GeoSpatial Computing

Accelerated Computing CUDA CUDA Programming and Performance
1

Using CUDA power to provide new geo is new topic that I started with viewshed and other algorithm writing.
Here we will be taking about different aspect of using CUDA in GEO

2

For first part I must say using GDAL library with CUDA is possible and its working in my case for creating viewshed.

#include “stdafx.h”
#include <stdio.h>
#include <cuda.h> → CUDA library
#include “gdal_priv.h” → GDAL library
#include <math.h>
#include <time.h>

3

int block_size = 10;
int n_blocks = N/block_size + (N%block_size == 0 ? 0:1);
los <<< n_blocks, block_size >>> (a_d,b_d, N,nXSize ,nYSize,XRef,YRef,ZRef); //line of sight
// Retrieve result from device and store it in host array
cudaMemcpy(a_h, a_d, sizeof(float)*N, cudaMemcpyDeviceToHost);
cudaMemcpy(b_h, b_d, sizeof(float)*N, cudaMemcpyDeviceToHost);

4

Sample program finished , I start to create GEOCloud using CUDA technology …