#include <iostream>
#include "Common.h"
#include "Tests.h"


Square::Square() {
	for (size_t index = 0; index < data_size; ++index)
	{
		gpuHostBuffer.push_back(static_cast<float>(index));
		sourceData.push_back(static_cast<float>(index));
	}
}
void Square::gpu_compute(cl::Context* context, cl::CommandQueue* queue, cl::Program* program, cl::Event* Event) {
	cl_int err = CL_SUCCESS;

	// Get the kernel handle
	cl::Kernel kernel(*program, "square", &err);
	CheckCLError(err);

	clInputBuffer = cl::Buffer(*context, CL_MEM_READ_ONLY, sizeof(float) * data_size, NULL, &err);
	queue->enqueueWriteBuffer(clInputBuffer,
		true, // Blocking!
		0, sizeof(float) * data_size, gpuHostBuffer.data());

	// Allocate the output data
	clResultBuffer = cl::Buffer(*context, CL_MEM_WRITE_ONLY, sizeof(float) * data_size, NULL, &err);

	// Set the kernel parameters	
	kernel.setArg(0, clInputBuffer); // kernel FV paraméterei sorrendben
	kernel.setArg(1, clResultBuffer);

	// Enqueue the kernel
	queue->enqueueNDRangeKernel(kernel,
		cl::NullRange,				// Indexek nem eloffszetelve
		cl::NDRange(data_size, 1),	// Minden elemet egy szál
		cl::NullRange,				// Workgroup méret? - ez az auto, ha nem indul, 1024-re, onnan csökkent, amig elindul
		NULL,						// 
		Event); // Ő jlezi hogy vége, lsd lent
}

void Square::cpu_compute()
{
	for (size_t index = 0; index < data_size; index++) {
		cpuResult.push_back(sourceData[index] * sourceData[index]);
	}
}

void Square::collect_results(cl::CommandQueue* queue) {
	queue->enqueueReadBuffer(clResultBuffer, true, 0, sizeof(float) * data_size, gpuHostBuffer.data());
}	

bool Square::validate_results() {
	for (size_t index = 0; index < data_size; index++) {
		if (cpuResult[index] != gpuHostBuffer[index]) {
			std::cout << "Wrong result at [" << index << "]: " << gpuHostBuffer[index] << "!=" << cpuResult[index] << std::endl;
			return false;
		}
	}
	std::cout << "Test \"Square\" completed." << std::endl;
	return true;
}