C ++ Console Conway的生命游戏非常缓慢的更新

时间:2017-12-21 21:14:03

标签: c++ performance conways-game-of-life

我目前正在研究Conway的C ++生活游戏的控制台版本。

Repo Link

问题在于我的Draw()方法正在尽可能快地绘制下一代。

我记得在某处读到使用GotoXY与获取控制台缓冲区相比非常慢。

问题是我没有一个概念如何实现控制台缓冲区来处理我的代码。

我并没有要求你们这样做,但我只是想让你看看我的Draw()和Update()方法,看看我是否正在做一些可怕的内存占用。 / p>

我的代码可能不是最好的,因为我仍然是#34;相当新的"到C ++,所以在批评之前要记住这一点。 :&#39)

细胞结构

struct Cell
{
int x, y; // Cell X, Y coordiantes
bool IsAlive; // Cell life state

string ToString()
{
    return "X: " + to_string(x) + "\tY" + to_string(y) + "\tAlive State: " + to_string(IsAlive);
}

void Die()
{
    IsAlive = false;
}

void Resurrect()
{
    IsAlive = true;
}
};

更新

void Update()
{
// Update Cells
CalculateNextGeneration(CellMap);
}

绘制

void Draw()
{
for (auto cell : CellMap) // Iterate through all cells in CellMap vector
{
    // Draw Cell if Alive
    // If a cell was alive upto 10th generation, display cell as '1'.
    if (cell.IsAlive)
        GotoXY(cell.x, cell.y, AliveCell);
    // If a cell is dead, display cell as ' '.
    else
        GotoXY(cell.x, cell.y, DeadCell);
}
}

CalculateNextGeneration

 /* 
TODO: Encapsulate IF Statements

Game Rules
1: Any Live Cell which has < 2 Live Neighbours, die [ Underpopulation ]
2: Any Live Cell which has 2 OR 3 Neighbours, live
3: Any Live Cell which has > 3 Neighbours, die [ Overpopulation ]
4: Any Dead Cell which has EXACTLY 3 Neighbours, resurrect [ Reporduciton ]

*/
void CalculateNextGeneration(vector<Cell> &map)
{
    for (auto& cell : map) // Iterate through all cells as references
    {
        if ((cell.IsAlive && GetAdjacentCellCount(cell, map) < 2) || (cell.IsAlive && GetAdjacentCellCount(cell, map) > 3)) // If current cell has < 2 neighbours OR current cell has > 3 neighbours, die [ Underpopulation & Overpopulation ] 
        {
            // Die
            cell.Die();
        }
        if (cell.IsAlive && GetAdjacentCellCount(cell, map) == 2 || GetAdjacentCellCount(cell, map) == 3) // If current cell has 2 OR 3 adjacent neighbours, live until next generation.
        {
            // Live Until Next Generation
        }
        if (!cell.IsAlive && GetAdjacentCellCount(cell, map) == 3) // If current cell has EXACTLY 3 adjacent neighbours, resurrect [ Reproduciton ]
        {
            // Resurrect
            cell.Resurrect();
        }
    }
}

GetAdjacentCellCount

/* Count Adjacent cells

Long version of counting adjacent cells.
TODO: Clean up code.

Example:

0 = Dead Cell
1 = Alive Cell
X = Current Cell

    1   0   1
    0   X   0
    1   1   0

Return would be 4 in this case. Since There are 4 alive cells surround the current cell (X).

The function doesn't consider the current cell's life state.

*/
int GetAdjacentCellCount(Cell &currentCell, vector<Cell> &map)
{
    int aliveCount = 0;

    int currentX = currentCell.x;
    int currentY = currentCell.y;

    vector<Cell> adjacentCells; // Create temporary vector with all adjacent cells

    adjacentCells.push_back(GetCellAtXY(currentX - 1, currentY - 1, map));  // - - // TOP LEFT CELL
    adjacentCells.push_back(GetCellAtXY(currentX, currentY - 1, map));      // 0 - // TOP MIDDLE CELL
    adjacentCells.push_back(GetCellAtXY(currentX + 1, currentY - 1, map));  // + - // TOP RIGHT CELL
    adjacentCells.push_back(GetCellAtXY(currentX + 1, currentY, map));      // + 0 // MIDDLE LEFT CELL
    adjacentCells.push_back(GetCellAtXY(currentX + 1, currentY + 1, map));  // + + // MIDDLE RIGHT CELL
    adjacentCells.push_back(GetCellAtXY(currentX, currentY + 1, map));      // 0 + // BOTTOM LEFT CELL
    adjacentCells.push_back(GetCellAtXY(currentX - 1, currentY + 1, map));  // - + // BOTTOM MIDDLE CELL
    adjacentCells.push_back(GetCellAtXY(currentX - 1, currentY, map));      // - - // BOTTOM RIGHT CELL

    for (auto adjCell : adjacentCells) // Iterate through all adjacent cells
    {
        if (adjCell.IsAlive == true) // Count how many are alive
            aliveCount++;
    }

    return aliveCount;
}

GetCellAtXY

Cell GetCellAtXY(int x, int y, vector<Cell> &map)
{
    Cell retrievedCell = { 0, 0, false }; // Create a default return cell

    for (auto cell : map) // Iterate through all cells in the map
    {
        if (cell.x == x && cell.y == y) // If Cell is found at coordinate X, Y
            retrievedCell = cell; // Set found Cell to return Cell
    }

    return retrievedCell;
}

1 个答案:

答案 0 :(得分:1)

您可能会花费大量时间重写GetAdjacentCellCount功能:

if (GetCellAtXY(currentX - 1, currentY - 1, map)  // - - // TOP LEFT CELL
   aliveCount++;
if (GetCellAtXY(currentX, currentY - 1, map));    // 0 - // TOP MIDDLE CELL
   aliveCount++;
if (GetCellAtXY(currentX + 1, currentY - 1, map)  // + - // TOP RIGHT CELL
   aliveCount++;
if (GetCellAtXY(currentX + 1, currentY, map)      // + 0 // MIDDLE LEFT CELL
   aliveCount++;
if (GetCellAtXY(currentX + 1, currentY + 1, map)  // + + // MIDDLE RIGHT CELL
   aliveCount++;
if (GetCellAtXY(currentX, currentY + 1, map)      // 0 + // BOTTOM LEFT CELL
   aliveCount++;
if (GetCellAtXY(currentX - 1, currentY + 1, map)  // - + // BOTTOM MIDDLE CELL
   aliveCount++;
if (GetCellAtXY(currentX - 1, currentY, map)      // - - // BOTTOM RIGHT CELL
   aliveCount++;

使用push_back填充向量会导致在每次迭代时对每个单元格进行多次堆分配。

我不知道这是否是您的代码的一部分是瓶颈与否,您必须要知道这一点,但这似乎是一个简单的改进。

修改

我发布的太快了。您的问题出在GetCellAtXY函数中。你循环(平均)一半的细胞来找到你的邻居。在每次迭代期间,每个单元格执行此操作8次!

相反,创建一个直接指向其8个邻居的单元格对象,例如:

struct Cell
{
  int x, y; // Cell X, Y coordiantes
  bool IsAlive; // Cell life state
  std::array<Cell*,8> neighbors;
}

然后你通过循环找到邻居一次(或者你在创建它们时填充它们,这可能更好)。请注意,当您设置A.neighbor[left]=&B时,您还需要设置B.neighbor[right]=&A

我相信你会得到没有指针的建议,使用指针不是正确的C ++。但我喜欢指针。

有很多选择:一个2D网格的单元格,你通过计算知道每个邻居的索引,一个std :: map,你用坐标的散列索引一个单元格等。

修改

这是一种可以为邻居编制索引的方法。这不一定是不那么冗长的方法,但它可以解决这些问题:

struct FieldSize {
  int x, y;
}
FieldSize fieldSize{ 40, 20 };

struct Cell {
  int x, y; // Cell X, Y coordiantes
  bool IsAlive; // Cell life state
  // ...
  bool HasLeftNeighbor() {
    return x != 0;
  }
  bool HasRightNeighbor() {
    return x != fieldSize.x-1;
  }
  bool HasTopNeighbor() {
    return y != 0;
  }
  bool HasTopLeftNeighbor() {
    return HasLeftNeighbor() && HasTopNeighbor();
  }
  // ... etc.
  int GetLeftNeighbor() {
    return (x-1) + y * fieldSize.x;
  }
  int GetTopNeighbor() {
    return x + (y-1) * fieldSize.x;
  }
  int GetTopLeftNeighbor() {
    return (x-1) + (y-1) * fieldSize.x;
  }
  // ... etc.
}

然后在GetAdjacentCellCount中:

if (HasLeftNeighbor() && map[GetLeftNeighbor()].IsAlive)
  aliveCount++;
// etc.

同样,这是非常冗长的,可以很容易地变得更加紧凑,也可能更高效,但我想强调你的逻辑。