Linearized Access for Contiguous Multidimensional Storage

Snippet

Linearized Access for Contiguous Multidimensional Storage

High-performance systems avoid nested pointers (like int***) because they cause cache misses. Instead, data is stored in a single contiguous block, and multidimensional coordinates are mapped to a linear index using row-major or column-major arithmetic.

snippet.cpp

cpp

class Volume3D {
    int* buffer;
    int w, h, d;
public:
    Volume3D(int x, int y, int z) : w(x), h(y), d(z) {
        buffer = new int[x * y * z];
    }
    int& at(int x, int y, int z) {
        return buffer[x + w * (y + h * z)];
    }
    ~Volume3D() { delete[] buffer; }
};

Breakdown

buffer = new int[x * y * z];

Allocates a single flat array to hold all 3D data points contiguously in memory.

return buffer[x + w * (y + h * z)];

Calculates the exact memory offset for 3D coordinates (x, y, z) using the strides of the dimensions.

Previous snippet Next snippet

More Expert C++ exercises →

From your library

Linearized Access for Contiguous Multidimensional Storage

Related