Question

I’m reading some book, and I’m given some timer implementation. However there is no real explanation about the timer in the book, and I’m wondering why he is doing certain things.

Code:

//  Desc:   Use this class to regulate code flow (for an update function say)
//          Instantiate the class with the frequency you would like your code
//          section to flow (like 10 times per second) and then only allow 
//          the program flow to continue if Ready() returns true

class Regulator
{
private:

  //the time period between updates 
  double m_dUpdatePeriod;

  //the next time the regulator allows code flow
  DWORD m_dwNextUpdateTime;


public:


  Regulator(double NumUpdatesPerSecondRqd)
  {
    m_dwNextUpdateTime = (DWORD)(timeGetTime()+RandFloat()*1000);

    if (NumUpdatesPerSecondRqd > 0)
    {
      m_dUpdatePeriod = 1000.0 / NumUpdatesPerSecondRqd; 
    }

    else if (isEqual(0.0, NumUpdatesPerSecondRqd))
    {
      m_dUpdatePeriod = 0.0;
    }

    else if (NumUpdatesPerSecondRqd < 0)
    {
      m_dUpdatePeriod = -1;
    }
  }


  //returns true if the current time exceeds m_dwNextUpdateTime
  bool isReady()
  {
    //if a regulator is instantiated with a zero freq then it goes into
    //stealth mode (doesn't regulate)
    if (isEqual(0.0, m_dUpdatePeriod)) return true;

    //if the regulator is instantiated with a negative freq then it will
    //never allow the code to flow
    if (m_dUpdatePeriod < 0) return false;

    DWORD CurrentTime = timeGetTime();

    //the number of milliseconds the update period can vary per required
    //update-step. This is here to make sure any multiple clients of this class
    //have their updates spread evenly
    static const double UpdatePeriodVariator = 10.0;

    if (CurrentTime >= m_dwNextUpdateTime)
    {
      m_dwNextUpdateTime = (DWORD)(CurrentTime + m_dUpdatePeriod + RandInRange(-UpdatePeriodVariator, UpdatePeriodVariator));

      return true;
    }

    return false;
  }
};

isEqual() checks whether the difference between two given doubles is smaller than 0.00…1

RandInRange() returns a random number within the given range

I don’t really understand the part why he is adding a random number with a given range to the m_dwNextUpdateTime. Even though he ‘explains’, it doesn’t seem to make any sense to me.

Can anyone help?

Answer 1

// This is here to make sure any multiple clients of this class
//have their updates spread evenly

The randomness is introduced to lower contention when you run the same code many times concurrently. If all clients use the exact same interval, they might all “wake up” at the exact same time, which can be unfortunate load-wise. Introducing some random variations makes them spread out their activity more.