Question

我正试图在D中实现生命游戏（这对我来说是新的）。它适用于尺寸为100x100,200x200,300x300的阵列，但当我达到400x400及更高版本时，可执行文件将停止工作。我该如何解决？这是我的代码：

import std.stdio;
import std.random;
import std.conv:to;
import std.file;
import std.conv:to;
import core.simd;

alias  to!(string) toString;

immutable side = 400;
immutable generations = 1000;

alias  to!(string) toString;

void main ()
{

//Initialize Game of Life

float[side+2][side+2] big_grid;
float[side][side] cell_grid;

//Input random numbers in the big grid

for ( int i = 1; i < side+1; i++ )
    for ( int k = 1; k < side+1; k++ )
    {
    big_grid[i][k] = uniform(0,2);
    }

//Wrap the cells to sides (toroidal) in the big grid

for (int k = 1; k < side+1; k++ )
    {
    big_grid[side+1][k] = big_grid[1][k];
    big_grid[0][k] = big_grid[side][k];
    big_grid[k][side + 1] = big_grid[k][1];
    big_grid[k][0] = big_grid[k][side];
    }

big_grid[side+1][side+1] = big_grid[1][1];
big_grid[0][0] = big_grid[side][side];
big_grid[0][side+1] = big_grid[side][1];
big_grid[side+1][0] = big_grid[1][side];

//Transfer cell information from big grid to cell grid

for ( int i = 1; i < side+1; i++ )
    for ( int k = 1; k < side+1; k++ )
    {
    cell_grid[i-1][k-1] = big_grid[i][k];
    }

File file = File("gen0.txt", "w");
for ( int i = 0; i < side; i++ )
    {
        for ( int k = 0; k < side; k++ )
        {
        file.write(cell_grid[i][k],",");
        }
        file.write("\n");
    }
file.close();

//Update the state of the cells

for (int d = 1; d < generations; d++ )
    {

    //Rules of Life

    for ( int i = 0; i < side; i++ )
        for ( int k = 0; k < side; k++ )
        {
        if (big_grid[i+1][k+1] == 0 && (big_grid[i][k]+big_grid[i][k+1]+big_grid[i][k+2]+big_grid[i+1][k]+big_grid[i+1][k+1]+big_grid[i+1][k+2]+big_grid[i+2][k]+big_grid[i+2][k+1]+big_grid[i+2][k+2] == 3))
            {
            cell_grid[i][k] = 1;
            }
        else if (big_grid[i+1][k+1] == 1 && ((big_grid[i][k]+big_grid[i][k+1]+big_grid[i][k+2]+big_grid[i+1][k]+big_grid[i+1][k+1]+big_grid[i+1][k+2]+big_grid[i+2][k]+big_grid[i+2][k+1]+big_grid[i+2][k+2] == 3) || (big_grid[i][k]+big_grid[i][k+1]+big_grid[i][k+2]+big_grid[i+1][k]+big_grid[i+1][k+1]+big_grid[i+1][k+2]+big_grid[i+2][k]+big_grid[i+2][k+1]+big_grid[i+2][k+2] == 4)))
            {
            cell_grid[i][k] = 1;
            }
        else
            {
            cell_grid[i][k] = 0;
            }
        }

    //Update big grid for next iteration

    for ( int i = 1; i < side+1; i++ )
        for ( int k = 1; k < side+1; k++ )
        {
        big_grid[i][k] = cell_grid[i-1][k-1];
        }

    for (int k = 1; k < side+1; k++ )
        {
        big_grid[side+1][k] = big_grid[1][k];
        big_grid[0][k] = big_grid[side][k];
        big_grid[k][side + 1] = big_grid[k][1];
        big_grid[k][0] = big_grid[k][side];
        }

    big_grid[side+1][side+1] = big_grid[1][1];
    big_grid[0][0] = big_grid[side][side];
    big_grid[0][side+1] = big_grid[side][1];
    big_grid[side+1][0] = big_grid[1][side];

    file = File("gen"~toString(d)~".txt", "w");
    for ( int i = 0; i < side; i++ )
    {
        for ( int k = 0; k < side; k++ )
        {
        file.write(cell_grid[i][k],",");
        }
        file.write("\n");
    }
    file.close();
    }

}

Answer 1

最可能的原因是使用Windows的默认堆栈限制，并且至少有两个路径可以解决它。

1。增加堆栈限制。

如果您在Windows上编译和运行程序，可执行文件通常会编译自己的最大堆栈大小，因此它更像是特定于语言的操作系统。

现在，对于编译的编程语言，它们的链接器通常提供增加最大堆栈大小的选项。例如，如果使用dmd（官方D编译器）并编译为32位目标，命令行将dmd -L/STACK:16777216 program.d将限制设置为16兆字节。 dmd链接器选项列表为here。

2。避免大量使用堆栈。

另一条路径是要注意，只需要本地普通旧数据变量的堆栈。数组float[side+2][side+2] big_grid;是静态数组，这意味着它被分配到声明为(side+2) * (side+2)连续4字节float s的位置

如果你改为使用动态数组auto big_grid = new float [] [] (side + 2, side + 2);，这将只在堆栈上分配size_t * 2个字节（8或16，具体取决于位数）其余的分配（一个浮动的动态数组的一维数组，以及size+2一维浮点数组）将在垃圾收集堆上发生。您可以详细了解不同类型的arrays in D，甚至可以阅读更多关于使用dynamic arrays的信息。

您还可以将数组声明为 thread-local ：只需将声明从函数范围移到模块范围即可。如果您想要一个真正的全局变量，请将__gshared添加到模块级声明：__gshared float[side+2][side+2] big_grid;。这样，您仍然可以在不在堆栈上分配时使用静态数组。

D编程语言中的生命游戏

1 个答案:

1。增加堆栈限制。

2。避免大量使用堆栈。