User`s guide

9 GPU Computing

9-20

Run CUDA or PTX Code on GPU

In this section...

“Overview” on page 9-20

“Create a CUDAKernel Object” on page 9-21

“Run a CUDAKernel” on page 9-26

“Complete Kernel Workflow” on page 9-28

Overview

This topic explains how to create an executable kernel from CU or PTX (parallel thread

execution) files, and run that kernel on a GPU from MATLAB. The kernel is represented

in MATLAB by a CUDAKernel object, which can operate on MATLAB array or gpuArray

variables.

The following steps describe the CUDAKernel general workflow:

Use compiled PTX code to create a CUDAKernel object, which contains the GPU

executable code.

Set properties on the CUDAKernel object to control its execution on the GPU.

Call feval on the CUDAKernel with the required inputs, to run the kernel on the

GPU.

MATLAB code that follows these steps might look something like this:

% 1. Create CUDAKernel object.

k = parallel.gpu.CUDAKernel('myfun.ptx','myfun.cu','entryPt1');

% 2. Set object properties.

k.GridSize = [8 1];

k.ThreadBlockSize = [16 1];

% 3. Call feval with defined inputs.

g1 = gpuArray(in1); % Input gpuArray.

g2 = gpuArray(in2); % Input gpuArray.

result = feval(k,g1,g2);