OpenCL C++ Hello World program - undefined reference in "cl.hpp"

Question

I am trying to run a simple Hello World program that should print "Hello World" using the GPU (Nvidia GTX 960 in my case). I am trying to compile this program using Code::Blocks on Windows 10, 64 bit. Apparently there are no newer driver versions out so I am stuck with OpenCL 1.2.

What I have done so far:

Installed the NVIDIA CUDA SDK
Set up Code::Blocks compiler to search the include directory and added the OpenCL.lib to the linker settings.
Copy-pasted the Hello World program from here:

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

Created the "opencl_hello_world.cl" file, as descriped in the link.

I am a noob in OpenCL and just wanted to test if it works. Every "Hello-World" code I've tried so far fails with different errors.

#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;

}

The code fails with the following output:

||=== 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 as far as I can see: All errors point me to different lines in the "cl.hpp" file (undefined reference to X where X is an OpenCL function).

The problem is that the source of these errors point me to the "cl.hpp" file, which I am not responsible for fixing. If there was an error when finding the library, I could understand it, however there seems to be something wrong with the "cl.hpp" file itself.

I tried including "CL/cl.h" instead, which results in even more errors like "vector was not declared in this scope", "cl has not been declared", "platforms was not declared in this scope", etc.

My questions are:

What's wrong with my code?
Are there known bugs in the cl.hpp file? Is this one of them?
What is the difference between and ?
How can I upgrade to the latest version of OpenCL? Every driver at the moment only has OpenCL 1.2 for some reason.
Is there any information on version changes in OpenCL? It seems like every tutorial I try is either outdated or only work on version 1.0, only works in C, etc.? Are there any tutorials for C++, OpenCL 1.2 in 2016?

@ColonelThirtyTwo Not really. Look at the file name in the error messages. The errors seem to come from inside the cl.hpp file. "main.o" is created without errors, it's when the compiler compiled "cl.hpp", that's when it fails. And the problem is, this file is not from me but from Nvidia. I wanted to ask if this error is somehow known. I would have posted this on the OpenCL forums if they let me. — BiggCrazy, Jun 05 '16 at 22:45
@FelixSchütt Because that's where the function declarations are - functions that are missing a definition - which is in a library you need to link against. — Quinchilion, Jun 05 '16 at 23:51

OpenCL C++ Hello World program - undefined reference in "cl.hpp"

0 Answers0