Question

我正在尝试运行一个简单的Hello World程序，该程序应打印＆＃34; Hello World＆＃34;使用GPU（在我的情况下为Nvidia GTX 960）。我正在尝试使用Windows 10,64位上的Code :: Blocks编译此程序。显然没有更新的驱动程序版本，所以我坚持使用OpenCL 1.2。

到目前为止我做了什么：

安装了NVIDIA CUDA SDK
设置Code :: Blocks编译器以搜索include目录，并将OpenCL.lib添加到链接器设置。
从这里复制粘贴Hello World程序：

http://dhruba.name/2012/10/06/opencl-cookbook-hello-world-using-cpp-host-binding/

创建了＆＃34; opencl_hello_world.cl＆＃34;文件，如链接中所述。

我是OpenCL的菜鸟，只想测试它是否有效。每一个＆＃34; Hello-World＆＃34;到目前为止，我尝试过的代码失败并出现了不同的错误。

#define __CL_ENABLE_EXCEPTIONS

#include <fstream>
#include <iostream>
#include <iterator>
#include <CL/cl.hpp>
#include <CL/opencl.h>

using namespace std;

int main () {

    vector<cl::Platform> platforms;
    vector<cl::Device> devices;
    vector<cl::Kernel> kernels;

    try {

        // create platform
        cl::Platform::get(&platforms);
        platforms[0].getDevices(CL_DEVICE_TYPE_GPU, &devices);

        // create context
        cl::Context context(devices);

        // create command queue
        cl::CommandQueue queue(context, devices[0]);

        // load opencl source
        ifstream cl_file("opencl_hello_world.cl");
        string cl_string(istreambuf_iterator<char>(cl_file), (istreambuf_iterator<char>()));
        cl::Program::Sources source(1, make_pair(cl_string.c_str(),
            cl_string.length() + 1));

        // create program
        cl::Program program(context, source);

        // compile opencl source
        program.build(devices);

        // load named kernel from opencl source
        cl::Kernel kernel(program, "hello_world");

        // create a message to send to kernel
        char* message = "Hello World!";
        int messageSize = 12;

        // allocate device buffer to hold message
        cl::Buffer buffer(CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR,
            sizeof(char) * messageSize, message);

        // set message as kernel argument
        kernel.setArg(0, buffer);
        kernel.setArg(1, sizeof(int), &messageSize);

        // execute kernel
        queue.enqueueTask(kernel);

        // wait for completion
        queue.finish();

        cout << endl;

    } catch (cl::Error e) {
        cout << endl << e.what() << " : " << e.err() << endl;
    }

    return 0;

}

代码失败，输出如下：

||=== Build: Debug in OpenCLTest (compiler: GNU GCC Compiler) ===|
obj\Debug\main.o||In function `getPlatformVersion':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1735|undefined reference to `clGetPlatformInfo@20'|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1737|undefined reference to `clGetPlatformInfo@20'|
obj\Debug\main.o||In function `getDevicePlatformVersion':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1744|undefined reference to `clGetDeviceInfo@20'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP13_cl_device_idE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1619|undefined reference to `clReleaseDevice@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP11_cl_contextE6retainES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1652|undefined reference to `clRetainContext@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP11_cl_contextE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1654|undefined reference to `clReleaseContext@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP17_cl_command_queueE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1663|undefined reference to `clReleaseCommandQueue@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP7_cl_memE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1672|undefined reference to `clReleaseMemObject@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP11_cl_programE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1690|undefined reference to `clReleaseProgram@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP10_cl_kernelE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1699|undefined reference to `clReleaseKernel@4'|
obj\Debug\main.o||In function `ZN2cl6detail16ReferenceHandlerIP9_cl_eventE7releaseES3_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|1708|undefined reference to `clReleaseEvent@4'|
obj\Debug\main.o||In function `ZNK2cl8Platform10getDevicesEyPSt6vectorINS_6DeviceESaIS2_EE':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2210|undefined reference to `clGetDeviceIDs@24'|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2216|undefined reference to `clGetDeviceIDs@24'|
obj\Debug\main.o||In function `ZN2cl8Platform3getEPSt6vectorIS0_SaIS0_EE':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2315|undefined reference to `clGetPlatformIDs@12'|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2322|undefined reference to `clGetPlatformIDs@12'|
obj\Debug\main.o||In function `ZN2cl7ContextC1ERKSt6vectorINS_6DeviceESaIS2_EEPiPFvPKcPKvjPvESC_S7_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2475|undefined reference to `clCreateContext@24'|
obj\Debug\main.o||In function `ZN2cl7ContextC1EyPiPFvPKcPKvjPvES6_S1_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|2586|undefined reference to `clCreateContextFromType@24'|
obj\Debug\main.o||In function `ZN2cl6BufferC1EyjPvPi':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|3158|undefined reference to `clCreateBuffer@24'|
obj\Debug\main.o||In function `ZN2cl6Kernel6setArgEjjPKv':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|4955|undefined reference to `clSetKernelArg@16'|
obj\Debug\main.o||In function `ZN2cl7ProgramC1ERKNS_7ContextERKSt6vectorISt4pairIPKcjESaIS8_EEPi':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|5054|undefined reference to `clCreateProgramWithSource@20'|
obj\Debug\main.o||In function `ZNK2cl7Program5buildERKSt6vectorINS_6DeviceESaIS2_EEPKcPFvP11_cl_programPvESB_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|5228|undefined reference to `clBuildProgram@24'|
obj\Debug\main.o||In function `ZN2cl6KernelC1ERKNS_7ProgramEPKcPi':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|5437|undefined reference to `clCreateKernel@12'|
obj\Debug\main.o||In function `ZN2cl12CommandQueueC1ERKNS_7ContextERKNS_6DeviceEyPi':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|5526|undefined reference to `clCreateCommandQueue@20'|
obj\Debug\main.o||In function `ZNK2cl12CommandQueue11enqueueTaskERKNS_6KernelEPKSt6vectorINS_5EventESaIS5_EEPS5_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|6334|undefined reference to `clEnqueueTask@20'|
obj\Debug\main.o||In function `ZNK2cl12CommandQueue6finishEv':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|6551|undefined reference to `clFinish@4'|
obj\Debug\main.o||In function `ZN2cl6Kernel6setArgINS_6BufferEEEijRKT_':|
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v7.5\include\CL\cl.hpp|4948|undefined reference to `clSetKernelArg@16'|
||error: ld returned 1 exit status|
||=== Build failed: 27 error(s), 0 warning(s) (0 minute(s), 0 second(s)) ===|

TL; DR，据我所见：所有错误都指向了＆＃34; cl.hpp＆＃34;中的不同行。 file（未定义引用X，其中X是OpenCL函数）。

问题是这些错误的根源指向了＆＃34; cl.hpp＆＃34;文件，我不负责修复。如果找到库时出错，我可以理解它，但是＆＃34; cl.hpp＆＃34;似乎有问题。归档。

我试过包括＆＃34; CL / cl.h＆＃34;相反，这会导致更多错误，例如＆＃34;向量未在此范围内声明＆＃34;，＆＃34; cl尚未声明＆＃34;，＆＃34;平台未在此范围内声明＆＃34; ;等等。

我的问题是：

我的代码出了什么问题？
cl.hpp文件中是否存在已知错误？这是其中之一吗？
和
如何升级到最新版本的OpenCL？目前，每个驱动程序由于某种原因只有OpenCL 1.2。
OpenCL中是否有关于版本更改的信息？似乎我尝试的每个教程要么已过时，要么仅适用于1.0版本，只适用于C等？ 2016年有没有关于C ++，OpenCL 1.2的教程？

OpenCL C ++ Hello World程序 - ＆＃34; cl.hpp＆＃34;

0 个答案: