gpuip Documentation

0.1

gpuip is a C++ cross-platform framework for Image Processing on the GPU architechure. It tries to simplify the image processing pipeline on the GPU and make it more generic across the thre most common environments: OpenCL, CUDA and OpenGL GLSL. It provides a simple interface to copy data from and to the GPU and makes it easy to compile and run GPU kernel code.

Example:

#include <gpuip.h>

void print_timings(const char * func_name, double ms, std::string * err)
{
    if (ms != GPUIP_ERROR) {
        printf("%s took %.2lf ms.\n", func_name, ms);
    } else {
        printf("Error in %s: %s\n", func_name, err->c_str());
    }
}

void use_gpuip()
{
    std::string err;
    float * data;
    unsigned int width, height;
    ReadImage(&data, &width, &height); // definied somewhere else

    if (!gpuip::ImageProcessor::CanCreateGpuEnvironment(gpuip::GLSL)) {
        // ... deal with error - throw exception, return function etc
    }
    gpuip::ImageProcessor::Ptr ip = gpuip::ImageProcessor::Create(gpuip::GLSL);
    ip->SetDimensions(width, height);
    gpuip::Buffer::Ptr b0 = ip->CreateBuffer("b0", gpuip::FLOAT, 4);
    gpuip::Buffer::Ptr b1 = ip->CreateBuffer("b1", gpuip::FLOAT, 4);
    gpuip::Kernel::Ptr kernel = gpuipip->CreateKernel("modify_red");
    kernel->code = GetKernelCode(); // definied somewhere else
    kernel->inBuffers.push_back(gpuip::Kernel::BufferLink(b0, "img"));
    kernel->outBuffers.push_back(gpuip::Kernel::BufferLink(b1, "out_img"));
    kernel->paramsFloat.push_back(gpuip::Parameter<float>("alpha", 0.4));
    print_timings("Build", ip->Build(&err), &err);
    print_timings("Allocate", ip->Allocate(&err), &err);
    print_timings("Copy", ip->Copy(b0, gpuip::Buffer::COPY_TO_GPU, data, &err), &err);
    print_timings("Run", ip->Run(&err), &err);
    print_timings("Copy", ip->Copy(b1, gpuip::Buffer::COPY_FROM_GPU, data, &err), &err);
}