matrix_select_batch.cu File Reference

Implementation of the Matrix::select_batch method for selecting a subset of the matrix. More...

#include "matrix.h"
#include <cuda_runtime.h>
#include <stdexcept>

Include dependency graph for matrix_select_batch.cu:

Go to the source code of this file.

Functions
__global__ void	selectBatchKernel (const double *src, double *dst, int src_cols, int dst_cols, int start_row, int start_col, int num_rows, int num_cols)
	CUDA kernel for selecting a subset of the matrix. More...

Detailed Description

Implementation of the Matrix::select_batch method for selecting a subset of the matrix.

Definition in file matrix_select_batch.cu.

Function Documentation

selectBatchKernel()

__global__ void selectBatchKernel	(	const double *	src,
		double *	dst,
		int	src_cols,
		int	dst_cols,
		int	start_row,
		int	start_col,
		int	num_rows,
		int	num_cols
	)

CUDA kernel for selecting a subset of the matrix.

Parameters

src	Pointer to the source matrix data.
dst	Pointer to the destination matrix data.
src_cols	Number of columns in the source matrix.
dst_cols	Number of columns in the destination matrix.
start_row	Starting row index.
start_col	Starting column index.
num_rows	Number of rows to select.
num_cols	Number of columns to select.

Definition at line 21 of file matrix_select_batch.cu.

                                                                                            {
    // Calculate global thread indices
    int row = blockIdx.y * blockDim.y + threadIdx.y;
    int col = blockIdx.x * blockDim.x + threadIdx.x;
 
    // Check if thread is within the selected subset bounds
    if (row < num_rows && col < num_cols) {
        // Calculate source and destination indices
        int src_idx = (start_row + row) * src_cols + (start_col + col);
        int dst_idx = row * dst_cols + col;
 
        // Copy the element from source to destination
        dst[dst_idx] = src[src_idx];
    }
}