5,321
edits
(→C) |
(→C++) |
||
Line 141: | Line 141: | ||
While you can use the C bindings in your C++ application, Khronos also provides a set of C++ bindings in <code>CL/cl2.hpp</code> which are much easier to use alongside std containers such as <code>std::vector</code>. When using C++ bindings, you also do not need to worry about releasing buffers since these are reference-counted. | While you can use the C bindings in your C++ application, Khronos also provides a set of C++ bindings in <code>CL/cl2.hpp</code> which are much easier to use alongside std containers such as <code>std::vector</code>. When using C++ bindings, you also do not need to worry about releasing buffers since these are reference-counted. | ||
<syntaxhighlight lang="cpp"> | |||
#include <CL/cl.hpp> | |||
#include <fstream> | |||
#include <iostream> | |||
int main(void) { | |||
int ret = 0; | |||
// Create the two input vectors | |||
const int LIST_SIZE = 1024; | |||
std::vector<int> A(LIST_SIZE); | |||
std::vector<int> B(LIST_SIZE); | |||
for (int i = 0; i < LIST_SIZE; i++) { | |||
A[i] = i; | |||
B[i] = LIST_SIZE - i; | |||
} | |||
// Load the kernel source code into the string source_str | |||
std::string source_str; | |||
{ | |||
std::ifstream file("vector_add_kernel.cl"); | |||
file.seekg(0, std::ios::end); | |||
source_str.resize(file.tellg()); | |||
file.seekg(0, std::ios::beg); | |||
file.read(&source_str[0], source_str.size()); | |||
} | |||
// Get platform and device information | |||
std::vector<cl::Platform> platforms; | |||
ret = cl::Platform::get(&platforms); | |||
std::vector<cl::Device> devices; | |||
ret = platforms[0].getDevices(CL_DEVICE_TYPE_ALL, &devices); | |||
// Create an OpenCL context | |||
cl::Context context(devices[0], NULL, NULL, NULL, &ret); | |||
// Create a command queue | |||
cl::CommandQueue command_queue(context, devices[0], 0UL, &ret); | |||
// Create memory buffers on the device for each vector | |||
cl::Buffer a_mem_obj(context, CL_MEM_READ_ONLY, LIST_SIZE * sizeof(int)); | |||
cl::Buffer b_mem_obj(context, CL_MEM_READ_ONLY, LIST_SIZE * sizeof(int)); | |||
cl::Buffer c_mem_obj(context, CL_MEM_READ_WRITE, LIST_SIZE * sizeof(int)); | |||
// Copy the lists A and B to their respective memory buffers | |||
ret = cl::copy(command_queue, A.begin(), A.end(), a_mem_obj); | |||
ret = cl::copy(command_queue, B.begin(), B.end(), b_mem_obj); | |||
// Create a program from the kernel source | |||
cl::Program program(context, source_str); | |||
// Build the program | |||
ret = program.build(std::vector<cl::Device>{devices[0]}); | |||
if (ret != CL_SUCCESS) { | |||
std::cerr << "Error building program" << std::endl; | |||
exit(EXIT_FAILURE); | |||
} | |||
// Create the OpenCL kernel | |||
cl::Kernel kernel(program, "vector_add", &ret); | |||
if (ret != CL_SUCCESS) { | |||
std::cerr << "Error creating kernel" << std::endl; | |||
exit(EXIT_FAILURE); | |||
} | |||
// Set the arguments of the kernel | |||
ret = kernel.setArg(0, sizeof(cl_mem), &a_mem_obj()); | |||
ret = kernel.setArg(1, sizeof(cl_mem), &b_mem_obj()); | |||
ret = kernel.setArg(2, sizeof(cl_mem), &c_mem_obj()); | |||
// Execute the OpenCL kernel on the list | |||
cl::NDRange global_item_size(LIST_SIZE); // Process the entire lists | |||
cl::NDRange local_item_size(64); // Divide work items into groups of 64 | |||
ret = command_queue.enqueueNDRangeKernel(kernel, 0, global_item_size, | |||
local_item_size, NULL, NULL); | |||
if (ret != CL_SUCCESS) { | |||
std::cerr << "Error starting kernel" << std::endl; | |||
exit(EXIT_FAILURE); | |||
} | |||
// Read the memory buffer C on the device to the local variable C | |||
std::vector<int> C(LIST_SIZE); | |||
ret = cl::copy(command_queue, c_mem_obj, C.begin(), C.end()); | |||
if (ret != CL_SUCCESS) { | |||
std::cerr << "Error copying C from gpu to memory " << ret << std::endl; | |||
exit(EXIT_FAILURE); | |||
} | |||
// Display the result to the screen | |||
for (int i = 0; i < LIST_SIZE; i++) | |||
printf("%d + %d = %d\n", A[i], B[i], C[i]); | |||
return 0; | |||
} | |||
</syntaxhighlight> | |||
===Julia=== | ===Julia=== |