Блог

Это сводит меня с ума. Похоже, что все и их мать разместили сообщение на форуме где-то в Интернете именно об этой проблеме и получили точно такое же решение, которое у меня не работает…

Это очень простой пример типа «привет, мир», который я скопировал из Интернета. Я назвал его cltest.c.

 #include <stdio.h>
#include <stdlib.h>

#ifdef __APPLE__
    #include <OpenCL/opencl.h>
#else
    #include <CL/cl.h>
#endif

#define MAX_SOURCE_SIZE (0x100000)

int main()
{
    const int LIST_SIZE = 1024;
    int* a = (int*) malloc(sizeof(int) * LIST_SIZE);
    int* b = (int*) malloc(sizeof(int) * LIST_SIZE);

    int i;  
    for (i = 0; i < LIST_SIZE; i  )
    {
        a[i] = i;
        b[i] = LIST_SIZE - i;   
    }
    
    //load the kernel source code into the array source_str
    FILE* fp;
    char* source_str;
    size_t source_size;
    
    fp = fopen("gpu.cl", "r");
    if (!fp)
    {
        printf("Failed to load kernel");
        exit(1);
    }
    source_str = (char *)malloc(MAX_SOURCE_SIZE);
    source_size = fread(source_str, 1, MAX_SOURCE_SIZE, fp);
    fclose(fp);
    
    //Get platform and device information
    cl_platform_id platform_id = NULL;
    cl_device_id device_id = NULL;
    cl_uint ret_num_devices;
    cl_uint ret_num_platforms;
    cl_int ret = clGetPlatformIDs(1, amp;platform_id, amp;ret_num_platforms);
    ret = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_GPU, 1, amp;device_id, amp;ret_num_devices);
    
    //create an OpenCL context
    cl_context context = clCreateContext(NULL, 1, amp;device_id, NULL, NULL, amp;ret);
    
    //create a command queue
    cl_command_queue command_queue = clCreateCommandQueue(context, device_id, 0, amp;ret);
    
    //create memory buffers on the device for each vector
    cl_mem a_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY, LIST_SIZE * sizeof(int), NULL, amp;ret);
    cl_mem b_mem_obj = clCreateBuffer(context, CL_MEM_READ_ONLY, LIST_SIZE * sizeof(int), NULL, amp;ret);
    cl_mem c_mem_obj = clCreateBuffer(context, CL_MEM_WRITE_ONLY, LIST_SIZE * sizeof(int), NULL, amp;ret);
    
    //copy a and b to their respective memory buffers
    ret = clEnqueueWriteBuffer(command_queue, a_mem_obj, CL_TRUE, 0, LIST_SIZE * sizeof(int), a, 0, NULL, NULL);
    ret = clEnqueueWriteBuffer(command_queue, b_mem_obj, CL_TRUE, 0, LIST_SIZE * sizeof(int), b, 0, NULL, NULL);
    
    //create a program from the kernel source
    cl_program program = clCreateProgramWithSource(context, 1, (const char **)amp;source_str, (const size_t *)amp;source_size, amp;ret);
    
    //build the program
    ret = clBuildProgram(program, 1, amp;device_id, NULL, NULL, NULL);
    
    //create the OpenCL kernel
    cl_kernel kernel = clCreateKernel(program, "vector_add", amp;ret);
    
    //set the arguments of the kernel
    ret = clSetKernelArg(kernel, 0, sizeof(cl_mem), (void *)amp;a_mem_obj);
    ret = clSetKernelArg(kernel, 1, sizeof(cl_mem), (void *)amp;b_mem_obj);
    ret = clSetKernelArg(kernel, 2, sizeof(cl_mem), (void *)amp;c_mem_obj);
    
    //execute the OpenCL kernel on the list
    size_t global_item_size = LIST_SIZE;
    size_t local_item_size = 64;
    ret = clEnqueueNDRangeKernel(command_queue, kernel, 1, NULL, amp;global_item_size, amp;local_item_size, 0, NULL, NULL);
    
    //read memory buffer c on the device to the local variable c
    int* c = (int *)malloc(sizeof(int) * LIST_SIZE);
    ret = clEnqueueReadBuffer(command_queue, c_mem_obj, CL_TRUE, 0, LIST_SIZE * sizeof(int), c, 0, NULL, NULL);
    
    //display the result to the screen
    for (i = 0; i < LIST_SIZE; i  )
    {
        printf("%d   %d = %dn", a[i], b[i], c[i]);
    }
    
    //clean up
    ret = clFlush(command_queue);
    ret = clFinish(command_queue);
    ret = clReleaseKernel(kernel);
    ret = clReleaseProgram(program);
    ret = clReleaseMemObject(a_mem_obj);
    ret = clReleaseMemObject(b_mem_obj);
    ret = clReleaseMemObject(c_mem_obj);
    ret = clReleaseContext(context);
    free(a);
    free(b);
    free(c);
    return 0;
}
  

Для сборки я использую команду

 gcc -v -Wl,--verbose -H cltest.c -I"C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0include" -L"C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0libx64" -lOpenCL
  

Это генерирует 1021 строку сообщений, но вот те, которые (я думаю) актуальны:

 (...)
#include "..." search starts here:
#include <...> search starts here:
 C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0include
 c:mingwbin../lib/gcc/mingw32/9.2.0/include
 c:mingwbin../lib/gcc/mingw32/9.2.0/../../../../include
 c:mingwbin../lib/gcc/mingw32/9.2.0/include-fixed
End of search list.
GNU C17 (MinGW.org GCC Build-20200227-1) version 9.2.0 (mingw32)
        compiled by GNU C version 9.2.0, GMP version 6.1.2, MPFR version 4.0.2, MPC version 1.1.0, isl version isl-0.21-GMP
(...)
. C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0include/CL/cl.h
.. C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0include/CL/cl_platform.h
... c:mingwlibgccmingw329.2.0includestdint.h
.... c:mingwincludestdint.h
..... c:mingwlibgccmingw329.2.0includestddef.h
... c:mingwlibgccmingw329.2.0includestddef.h
(...)
C:UsersDevsmanAppDataLocalTempccwXijlf.o
attempt to open C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0libx64/libOpenCL.dll.a failed
attempt to open C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0libx64/OpenCL.dll.a failed
attempt to open C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0libx64/libOpenCL.a failed
attempt to open C:Program FilesNVIDIA GPU Computing ToolkitCUDAv11.0libx64/OpenCL.lib succeeded
(...)
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x11b): undefined reference to `clGetPlatformIDs@12'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x152): undefined reference to `clGetDeviceIDs@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x18a): undefined reference to `clCreateContext@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x1b9): undefined reference to `clCreateCommandQueue@20'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x1f7): undefined reference to `clCreateBuffer@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x235): undefined reference to `clCreateBuffer@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x273): undefined reference to `clCreateBuffer@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x2c8): undefined reference to `clEnqueueWriteBuffer@36'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x31d): undefined reference to `clEnqueueWriteBuffer@36'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x34b): undefined reference to `clCreateProgramWithSource@20'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x383): undefined reference to `clBuildProgram@24'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x3a3): undefined reference to `clCreateKernel@12'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x3cb): undefined reference to `clSetKernelArg@16'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x3f3): undefined reference to `clSetKernelArg@16'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x41b): undefined reference to `clSetKernelArg@16'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x476): undefined reference to `clEnqueueNDRangeKernel@36'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x4dc): undefined reference to `clEnqueueReadBuffer@36'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x54d): undefined reference to `clFlush@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x55e): undefined reference to `clFinish@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x56f): undefined reference to `clReleaseKernel@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x580): undefined reference to `clReleaseProgram@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x591): undefined reference to `clReleaseMemObject@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x5a2): undefined reference to `clReleaseMemObject@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x5b3): undefined reference to `clReleaseMemObject@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: C:UsersDevsmanAppDataLocalTempccwXijlf.o:cltest.c:(.text 0x5c4): undefined reference to `clReleaseContext@4'
c:/mingw/bin/../lib/gcc/mingw32/9.2.0/../../../../mingw32/bin/ld.exe: link errors found, deleting executable `a.exe'
(...)
  

Я несколько раз менял порядок каждого аргумента, я дважды, трижды и вчетверо проверял указанные каталоги на наличие соответствующих файлов, я снял флаг -o на случай, если это как-то мешает, я попытался скомпилировать cltest.c в объект, а затем связать их в отдельном вызове, и у меня закончились идеи! Надеюсь, есть простое исправление, которое настолько простое / очевидное, что я его упускаю. Заранее спасибо.