VideoEditorBeta/flatLight.cpp at master · erfan-ops/VideoEditorBeta · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
#include "flatLight.h"
#include "flatLight_launcher.cuh"
#include "globals.h"
#include "utils.h"
#include "SourcesOpenCL.h"

void (FlatLightProcessor::* FlatLightProcessor::processFunc)() const = nullptr;
void (FlatLightProcessor::* FlatLightProcessor::processFuncRGBA)() const = nullptr;

cl_kernel FlatLightProcessor::m_openclKernel = nullptr;
cl_kernel FlatLightProcessor::m_openclKernelRGBA = nullptr;

FlatLightProcessor::FlatLightProcessor(int size, int nPixels, float lightness)
    : imgSize(size), m_nPixels(nPixels), m_lightness(lightness) {
    m_img = new unsigned char[this->imgSize];

    // Allocate buffers for CUDA or OpenCL
    if (isCudaAvailable()) {
        cudaStreamCreate(&this->m_cudaStream);
        allocateCUDA();
        blockSize = 1024;
        gridSize = (m_nPixels + blockSize - 1) / blockSize;
    }
    else {
        allocateOpenCL();
    }
}

FlatLightProcessor::~FlatLightProcessor() {
    // Free image and color palette buffers
    delete[] m_img;

    // Free CUDA buffers
    if (isCudaAvailable()) {
        cudaFree(d_img);
        cudaStreamDestroy(m_cudaStream);
    }
    else {
        clReleaseMemObject(m_imgBuf);
    }
}

void FlatLightProcessor::allocateCUDA() {
    cudaMallocAsync(&d_img, imgSize, m_cudaStream);
}

void FlatLightProcessor::allocateOpenCL() {
    cl_int err;
    m_imgBuf = clCreateBuffer(globalContextOpenCL, CL_MEM_READ_WRITE, imgSize, nullptr, &err);
}

void FlatLightProcessor::process() const {
    (this->*processFunc)();
}

void FlatLightProcessor::processRGBA() const {
    (this->*processFuncRGBA)();
}

void FlatLightProcessor::processCUDA() const {
    cudaMemcpyAsync(d_img, m_img, imgSize, cudaMemcpyHostToDevice, m_cudaStream);

    flatLight_CUDA(gridSize, blockSize, m_cudaStream, d_img, m_nPixels, m_lightness);

    cudaMemcpyAsync(m_img, d_img, imgSize, cudaMemcpyDeviceToHost, m_cudaStream);
    cudaStreamSynchronize(m_cudaStream);
}

void FlatLightProcessor::processRGBA_CUDA() const {
    cudaMemcpyAsync(d_img, m_img, imgSize, cudaMemcpyHostToDevice, m_cudaStream);

    flatLightRGBA_CUDA(gridSize, blockSize, m_cudaStream, d_img, m_nPixels, m_lightness);

    cudaMemcpyAsync(m_img, d_img, imgSize, cudaMemcpyDeviceToHost, m_cudaStream);
    cudaStreamSynchronize(m_cudaStream);
}

void FlatLightProcessor::processOpenCL() const {
    // 1. Copy input image to device
    clEnqueueWriteBuffer(globalQueueOpenCL,
        m_imgBuf,
        CL_FALSE,  // Non-blocking write
        0,
        imgSize,
        m_img,
        0, nullptr, nullptr);

    clSetKernelArg(m_openclKernel, 0, sizeof(cl_mem), &m_imgBuf);
    clSetKernelArg(m_openclKernel, 1, sizeof(int), &m_nPixels);
    clSetKernelArg(m_openclKernel, 2, sizeof(float), &m_lightness);

    size_t globalSize = m_nPixels;
    cl_event kernelEvent;
    clEnqueueNDRangeKernel(globalQueueOpenCL,
        m_openclKernel,
        1,
        nullptr,
        &globalSize,
        nullptr,
        0, nullptr, &kernelEvent);

    // 4. Read back results (wait for kernel to complete)
    clEnqueueReadBuffer(globalQueueOpenCL,
        m_imgBuf,
        CL_TRUE,  // Blocking read
        0,
        imgSize,
        m_img,
        1, &kernelEvent, nullptr);

    // Release the kernel event
    clReleaseEvent(kernelEvent);
}

void FlatLightProcessor::processRGBA_OpenCL() const {
    // 1. Copy input image to device
    clEnqueueWriteBuffer(globalQueueOpenCL,
        m_imgBuf,
        CL_FALSE,  // Non-blocking write
        0,
        imgSize,
        m_img,
        0, nullptr, nullptr);

    clSetKernelArg(m_openclKernelRGBA, 0, sizeof(cl_mem), &m_imgBuf);
    clSetKernelArg(m_openclKernelRGBA, 1, sizeof(int), &m_nPixels);
    clSetKernelArg(m_openclKernelRGBA, 2, sizeof(float), &m_lightness);

    size_t globalSize = m_nPixels;
    cl_event kernelEvent;
    clEnqueueNDRangeKernel(globalQueueOpenCL,
        m_openclKernelRGBA,
        1,
        nullptr,
        &globalSize,
        nullptr,
        0, nullptr, &kernelEvent);

    // 4. Read back results (wait for kernel to complete)
    clEnqueueReadBuffer(globalQueueOpenCL,
        m_imgBuf,
        CL_TRUE,  // Blocking read
        0,
        imgSize,
        m_img,
        1, &kernelEvent, nullptr);

    // Release the kernel event
    clReleaseEvent(kernelEvent);
}

void FlatLightProcessor::setImage(const unsigned char* img) {
    memcpy(m_img, img, imgSize);
}

void FlatLightProcessor::upload(unsigned char* Dst) {
    memcpy(Dst, m_img, imgSize);
}

void FlatLightProcessor::init() {
    if (isCudaAvailable()) {
        FlatLightProcessor::processFunc = &FlatLightProcessor::processCUDA;
        FlatLightProcessor::processFuncRGBA = &FlatLightProcessor::processRGBA_CUDA;
    }
    else {
        FlatLightProcessor::processFunc = &FlatLightProcessor::processOpenCL;
        FlatLightProcessor::processFuncRGBA = &FlatLightProcessor::processRGBA_OpenCL;

        FlatLightProcessor::m_openclKernel = openclUtils::createKernelFromSource(globalContextOpenCL, globalDeviceOpenCL, flatLightOpenCLKernelSource, "flatLight_kernel");
        FlatLightProcessor::m_openclKernelRGBA = openclUtils::createKernelFromSource(globalContextOpenCL, globalDeviceOpenCL, flatLightRGBAOpenCLKernelSource, "flatLightRGBA_kernel");
    }
}