 7 8
    1000
    2
    Network RTT (ms)
    10
    LTE 3G Adverse 3G
    2G
    Wi-Fi
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

35. 38
    32
    26
    20
    14
    8
    2
    Load time (s)
    10
    2 3 4 5 6 7 8
    100
    2 3 4 5 6 7 8
    1000
    2
    Network RTT (ms)
    10
    Compute
    LTE 3G Adverse 3G
    2G
    Wi-Fi
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

36. Measured Power (W)
    0.0
    3.0
    6.0
    9.0
    2009
    2010
    2011
    2012
    2013
    2014
    2015
    Screen Radio CPU
    11
    Compute
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    

    View Slide

37. 12
    Compute
    Is This (Just) a Network Issue? [IEEE MICRO 2015]
    Measured Power (W)
    0.0
    3.0
    6.0
    9.0
    2009
    2010
    2011
    2012
    2013
    2014
    2015
    Screen Radio CPU
    

    View Slide

38. Compute
    13
    Web Browsing from a Compute Perspective
    

    View Slide

39. 14
    Web Browsing from a Compute Perspective
    Compute
    

    View Slide

40. Frameworks and Libraries
    HTML JavaScript
    CSS
    Language Runtime
    Styling
    Security
    Local
    Storage
    User
    Input
    Layout
    Render
    Runtime
    15
    Web Browsing from a Compute Perspective
    Compute
    Architecture
    Application
    

    View Slide

41. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    

    View Slide

42. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    [ISCA 2014]
    [TOCS 2017]
    

    View Slide

43. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    GreenWeb
    Language Support
    for Quality-of-Experience
    [PLDI 2016]
    [ISCA 2014]
    [TOCS 2017]
    

    View Slide

44. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    GreenWeb
    Language Support
    for Quality-of-Experience
    [PLDI 2016]
    [ISCA 2014]
    [TOCS 2017]
    

    View Slide

45. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    GreenWeb
    Language Support
    for Quality-of-Experience
    [PLDI 2016]
    [ISCA 2014]
    [TOCS 2017]
    

    View Slide

46. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    WebRT
    Fast, Energy-Efficient
    Mobile Web Runtime
    GreenWeb
    Language Support
    for Quality-of-Experience
    [PLDI 2016]
    [ISCA 2014]
    [TOCS 2017]
    [HPCA 2013]
    [HPCA 2015]
    [ISCA 2019]
    

    View Slide

47. Runtime
    16
    Cross-Layer Optimizations
    Architecture
    Application
    WebCore
    Web-specific
    Processor Architecture
    WebRT
    Fast, Energy-Efficient
    Mobile Web Runtime
    GreenWeb
    Language Support
    for Quality-of-Experience
    [PLDI 2016]
    [ISCA 2014]
    [TOCS 2017]
    [HPCA 2013]
    [HPCA 2015]
    [ISCA 2019]
    

    View Slide

48. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions
    

    View Slide

49. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    

    View Slide

50. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    click
    touchstart
    touchmove
    scroll
    

    View Slide

51. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    click
    touchstart
    touchmove
    scroll
    Event Queue
    

    View Slide

52. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    click
    touchstart
    touchmove
    scroll
    Event
    Loop
    Event Queue
    CPU
    

    View Slide

53. 17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    click
    touchstart
    touchmove
    scroll
    Event is the atomic unit of execution; optimize
    latency/energy at the event-level.
    Event
    Loop
    Event Queue
    CPU
    

    View Slide

54. ▸ Observation: Events have different latency
    slacks that enable energy optimizations
    17
    Mobile Applications are Event-Driven
    Touching
    Moving
    Interactions Events
    click
    touchstart
    touchmove
    scroll
    Event
    Loop
    Event Queue
    CPU
    

    View Slide

55. ▸ Observation: Events have different latency
    slacks that enable energy optimizations
    18
    Event-based Web Runtime Optimization
    

    View Slide

56. ▸ Observation: Events have different latency
    slacks that enable energy optimizations
    18
    ▸ Mechanism: Provide just enough energy to
    meet QoE requirement for different events
    Event-based Web Runtime Optimization
    

    View Slide

57. ▸ Observation: Events have different latency
    slacks that enable energy optimizations
    18
    ▸ Mechanism: Provide just enough energy to
    meet QoE requirement for different events
    ▸ Implementation: Map events to different
    heterogeneous hardware configurations
    Event-based Web Runtime Optimization
    

    View Slide

58. Event-Level Characterization
    !19
    

    View Slide

59. Event-Level Characterization
    !19
    

    View Slide

60. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    

    View Slide

61. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    

    View Slide

62. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    keyup
    

    View Slide

63. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    keyup
    

    View Slide

64. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    keyup
    

    View Slide

65. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    change
    keyup
    

    View Slide

66. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    change
    Small
    
    Slack
    keyup
    

    View Slide

67. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    change
    Small
    
    Slack
    click
    keyup
    

    View Slide

68. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    change
    Small
    
    Slack
    No Slack
    click
    keyup
    

    View Slide

69. Event-Level Characterization
    !19
    150
    100
    50
    0
    Event Latency (ms)
    Events
    Large Slack
    change
    Small
    
    Slack
    No Slack
    click
    keyup
    ▸ Wide distribution of event latencies.
    Events exhibit different slacks.
    ▹ How to exploit event slacks?
    

    View Slide

70. !20
    Event-based Scheduler (EBS)
    

    View Slide

71. !20
    Event-based Scheduler (EBS)
    ▸ Goal: For each event, find the most energy-efficient
    hardware configuration that meets the latency target
    

    View Slide

72. !20
    Event-based Scheduler (EBS)
    Thread Scheduling
    

    View Slide

73. !20
    Event-based Scheduler (EBS)
    Thread Scheduling
    

    View Slide

74. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    

    View Slide

75. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    Throughput Fairness
    

    View Slide

76. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    Throughput Fairness
    Events-based Scheduling
    

    View Slide

77. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    Throughput Fairness
    Events-based Scheduling
    Event Queue
    

    View Slide

78. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    Throughput Fairness
    Event-based Scheduler
    Events-based Scheduling
    Event Queue
    

    View Slide

79. !20
    Event-based Scheduler (EBS)
    Thread-based Scheduler
    Thread Scheduling
    Throughput Fairness
    Event-based Scheduler
    Events-based Scheduling
    Event
    Latency
    Event
    Energy
    Event Queue
    

    View Slide

80. !21
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    ▸ Widely used in commodity devices
    

    View Slide

81. !22
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    ▸ Widely used in commodity devices
    Energy Consumption
    Performance
    Big Core
    Small Core
    

    View Slide

82. !22
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    ▸ Widely used in commodity devices
    Energy Consumption
    Performance
    Big Core
    Small Core
    Voltage/
    Frequency
    Levels
    

    View Slide

83. !22
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    ▸ Widely used in commodity devices
    Energy Consumption
    Performance
    Big Core
    Small Core
    

    View Slide

84. !22
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    ▸ Widely used in commodity devices
    Energy Consumption
    Performance
    Big Core
    Small Core
    

    View Slide

85. !23
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    Memory
    Operation
    CPU
    Operation
    Tmemory
    Ndependent
    f
    Event
    Latency
    Xie, et al., Compile-Time Dynamic Voltage Scaling Settings: Opportunities and Limits, PLDI’03
    

    View Slide

86. !23
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    Memory
    Operation
    CPU
    Operation
    Tmemory
    Ndependent
    f
    Event
    Latency
    Xie, et al., Compile-Time Dynamic Voltage Scaling Settings: Opportunities and Limits, PLDI’03
    Event
    Latency
    =
    

    View Slide

87. !23
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    Memory
    Operation
    CPU
    Operation
    Tmemory
    Ndependent
    f
    Event
    Latency
    Xie, et al., Compile-Time Dynamic Voltage Scaling Settings: Opportunities and Limits, PLDI’03
    Event
    Latency
    = Tmemory
    +
    

    View Slide

88. !23
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    Memory
    Operation
    CPU
    Operation
    Tmemory
    Ndependent
    f
    Event
    Latency
    Xie, et al., Compile-Time Dynamic Voltage Scaling Settings: Opportunities and Limits, PLDI’03
    Event
    Latency
    = Tmemory
    + Ndependent / f
    

    View Slide

89. !23
    Leveraging Heterogeneous Hardware
    ▸ Offer a large performance-energy trade-off space
    Memory
    Operation
    CPU
    Operation
    Tmemory
    Ndependent
    f
    Event
    Latency
    Xie, et al., Compile-Time Dynamic Voltage Scaling Settings: Opportunities and Limits, PLDI’03
    Event
    Latency
    = Tmemory
    + Ndependent / f
    

    View Slide

90. Evaluation Methodology
    ▸ Implemented inside Google Chromium Web browser
    ▸ Representative hardware platform
    ▹ Exynos 5410 SoC (A15 + A7)
    ▸UI-level record and replay for reproducibility.
    ▹ Mosaic [ISPASS 2015] https://github.com/Matthalp/mosaic
    24
    

    View Slide

91. 25
    Evaluation Results
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    

    View Slide

92. 26
    Evaluation Results
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    Android
    

    View Slide

93. 0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    Android
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    EBS
    27
    Evaluation Results
    29.2%
    Energy Savings
    

    View Slide

94. 0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    Android
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    EBS
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    28
    Evaluation Results
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    

    View Slide

95. 0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    Android
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    EBS
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    29
    Evaluation Results
    

    View Slide

96. 0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    Android
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    0
    0.2
    0.4
    0.6
    0.8
    1
    EBS
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Perf
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    Norm. QoS Violation
    0
    0.2
    0.4
    0.6
    0.8
    1
    1.2
    Norm. Energy
    0
    0.2
    0.4
    0.6
    0.8
    1
    30
    Evaluation Results
    0.8% More
    QoS violations
    

    View Slide

97. Rethink Event-based Scheduling
    31
    Event-based Scheduler
    Event
    Latency
    Event
    Energy
    Event Queue
    

    View Slide

98. Rethink Event-based Scheduling
    31
    Event-based Scheduler
    Event
    Latency
    Event
    Energy
    Pending Events
    Event Queue
    

    View Slide

99. Rethink Event-based Scheduling
    31
    Event-based Scheduler
    Event
    Latency
    Event
    Energy
    Pending Events
    onclick
    GC task
    Timer event
    Event Queue
    Time
    Event
    Queue
    Pending
    Events
    

    View Slide

100. Rethink Event-based Scheduling
     31
     Event-based Scheduler
     Event
     Latency
     Event
     Energy
     Reactive Strategy
     Pending Events
     onclick
     GC task
     Timer event
     Event Queue
     Time
     Event
     Queue
     Pending
     Events
     

     View Slide

101. Rethink Event-based Scheduling
     31
     Event-based Scheduler
     Event
     Latency
     Event
     Energy
     Reactive Strategy
     Pending Events
     onclick
     GC task
     Timer event
     ontouchmove
     onsubmit
     Event Queue
     Time
     Event
     Queue
     Pending
     Events
     

     View Slide

102. Rethink Event-based Scheduling
     31
     Event-based Scheduler
     Event
     Latency
     Event
     Energy
     Proactive Strategy
     Pending &
     Speculative Events
     onclick
     GC task
     Timer event
     ontouchmove
     onsubmit
     Event Queue
     Time
     Event
     Queue
     Pending
     Events
     Speculative
     Events
     

     View Slide

103. Inefficiency of Current Schedulers
     32
     

     View Slide

104. Inefficiency of Current Schedulers
     32
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     

     View Slide

105. Inefficiency of Current Schedulers
     32
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1
     OS
     

     View Slide

106. Inefficiency of Current Schedulers
     32
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 slack
     OS
     

     View Slide

107. Inefficiency of Current Schedulers
     32
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     slack
     OS
     

     View Slide

108. Inefficiency of Current Schedulers
     33
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     EBS
     OS
     

     View Slide

109. Inefficiency of Current Schedulers
     33
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     E1
     EBS
     OS
     

     View Slide

110. Inefficiency of Current Schedulers
     33
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     E1 E2
     EBS
     OS
     

     View Slide

111. Inefficiency of Current Schedulers
     33
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     E1 E2
     ?
     EBS
     OS
     

     View Slide

112. Inefficiency of Current Schedulers
     34
     Time
     Oracle
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     E1 E2
     EBS
     OS
     

     View Slide

113. Inefficiency of Current Schedulers
     34
     Time
     Oracle
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     E1 E2
     E1 E2
     EBS
     OS
     

     View Slide

114. Inefficiency of Current Schedulers
     35
     Time
     input 1 input 2
     QoS
     
     Deadline 1
     QoS
     
     Deadline 2
     E1 E2
     Schedule across both
     pending and future events
     Oracle
     

     View Slide

115. PES: Proactive Event Scheduler
     36
     

     View Slide

116. PES: Proactive Event Scheduler
     36
     Prediction
     Web Program
     
     Analysis
     Machine
     
     Learning
     

     View Slide

117. PES: Proactive Event Scheduler
     36
     Prediction
     Web Program
     
     Analysis
     Machine
     
     Learning
     Scheduling
     Constrained Optimization
     

     View Slide

118. PES: Proactive Event Scheduler
     36
     Prediction
     Web Program
     
     Analysis
     Machine
     
     Learning
     Scheduling
     Constrained Optimization
     

     View Slide

119. Event Sequence Learning
     37
     Prediction
     Model
     

     View Slide

120. Event Sequence Learning
     37
     Time
     E1 E2 E3
     Event Sequence
     Prediction
     Model
     

     View Slide

121. Event Sequence Learning
     37
     Time
     E1 E2 E3
     Event Sequence
     Prediction
     Model
     

     View Slide

122. Event Sequence Learning
     37
     Time
     E1 E2 E3
     Event Sequence
     E4
     Prediction
     Model
     

     View Slide

123. Event Sequence Learning
     37
     Time
     E1 E2 E3
     Event Sequence
     E4
     Prediction
     Model
     

     View Slide

124. Recurrent Prediction
     38
     Time
     E1 E2 E3 E4
     Event Sequence
     Prediction
     Model
     

     View Slide

125. Recurrent Prediction
     38
     Time
     E1 E2 E3 E4
     Event Sequence
     Prediction
     Model
     

     View Slide

126. Recurrent Prediction
     38
     Time
     E1 E2 E3 E4
     Event Sequence
     E5
     Prediction
     Model
     

     View Slide

127. Recurrent Prediction
     38
     Time
     E1 E2 E3 E4
     Event Sequence
     E5
     Prediction
     Model
     

     View Slide

128. Recurrent Prediction
     38
     Time
     E1 E2 E3 E4
     Event Sequence
     E5
     …
     Prediction
     Model
     

     View Slide

129. Prediction Model
     39
     Prediction
     Model
     

     View Slide

130. Prediction Model
     40
     Prediction
     Model
     The distance of click
     The number of scrolls
     The number of navigations
     Features
     
     encoding
     
     past
     interactions
     …
     

     View Slide

131. Prediction Model
     41
     Features
     
     encoding
     
     past
     
     interactions
     ln(
     p
     1 − p
     ) =
     x
     β
     Prediction
     Model
     

     View Slide

132. Prediction Model
     42
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Prediction
     Model
     0.10
     0.12
     0.58
     0.07
     0.02
     

     View Slide

133. Prediction Model
     42
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Prediction
     Model
     ✔
     

     View Slide

134. Prediction Model
     42
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Prediction
     Model
     All Event Types
     

     View Slide

135. Prediction Model
     42
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Prediction
     Model
     Filter Unlikely Events!
     

     View Slide

136. 43
     Program State Analysis
     

     View Slide

137. 43
     Program State Analysis
     

     View Slide

138. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Program State Analysis
     DOM Tree
     

     View Slide

139. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Program State Analysis
     DOM Tree
     Viewport
     

     View Slide

140. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Viewport
     Program State Analysis
     DOM Tree
     Viewport
     

     View Slide

141. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Viewport
     Program State Analysis
     DOM Tree
     Viewport
     

     View Slide

142. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Viewport
     Program State Analysis
     DOM Tree
     Viewport
     

     View Slide

143. 43
     
     
     
     
     
     …
     
     
     
     
     …
     Viewport
     Program State Analysis
     DOM Tree
     Viewport
     

     View Slide

144. Integrating Program States
     44
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Prediction
     Model
     

     View Slide

145. Integrating Program States
     45
     Features
     
     encoding
     
     past
     
     interactions
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     Current

     application
     
     states
     +
     +
     Prediction
     Model
     

     View Slide

146. Overview of the Predictor
     46
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     +
     Prediction
     Model
     ~10 us
     Features
     
     encoding
     
     past
     
     interactions
     Current

     application
     
     states
     +
     

     View Slide

147. Overview of the Predictor
     46
     Click
     ScrollUp
     ScrollDown
     ZoomIn
     ZoomOut
     …
     +
     Prediction
     Model
     Stop until the cumulative confidence of the
     predicted event sequence is below a threshold
     ~10 us
     Features
     
     encoding
     
     past
     
     interactions
     Current

     application
     
     states
     +
     

     View Slide

148. PES: Proactive Event Scheduler
     47
     Prediction
     Web Program
     
     Analysis
     Machine
     
     Learning
     Scheduling
     Constrained Optimization
     

     View Slide

149. Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     ▸ Goal: minimize total energy while meeting deadlines
     

     View Slide

150. Objective:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     ▸ Goal: minimize total energy while meeting deadlines
     N
     ∑
     i
     Min. Energy (i)
     

     View Slide

151. Objective:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     N
     ∑
     i
     △Texe (i) x
     Min.
     

     View Slide

152. Objective:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     

     View Slide

153. Objective:
     Constraints:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     

     View Slide

154. Objective:
     Constraints:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     Order: ≤
     Tend (i) Tstart (i+1)
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     

     View Slide

155. Objective:
     Constraints:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     Tstart 1 Tstart 2 Tstart 3
     TQoS 1 TQoS 2 TQoS 3
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     

     View Slide

156. Objective:
     Constraints:
     Constrained Optimization Formulation
     48
     E1 E2 E3
     Time
     Tstart 1 Tstart 2 Tstart 3
     TQoS 1 TQoS 2 TQoS 3
     △Texe 1 △Texe 2 △Texe 3
     ▸ Goal: minimize total energy while meeting deadlines
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     Scheduling knobs:
     
     Big/little + DVFS for
     each event.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     

     View Slide

157. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     △Texe (i) =
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

158. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     △Texe (i) = Tmemory
     +
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

159. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     Tcpu
     △Texe (i) = Tmemory
     +
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

160. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     △Texe (i) = Tmemory
     + Ncycles / f (i)
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

161. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Power (i)
     Min.
     △Texe (i) = Tmemory
     + Constants
     Ncycles / f (i)
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

162. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x
     Min.
     △Texe (i) = Tmemory
     + Constants
     Ncycles / f (i)
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     Pmap (i)
     Pmap (i) = Freq2Power (f(i))
     

     View Slide

163. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     49
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x
     Min.
     △Texe (i) = Tmemory
     + Constants
     Offline profile
     Ncycles / f (i)
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     Pmap (i)
     Pmap (i) = Freq2Power (f(i))
     

     View Slide

164. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     

     View Slide

165. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j) ⍺ (i, j) in {0,1}
     

     View Slide

166. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j) ⍺ (i, j) in {0,1}
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     

     View Slide

167. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j) ⍺ (i, j) in {0,1}
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

168. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j) ⍺ (i, j) in {0,1}
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     Only one DVFS
     setting is chosen
     
     for each event
     

     View Slide

169. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

170. Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

171. Overhead:
     Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

172. Overhead:
     10 DVFS configurations
     Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

173. Overhead:
     10 DVFS configurations
     8 events look ahead
     Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

174. Overhead:
     80 variables in ILP
     Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     

     View Slide

175. Overhead:
     Runtime overhead: 10ms
     80 variables in ILP
     Each Event:
     Objective:
     Constraints:
     Problem Formulation
     50
     ▸ Scheduling Problem → Constrained Optimization.
     N
     ∑
     i
     △Texe (i) x Pmap (i)
     Min.
     Order: ≤
     Tend (i) Tstart (i+1)
     Deadline: ≤
     Tstart (i) △Texe (i) TQoS (i)
     +
     { Ncycles / f (i, j) }
     △Texe (i) =Tmemory
     +M
     ∑
     j=0
     * ⍺ (i, j)
     Integer Linear
     Programing!
     Pmap (i) = Freq2Power ( f(i, j) ) * ⍺ (i, j)
     M
     ∑
     j=0
     1
     =
     ⍺ (i, j)
     With the constraint:
     M
     ∑
     j=0
     Dynamic programing
     

     View Slide

176. Putting Things Together
     51
     Web Application
     Rendering Engine
     Time
     

     View Slide

177. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Time
     

     View Slide

178. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Events
     Time
     Prediction
     

     View Slide

179. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Events
     Scheduler
     Predictions
     Time
     Prediction
     

     View Slide

180. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Events
     Speculative
     
     Schedules
     Scheduler
     Predictions
     Time
     Prediction Schedule
     

     View Slide

181. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Events
     Speculative
     
     Schedules
     Scheduler
     Predictions
     Time
     Prediction Schedule F1
     Pending
     Frame Buffer
     F2 F3
     

     View Slide

182. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Controller
     Events
     Speculative
     
     Schedules
     Pending
     
     Frames
     Scheduler
     Predictions
     Time
     Prediction Schedule
     F1
     Pending
     Frame Buffer F2 F3
     

     View Slide

183. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Controller
     Events
     Speculative
     
     Schedules
     Pending
     
     Frames
     Scheduler
     Predictions
     Time
     Prediction Schedule
     F1
     Pending
     Frame Buffer F2 F3
     

     View Slide

184. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Controller
     Events
     Speculative
     
     Schedules
     Pending
     
     Frames
     Scheduler
     Predictions
     Commit
     Time
     Prediction Schedule
     F1
     Pending
     Frame Buffer F2 F3
     ✔ ✔
     

     View Slide

185. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Controller
     Events
     Speculative
     
     Schedules
     Pending
     
     Frames
     Recover
     Scheduler
     Predictions
     Commit
     Time
     Prediction Schedule
     F1
     Pending
     Frame Buffer F2 F3
     ✔ ✔ ✘
     

     View Slide

186. PES
     Putting Things Together
     51
     Web Application
     Rendering Engine
     Predictor
     Controller
     Events
     Speculative
     
     Schedules
     Pending
     
     Frames
     Recover
     Scheduler
     Predictions
     Commit
     Time
     Prediction Schedule
     F1
     Pending
     Frame Buffer F2 F3
     ✔ ✔ ✘ F3
     ✔
     

     View Slide

187. Evaluation
     52
     52
     ▸Traces
     ▹Training: 10 users, more than 100 traces.
     ▹Testing: 36 traces, 3 for each Web application.
     ▸Baseline Mechanisms
     ▹Interactive governor (Interactive) — Android default
     ▹EBS: a state-of-the-art reactive event-based scheduler
     ▹Oracle: optimal scheduler
     ▸Metrics
     ▹Energy consumption
     ▹QoS violation
     

     View Slide

188. High Prediction Accuracy, Low Mis-Prediction Penalty
     53
     Prediction
     Accuracy (%)
     60
     70
     80
     90
     100
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     

     View Slide

189. High Prediction Accuracy, Low Mis-Prediction Penalty
     53
     Prediction
     Accuracy (%)
     60
     70
     80
     90
     100
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Mis-prediction
     Waste (ms)
     0
     10
     20
     30
     40
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     

     View Slide

190. High Prediction Accuracy, Low Mis-Prediction Penalty
     53
     Prediction
     Accuracy (%)
     60
     70
     80
     90
     100
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Mis-prediction
     Waste (ms)
     0
     10
     20
     30
     40
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     

     View Slide

191. High Prediction Accuracy, Low Mis-Prediction Penalty
     53
     Prediction
     Accuracy (%)
     60
     70
     80
     90
     100
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Mis-prediction
     Waste (ms)
     0
     10
     20
     30
     40
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     92% Prediction Accuracy and 20 ms of Mis-Prediction Waste
     

     View Slide

192. Experimental Result
     54
     

     View Slide

193. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

194. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

195. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

196. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

197. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

198. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     QoS Violation
     0
     0.15
     0.3
     0.45
     0.6
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

199. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     QoS Violation
     0
     0.15
     0.3
     0.45
     0.6
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

200. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     QoS Violation
     0
     0.15
     0.3
     0.45
     0.6
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

201. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     QoS Violation
     0
     0.15
     0.3
     0.45
     0.6
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     

     View Slide

202. Experimental Result
     54
     Norn. Energy
     0
     0.25
     0.5
     0.75
     1
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     QoS Violation
     0
     0.15
     0.3
     0.45
     0.6
     163
     msn
     slashdot
     youtube
     google
     amazon
     ebay
     sina
     espn
     bbc
     cnn
     twitter
     Interactive EBS PES Oracle
     61% less QoS Violation and 26% of Energy Reduction
     

     View Slide

203. Event-driven Processing is a
     Fundamental and Prevalent Paradigm
     Edge Computing Database Crowdsourced Sensor Network
     Mobile Applications Virtual Reality Internet-of-things Cloud Computing
     

     View Slide

204. What’s Next?
     

     View Slide

205. “Microkernel”-based Browser
     ▸ A web app/website exercises only a small portion
     of the Web specification. There are “hot” tags,
     CSS properties, etc.
     ▸ Design a minimalistic browser core, and load the
     rest only when requested by a specific webpage.
     57
     https://www.advancedwebranking.com/html/ https://maqentaer.com/devopera-static-backup/http/dev.opera.com/articles/view/mama-css-syntax/index.html
     

     View Slide

206. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     58
     

     View Slide

207. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     58
     

     View Slide

208. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     58
     

     View Slide

209. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     ▸ Still, focus on performance analysis and inspection without giving
     developers directly control over user-perceived performance.
     58
     

     View Slide

210. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     ▸ Still, focus on performance analysis and inspection without giving
     developers directly control over user-perceived performance.
     ▹ Developers are given a set of high-level, coarse-grained guidelines
     58
     

     View Slide

211. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     ▸ Still, focus on performance analysis and inspection without giving
     developers directly control over user-perceived performance.
     ▹ Developers are given a set of high-level, coarse-grained guidelines
     ▸ Proposal: empower developers to directly express their requirements of
     user-centric performance goals (up to the browser to deliver).
     58
     

     View Slide

212. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     ▸ Still, focus on performance analysis and inspection without giving
     developers directly control over user-perceived performance.
     ▹ Developers are given a set of high-level, coarse-grained guidelines
     ▸ Proposal: empower developers to directly express their requirements of
     user-centric performance goals (up to the browser to deliver).
     ▹ Paint order: developers identify “hero” elements and specify the order
     in which hero elements are painted. div#hero {paint-order:1}
     58
     

     View Slide

213. User-Centric Language (Extensions)
     ▸ Web performance is a comprehensive user-centric objective that can be
     not captured by one single metric.
     ▹ Old days: time when unload is fired, doesn’t corresponds to UX
     ▹ Now: First Meaningful Paint (FMP), Time to Interactive (TTI), etc.
     ▸ Still, focus on performance analysis and inspection without giving
     developers directly control over user-perceived performance.
     ▹ Developers are given a set of high-level, coarse-grained guidelines
     ▸ Proposal: empower developers to directly express their requirements of
     user-centric performance goals (up to the browser to deliver).
     ▹ Paint order: developers identify “hero” elements and specify the order
     in which hero elements are painted. div#hero {paint-order:1}
     ▹ Interactive state: developers specify the kind of interaction that
     should ideally be granted when a particular element is painted.
     div#menu{istate:touchstart,50}
     58
     

     View Slide