Difference between revisions of "GPU610/TeamKCM"

From CDOT Wiki
Jump to: navigation, search
 
(59 intermediate revisions by 3 users not shown)
Line 2: Line 2:
 
= Team KCM =
 
= Team KCM =
 
== Team Members ==  
 
== Team Members ==  
−
# [mailto:bkim35@myseneca.ca?subject=gpu610 Byunghi Kim], Some responsibility  
+
# [mailto:bkim35@myseneca.ca?subject=gpu610 Byungho Kim], Some responsibility  
 
# [mailto:tchung22@myseneca.ca?subject=gpu610 Taeyang Chung], Some responsibility
 
# [mailto:tchung22@myseneca.ca?subject=gpu610 Taeyang Chung], Some responsibility
 
# [mailto:symoon@myseneca.ca?subject=gpu610 SeungYeon Moon], Some responsibility
 
# [mailto:symoon@myseneca.ca?subject=gpu610 SeungYeon Moon], Some responsibility
Line 10: Line 10:
 
== Progress ==
 
== Progress ==
 
=== Assignment 1 ===
 
=== Assignment 1 ===
−
====Byunghi's Findings====
+
====Byungho's Findings====
−
......
+
=====Introduction=====
 +
To enhance image quality, most of S/W use convolution algorithm with sharpen kernel. The convolution algorithm is based on pixel window and kernel multiplication and the kernel which consists of 3 x 3 matrix or 5 x 5 matrix. Therefore, as we learned at the GPU610 class, I can speed up the program using CUDA library
 +
 
 +
 
 +
 
 +
{| class="wikitable"
 +
|+
 +
! Original Image !! Enhanced Image
 +
|-
 +
| [[Image:Original.png|frame]] || [[Image:Enhanced.png|frame]]
 +
|}
 +
 
 +
These two images show the result of the S/W. Left picture is the original image and right picture is obtained by the S/W. 3x3 matrix sharpen kernel is used.
 +
 
 +
=====Convolution=====
 +
 
 +
{| border="1" style="border-collapse:collapse" class="wikitable"
 +
|+ 6 x 4 Pixel Small Image
 +
|-
 +
| IP<sub>11</sub> || IP<sub>12</sub> || IP<sub>13</sub> || IP<sub>14</sub> || IP<sub>15</sub> || IP<sub>16</sub>
 +
|-
 +
| IP<sub>21</sub> || IP<sub>22</sub> || IP<sub>23</sub> || IP<sub>24</sub> || IP<sub>25</sub> || IP<sub>26</sub>
 +
|-
 +
| IP<sub>31</sub> || IP<sub>32</sub> || IP<sub>33</sub> || IP<sub>34</sub> || IP<sub>35</sub> || IP<sub>36</sub>
 +
|-
 +
| IP<sub>41</sub> || IP<sub>42</sub> || IP<sub>43</sub> || IP<sub>44</sub> || IP<sub>45</sub> || IP<sub>46</sub>
 +
|-
 +
|}
 +
 
 +
 
 +
 
 +
{|  border="1" style="border-collapse:collapse"  class="wikitable"
 +
|+ Kernel
 +
|-
 +
|  K<sub>0</sub> || K<sub>1</sub> || K<sub>2</sub>
 +
|-
 +
|  K<sub>3</sub> || K<sub>4</sub> || k<sub>5</sub>
 +
|-
 +
|  k<sub>6</sub> || k<sub>7</sub> || k<sub>8</sub>
 +
|-
 +
|}
 +
 
 +
 
 +
The 33th output pixel will be given by
 +
 
 +
OP<sub>33</sub> = IP<sub>22</sub> x K<sub>0</sub> + IP<sub>23</sub> x K<sub>1</sub> + IP<sub>24</sub> x K<sub>2</sub> + IP<sub>32</sub> x K<sub>3</sub> + IP<sub>33</sub> x K<sub>4</sub> + IP<sub>34</sub> x K<sub>5</sub> + IP<sub>42</sub> x K<sub>6</sub> + IP<sub>43</sub> x K<sub>7</sub> + IP<sub>44</sub> x K<sub>8</sub>
 +
 
 +
=====Analysis=====
 +
The profile shows that filter function occupies the most of processing time.
 +
<pre>
 +
  %  cumulative  self              self    total         
 +
time  seconds  seconds    calls  ns/call  ns/call  name 
 +
93.59      1.46    1.46                            filter(image_t*, double*, int, double, double)
 +
  6.41      1.56    0.10  5038848    19.85    19.85  put_pixel_unsafe(image_t*, unsigned int, unsigned int, unsigned char, unsigned char, unsigned char)
 +
  0.00      1.56    0.00        2    0.00    0.00  alloc_img(unsigned int, unsigned int)
 +
  0.00      1.56    0.00        1    0.00    0.00  _GLOBAL__sub_I__Z7get_ppmP8_IO_FILE
 +
  0.00      1.56    0.00        1    0.00    0.00  get_ppm(_IO_FILE*)
 +
</pre>
 +
 
 +
 
 +
In the filter function, 3 x 3 mask walk through all RGB pixels and calculate convolution of them.
 +
<pre>
 +
image filter(image im, double *K, int Ks, double divisor, double offset)
 +
{ 
 +
    image oi;
 +
    unsigned int ix, iy, l;
 +
    int kx, ky;
 +
    double cp[3];
 +
    oi = alloc_img(im->width, im->height);
 +
    if ( oi != NULL ) {
 +
        for(ix=0; ix < im->width; ix++) {
 +
            for(iy=0; iy < im->height; iy++) {
 +
                cp[0] = cp[1] = cp[2] = 0.0;
 +
                for(kx=-Ks; kx <= Ks; kx++) {
 +
                    for(ky=-Ks; ky <= Ks; ky++) {
 +
                        for(l=0; l<3; l++)
 +
                            cp[l] += (K[(kx+Ks) +
 +
                                    (ky+Ks)*(2*Ks+1)]/divisor) *
 +
                                ((double)GET_PIXEL_CHECK(im, ix+kx, iy+ky, l)) + offset;
 +
                    }
 +
                }
 +
                for(l=0; l<3; l++)
 +
                    cp[l] = (cp[l]>255.0) ? 255.0 : ((cp[l]<0.0) ? 0.0 : cp[l]) ;
 +
                put_pixel_unsafe(oi, ix, iy,
 +
                        (color_component)cp[1],
 +
                        (color_component)cp[2],
 +
                        (color_component)cp[0]);
 +
            }
 +
        }
 +
        return oi;
 +
    }
 +
  return NULL;
 +
}
 +
</pre>
 +
 
 
====Taeyang's Findings====
 
====Taeyang's Findings====
−
......
+
My program is to solve the heat transfer problem using explicit method. I found this source code at http://www.pengfeidu.net/Media/pdf/1dUnsteadyHeatTransferCode.pdf
 +
 
 +
The 'Solve()' function takes 3 arguments and as t-value decreases the processing time increases as shown below
 +
[[File:A1_3.png]]
 +
 
 +
 
 +
[[File:A1_4.png]]
 +
 
 +
The 'Solve()' function:
 +
 
 +
      void solve(){
 +
                int j=0;
 +
                while(j<tTotal/dt){
 +
                        solutionNew[0]=10;
 +
                        solutionNew[xGridNum]=120;
 +
                        for(int k=1;k<xGridNum;k++){
 +
                            solutionNew[k] = solutionLast[k]+dt*(K/dx/dx*
 +
                                      (solutionLast[k-1]+solutionLast[k+1]-2.*solutionLast[k]
 +
                                      +source(x[k],j*dt)));
 +
                        }
 +
                        j++;
 +
                        solutionLast=solutionNew;
 +
                }
 +
        }
 +
 
 +
By Implementing parallel solution, I believe I can shorten the processing time.
 +
 
 
====SeungYeon's Findings====
 
====SeungYeon's Findings====
 
There are many types of image processing or operations can be done. Some of examples are rotating,re-sizing,blurring, etc...   
 
There are many types of image processing or operations can be done. Some of examples are rotating,re-sizing,blurring, etc...   
Line 25: Line 145:
 
=====Here is two test runs using same image but in different file size=====
 
=====Here is two test runs using same image but in different file size=====
  
−
=====Test run with an image size of 500KB=====
+
====Test run with an image size of 500KB====
 
[[File:Test_run_500KB_total.png]]
 
[[File:Test_run_500KB_total.png]]
  
 +
Total Time takes is 12 seconds
  
  
Line 33: Line 154:
  
  
−
=====Test run with an image size of 2MB=====
+
====Test run with an image size of 2MB====
 
[[File:Test_run_2MB_total.png]]
 
[[File:Test_run_2MB_total.png]]
  
 +
Total Time takes is 14 seconds
  
  
 
[[File:Test_run_2MB.png]]
 
[[File:Test_run_2MB.png]]
 +
 +
 +
====SeungYeons' observation for As_1====
 +
Changing contrast and brightness of an image in the function is implemented with a scale * matrix operation.
 +
M_A(x,y) = ALPHA * M_B(x,y) + BETA;
 +
  
 
=== Assignment 2 ===
 
=== Assignment 2 ===
 +
For our Assignment 2 we decided to work with heat transfer problem.
 +
 +
To compare the speed difference between the original code and parallelized code, we first profiled the original code in matrix environment an parallelized code in school computer. (using gporf and nvvp)
 +
 +
==== Profile results ====
 +
Each Program was tested with 5 different input data.(500,1000,2500,5000,10000)
 +
 +
=====Profile #1 (Original code) =====
 +
Sample Data = 500
 +
 +
[[File:500-Original.png]]
 +
 +
 +
Sample Data = 2500
 +
 +
[[File:2500-Original.png]]
 +
 +
 +
Sample Data = 10000
 +
 +
[[File:10000-Original.png]]
 +
 +
 +
=====Profile #2 (GPU implementation version) =====
 +
Sample Data = 500
 +
 +
[[File:500.png]]
 +
 +
 +
 +
Sample Data = 2500
 +
 +
[[File:2500.png]]
 +
 +
 +
Sample Data = 10000
 +
 +
[[File:10000.png]]
 +
 +
 +
===== Summary of profiling =====
 +
We tested CPU and GPU implementation version with same data sizes and showed significant speedup in result after using kernel.
 +
 +
 +
==== Summary of profiling in Chart ====
 +
 +
 +
[[File:Assignment2.png]]
 +
 +
 +
==== Charts with some of main operation(function) calls for CPU version and GPU(kernel) ====
 +
 +
===== Original =====
 +
 +
[[File:Originalcode.png]]
 +
 +
===== GPU (kernel) =====
 +
 +
[[File:gpu(kernal).png]]
 +
 +
==== Screen shot of solve function in CPU and GPU(kernel) versions ====
 +
 +
===== CPU =====
 +
 +
[[File:Original-code.png]]
 +
 +
===== GPU(Solve) and Kernel=====
 +
 +
[[File:GPU(kernel).png]]
 +
 +
[[File:gpu-code.png]]
 +
 +
 +
 
=== Assignment 3 ===
 
=== Assignment 3 ===
 +
For assignment 3, we were checking everything we can do to improve the performance and found 2 thing we can do.
 +
Firstly, in the while loop, there were 6 times of memory copy functions called, and we found that we can reduce 6 times to 1 time by using device address pointer switching.
 +
Furthermore, we found that if the sample number n is less that 1024, we can use shared memory in the kernel.
 +
 +
==== Reducing Memory Copy Calls====
 +
 +
===== Assignment 2 Code =====
 +
<syntaxhighlight lang="cpp">
 +
void solve(){
 +
    int j=0;
 +
    int d;
 +
 +
    cudaDeviceProp prop;
 +
    cudaGetDevice(&d);
 +
    cudaGetDeviceProperties(&prop, d);
 +
    unsigned ntpb = prop.maxThreadsDim[0];
 +
    unsigned ntpg = ntpb * prop.maxGridSize[0];
 +
    if (xGridNum > ntpg) {
 +
        xGridNum = ntpg;
 +
        std::cout << "n reduced to " << xGridNum << std::endl;
 +
    }
 +
 +
    double* d_x;
 +
    double* d_solutionLast;
 +
    double* d_solutionNew;
 +
 +
    cudaMalloc((void**)&d_x, (xGridNum + 1) * sizeof(float));     
 +
    cudaMalloc((void**)&d_solutionLast, xGridNum * sizeof(double));
 +
    cudaMalloc((void**)&d_solutionNew, xGridNum * sizeof(double));
 +
 +
    while(j<tTotal/dt){
 +
        solutionNew[0]=10;
 +
        solutionNew[xGridNum]=120;
 +
 +
        cudaMemcpy(d_x, x, (xGridNum + 1) * sizeof(float), cudaMemcpyHostToDevice);
 +
        cudaMemcpy(d_solutionLast, solutionLast, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
 +
        cudaMemcpy(d_solutionNew, solutionNew, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
 +
 +
     
 +
 +
        kernel<<<(xGridNum + ntpb - 1) / ntpb, ntpb>>>(d_solutionNew, d_solutionLast, d_x, xGridNum, dt, dx, K, j);
 +
        j++;
 +
 +
        cudaMemcpy(x, d_x, (xGridNum + 1) * sizeof(float), cudaMemcpyDeviceToHost);
 +
        cudaMemcpy(solutionLast, d_solutionLast, xGridNum * sizeof(double), cudaMemcpyDeviceToHost);
 +
        cudaMemcpy(solutionNew, d_solutionNew, xGridNum * sizeof(double), cudaMemcpyDeviceToHost);
 +
 +
        solutionLast=solutionNew;
 +
    }
 +
 +
    for(j = 0; j <= xGridNum; j++){
 +
      std::cout << solutionLast[j] << endl;
 +
    }
 +
 +
    cudaFree(d_x);
 +
    cudaFree(d_solutionLast);
 +
    cudaFree(d_solutionNew);
 +
}
 +
</syntaxhighlight>
 +
 +
 +
===== Improved Code =====
 +
<syntaxhighlight lang="cpp">
 +
void solve(){
 +
    int j=0;
 +
    int d;
 +
 
 +
    cudaDeviceProp prop;
 +
    cudaGetDevice(&d);
 +
    cudaGetDeviceProperties(&prop, d);
 +
    unsigned ntpb = prop.maxThreadsDim[0];
 +
    unsigned ntpg = ntpb * prop.maxGridSize[0];
 +
    if (xGridNum > ntpg) {
 +
        xGridNum = ntpg;
 +
        std::cout << "n reduced to " << xGridNum << std::endl;
 +
    }
 +
 +
    double* d_x;
 +
    double* d_solutionA;
 +
    double* d_solutionB;
 +
 +
    cudaMalloc((void**)&d_x, (xGridNum + 1) * sizeof(float));     
 +
    cudaMalloc((void**)&d_solutionA, xGridNum * sizeof(double));
 +
    cudaMalloc((void**)&d_solutionB, xGridNum * sizeof(double));
 +
 
 +
    double* d_solutionTemp;
 +
 +
    cudaMemcpy(d_x, x, (xGridNum + 1) * sizeof(float), cudaMemcpyHostToDevice);
 +
    cudaMemcpy(d_solutionA, solutionNew, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
 +
    cudaMemcpy(d_solutionB, solutionLast, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
 +
 +
    std::ofstream myfile;
 +
    myfile.open ("output.txt");                 
 +
 
 +
    // A = new, B = last
 +
    while( j < tTotal / dt){                   
 +
        kernel<<<(xGridNum + ntpb - 1) / ntpb, ntpb>>>(d_solutionA, d_solutionB, d_x, xGridNum, dt, dx, K, j);
 +
        j++;     
 +
        cudaMemcpy(solutionNew, d_solutionA, xGridNum * sizeof(double), cudaMemcpyDeviceToHost);
 +
        /*
 +
        myfile << "Time" << tTotal/dt << std::endl;
 +
        for(int i = 0; i <= xGridNum; i++){                                 
 +
          myfile << solutionNew[i] << ":";
 +
        }
 +
        */
 +
        d_solutionTemp = d_solutionA;
 +
        d_solutionA = d_solutionB;
 +
        d_solutionB = d_solutionTemp;
 +
    }
 +
 +
    myfile.close();
 +
 
 +
    cudaFree(d_x);
 +
    cudaFree(d_solutionA);
 +
    cudaFree(d_solutionB);
 +
 +
    cudaDeviceReset(); // Important
 +
}
 +
</syntaxhighlight>
 +
 +
 +
==== Using Shared Memory ====
 +
 +
===== Assignment 2 Code =====
 +
<syntaxhighlight lang="cpp">
 +
__global__ void kernel(double* solutionNew, double* solutionLast, double* x, int n, float dt, float dx, float K, int j){
 +
        int k = blockIdx.x * blockDim.x + threadIdx.x;
 +
 +
        if( k < n){
 +
                k += 1;
 +
                solutionNew[k] = solutionLast[k]+dt*(K/dx/dx*
 +
                                (solutionLast[k-1]+solutionLast[k+1]-2.*solutionLast[k]
 +
                                +(x[k]*j*dt*1000))
 +
                                );
 +
        }
 +
}
 +
</syntaxhighlight>
 +
===== Improved Code =====
 +
<syntaxhighlight lang="cpp">
 +
__global__ void kernel(double* solutionNew, double* solutionLast, double* x, int n, float dt, float dx, float K, int j){
 +
        int k = blockIdx.x * blockDim.x + threadIdx.x;
 +
     
 +
        double *t_c;
 +
        __shared__ double s_c[1024];
 +
 
 +
        if ( n < 1024 && k < (n + 1))
 +
            s_c[k] = solutionLast[k];
 +
        __syncthreads();
 +
             
 +
        if( n < 1024) t_c = s_c;
 +
        else t_c = solutionLast;
 +
 +
                     
 +
        if( k == 0){
 +
            solutionNew[k] = 10;
 +
        } else if( k == n){
 +
            solutionNew[k] = 120;
 +
        }
 +
                     
 +
        if( k < n){
 +
                k += 1;
 +
                solutionNew[k] = t_c[k]+dt*(K/dx/dx*
 +
                                (t_c[k-1]+t_c[k+1]-2.*t_c[k]
 +
                                +(x[k]*j*dt*1000))
 +
                                );
 +
        }             
 +
}
 +
</syntaxhighlight>
 +
 +
 +
==== Profile Result ====
 +
 +
===== Sample Data = 500 =====
 +
[[File:as3v1-500.png]]
 +
 +
===== Sample Data = 1000 =====
 +
[[File:as3v1-2500.png]]
 +
 +
 +
===== Sample Data = 5000 =====
 +
[[File:as3v1-5000.png]]
 +
 +
 +
===== Sample Data =  10000 =====
 +
[[File:as3v1-10000.png]]
 +
 +
==== Chart ====
 +
[[File:a3 duration chart.png]]
 +
 +
 +
=== Assignment 3 v2 ===
 +
==== Description of v2 ====
 +
Changed Double precision to single precision float
 +
 +
==== Sample Date = 500 ====
 +
[[File:as3v2-500.png]]
 +
 +
==== Sample Date = 1000 ====
 +
[[File:as3v2-1000.png]]
 +
 +
==== Sample Date = 5000 ====
 +
[[File:as3v2-5000.png]]
 +
 +
==== Sample Date = 10000 ====
 +
[[File:as3v2-10000.png]]
 +
 +
==== Char ====
 +
[[File:Table.png]]
 +
 +
 +
 +
==== Conclusions / Problems Encountered ====
 +
Using CUDA, Our team achieved around 8000% speed up in total run time compare to original project and final result. We were certain that by implementing kernel to the original project will result in huge speed up because a calculation was done in a loop(serial) to get each heat points in specific time. To accomplish the project, first we focused on understanding the original project to find out the "hot spot" in the code and different variables and their uses. Working with heat equation problem was challenging, because we had to understand how heat points are calculated with a equation and had to find out which part/variables are needed and not to develop a kernel. The original project did not have any dependent variables given from user input, so we had to decide which part of the equation we want to be dependent. As we develop kernel and work with different resources in the program we had to decide how to work with and approach different memories in the program(accessing global memory, using shared memory..etc).One of the principle logic the program used is, to get a new heat value a last heat value is needed for calculation. So, when developing kernel our team had to figure out how many times memory copies are need and what parts can be done in GPU side(shared,register memories...) In result, we managed to minimize the number of memory copies from CPU to GPU and use more shared and register memories.

Latest revision as of 14:20, 5 December 2014


GPU610/DPS915 | Student List | Group and Project Index | Student Resources | Glossary

Team KCM

Team Members

  1. Byungho Kim, Some responsibility
  2. Taeyang Chung, Some responsibility
  3. SeungYeon Moon, Some responsibility

Email All

Progress

Assignment 1

Byungho's Findings

Introduction

To enhance image quality, most of S/W use convolution algorithm with sharpen kernel. The convolution algorithm is based on pixel window and kernel multiplication and the kernel which consists of 3 x 3 matrix or 5 x 5 matrix. Therefore, as we learned at the GPU610 class, I can speed up the program using CUDA library


Original Image Enhanced Image
Original.png
Enhanced.png

These two images show the result of the S/W. Left picture is the original image and right picture is obtained by the S/W. 3x3 matrix sharpen kernel is used.

Convolution
6 x 4 Pixel Small Image
IP11 IP12 IP13 IP14 IP15 IP16
IP21 IP22 IP23 IP24 IP25 IP26
IP31 IP32 IP33 IP34 IP35 IP36
IP41 IP42 IP43 IP44 IP45 IP46


Kernel
K0 K1 K2
K3 K4 k5
k6 k7 k8


The 33th output pixel will be given by

OP33 = IP22 x K0 + IP23 x K1 + IP24 x K2 + IP32 x K3 + IP33 x K4 + IP34 x K5 + IP42 x K6 + IP43 x K7 + IP44 x K8

Analysis

The profile shows that filter function occupies the most of processing time. 

  %   cumulative   self              self     total          
 time   seconds   seconds    calls  ns/call  ns/call  name   
 93.59      1.46     1.46                             filter(image_t*, double*, int, double, double)
  6.41      1.56     0.10  5038848    19.85    19.85  put_pixel_unsafe(image_t*, unsigned int, unsigned int, unsigned char, unsigned char, unsigned char)
  0.00      1.56     0.00        2     0.00     0.00  alloc_img(unsigned int, unsigned int)
  0.00      1.56     0.00        1     0.00     0.00  _GLOBAL__sub_I__Z7get_ppmP8_IO_FILE
  0.00      1.56     0.00        1     0.00     0.00  get_ppm(_IO_FILE*)


In the filter function, 3 x 3 mask walk through all RGB pixels and calculate convolution of them. 

image filter(image im, double *K, int Ks, double divisor, double offset)
{  
    image oi;
    unsigned int ix, iy, l;
    int kx, ky;
    double cp[3];
    oi = alloc_img(im->width, im->height);
    if ( oi != NULL ) {
        for(ix=0; ix < im->width; ix++) {
            for(iy=0; iy < im->height; iy++) {
                cp[0] = cp[1] = cp[2] = 0.0;
                for(kx=-Ks; kx <= Ks; kx++) {
                    for(ky=-Ks; ky <= Ks; ky++) {
                        for(l=0; l<3; l++)
                            cp[l] += (K[(kx+Ks) +
                                    (ky+Ks)*(2*Ks+1)]/divisor) *
                                ((double)GET_PIXEL_CHECK(im, ix+kx, iy+ky, l)) + offset;
                    }
                }
                for(l=0; l<3; l++)
                    cp[l] = (cp[l]>255.0) ? 255.0 : ((cp[l]<0.0) ? 0.0 : cp[l]) ;
                put_pixel_unsafe(oi, ix, iy,
                        (color_component)cp[1],
                        (color_component)cp[2],
                        (color_component)cp[0]);
            }
        }
        return oi;
    }
   return NULL;
}

Taeyang's Findings

My program is to solve the heat transfer problem using explicit method. I found this source code at http://www.pengfeidu.net/Media/pdf/1dUnsteadyHeatTransferCode.pdf

The 'Solve()' function takes 3 arguments and as t-value decreases the processing time increases as shown below A1 3.png


A1 4.png

The 'Solve()' function: 

      void solve(){
               int j=0;
               while(j<tTotal/dt){
                       solutionNew[0]=10;
                       solutionNew[xGridNum]=120;
                       for(int k=1;k<xGridNum;k++){
                            solutionNew[k] = solutionLast[k]+dt*(K/dx/dx*
                                      (solutionLast[k-1]+solutionLast[k+1]-2.*solutionLast[k]
                                      +source(x[k],j*dt)));
                       }
                       j++;
                       solutionLast=solutionNew;
               }
       }

By Implementing parallel solution, I believe I can shorten the processing time.

SeungYeon's Findings

There are many types of image processing or operations can be done. Some of examples are rotating,re-sizing,blurring, etc... For Assignment 1, I decided to work with one of the operation that can be done with an image in c++, which deals with brightness and contrast of an image. I was able to find a open source code from "openCV" website it stands for Open Source Computer Vision, is open source libraries developed by Intel for image and video processing. The libraries can be installed in any platform, and i was able to install it in my windows platform, and all test runs all done using visual studio 2013, Test file is compiled and profiled using performance wizard vs13. Following website openCV.org is an official documentation website of openCV libraries, with code examples for many image and video type operations

The open source code for the program requires 3 user inputs which are path to an image and alpha,beta values which will be multiplied and added to each pixels' in the image. To make sure test run is accurate, i had to change some code in the program and hard code in the values for alpha and beta.

Here is two test runs using same image but in different file size

Test run with an image size of 500KB

Test run 500KB total.png

Total Time takes is 12 seconds


Test run 500KB.png


Test run with an image size of 2MB

Test run 2MB total.png

Total Time takes is 14 seconds


Test run 2MB.png


SeungYeons' observation for As_1

Changing contrast and brightness of an image in the function is implemented with a scale * matrix operation. M_A(x,y) = ALPHA * M_B(x,y) + BETA;


Assignment 2

For our Assignment 2 we decided to work with heat transfer problem.

To compare the speed difference between the original code and parallelized code, we first profiled the original code in matrix environment an parallelized code in school computer. (using gporf and nvvp)

Profile results

Each Program was tested with 5 different input data.(500,1000,2500,5000,10000)

Profile #1 (Original code)

Sample Data = 500

500-Original.png


Sample Data = 2500

2500-Original.png


Sample Data = 10000

10000-Original.png


Profile #2 (GPU implementation version)

Sample Data = 500

500.png


Sample Data = 2500

2500.png


Sample Data = 10000

10000.png


Summary of profiling

We tested CPU and GPU implementation version with same data sizes and showed significant speedup in result after using kernel.


Summary of profiling in Chart

Assignment2.png


Charts with some of main operation(function) calls for CPU version and GPU(kernel)

Original

Originalcode.png

GPU (kernel)

Gpu(kernal).png

Screen shot of solve function in CPU and GPU(kernel) versions

CPU

Original-code.png

GPU(Solve) and Kernel

GPU(kernel).png

Gpu-code.png


Assignment 3

For assignment 3, we were checking everything we can do to improve the performance and found 2 thing we can do. Firstly, in the while loop, there were 6 times of memory copy functions called, and we found that we can reduce 6 times to 1 time by using device address pointer switching. Furthermore, we found that if the sample number n is less that 1024, we can use shared memory in the kernel.

Reducing Memory Copy Calls

Assignment 2 Code
void solve(){
    int j=0;
    int d;

    cudaDeviceProp prop;
    cudaGetDevice(&d);
    cudaGetDeviceProperties(&prop, d);
    unsigned ntpb = prop.maxThreadsDim[0];
    unsigned ntpg = ntpb * prop.maxGridSize[0];
    if (xGridNum > ntpg) {
        xGridNum = ntpg;
        std::cout << "n reduced to " << xGridNum << std::endl;
    }

    double* d_x;
    double* d_solutionLast;
    double* d_solutionNew;

    cudaMalloc((void**)&d_x, (xGridNum + 1) * sizeof(float));       
    cudaMalloc((void**)&d_solutionLast, xGridNum * sizeof(double));
    cudaMalloc((void**)&d_solutionNew, xGridNum * sizeof(double));

    while(j<tTotal/dt){
        solutionNew[0]=10;
        solutionNew[xGridNum]=120;

        cudaMemcpy(d_x, x, (xGridNum + 1) * sizeof(float), cudaMemcpyHostToDevice);
        cudaMemcpy(d_solutionLast, solutionLast, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
        cudaMemcpy(d_solutionNew, solutionNew, xGridNum * sizeof(double), cudaMemcpyHostToDevice);

       

        kernel<<<(xGridNum + ntpb - 1) / ntpb, ntpb>>>(d_solutionNew, d_solutionLast, d_x, xGridNum, dt, dx, K, j);
        j++;

        cudaMemcpy(x, d_x, (xGridNum + 1) * sizeof(float), cudaMemcpyDeviceToHost);
        cudaMemcpy(solutionLast, d_solutionLast, xGridNum * sizeof(double), cudaMemcpyDeviceToHost);
        cudaMemcpy(solutionNew, d_solutionNew, xGridNum * sizeof(double), cudaMemcpyDeviceToHost);

        solutionLast=solutionNew;
    }

    for(j = 0; j <= xGridNum; j++){
      std::cout << solutionLast[j] << endl;
    }

    cudaFree(d_x);
    cudaFree(d_solutionLast);
    cudaFree(d_solutionNew);
}


Improved Code
void solve(){
    int j=0;
    int d;
  
    cudaDeviceProp prop;
    cudaGetDevice(&d);
    cudaGetDeviceProperties(&prop, d);
    unsigned ntpb = prop.maxThreadsDim[0];
    unsigned ntpg = ntpb * prop.maxGridSize[0];
    if (xGridNum > ntpg) {
        xGridNum = ntpg;
        std::cout << "n reduced to " << xGridNum << std::endl;
    }

    double* d_x;
    double* d_solutionA;
    double* d_solutionB;

    cudaMalloc((void**)&d_x, (xGridNum + 1) * sizeof(float));       
    cudaMalloc((void**)&d_solutionA, xGridNum * sizeof(double));
    cudaMalloc((void**)&d_solutionB, xGridNum * sizeof(double));
   
    double* d_solutionTemp;

    cudaMemcpy(d_x, x, (xGridNum + 1) * sizeof(float), cudaMemcpyHostToDevice);
    cudaMemcpy(d_solutionA, solutionNew, xGridNum * sizeof(double), cudaMemcpyHostToDevice);
    cudaMemcpy(d_solutionB, solutionLast, xGridNum * sizeof(double), cudaMemcpyHostToDevice);

    std::ofstream myfile;
    myfile.open ("output.txt");                   
  
    // A = new, B = last
    while( j < tTotal / dt){                    
        kernel<<<(xGridNum + ntpb - 1) / ntpb, ntpb>>>(d_solutionA, d_solutionB, d_x, xGridNum, dt, dx, K, j);
        j++;       
        cudaMemcpy(solutionNew, d_solutionA, xGridNum * sizeof(double), cudaMemcpyDeviceToHost); 
        /*
        myfile << "Time" << tTotal/dt << std::endl;
        for(int i = 0; i <= xGridNum; i++){                                   
          myfile << solutionNew[i] << ":";
        }
        */
        d_solutionTemp = d_solutionA;
        d_solutionA = d_solutionB;
        d_solutionB = d_solutionTemp;
    }

    myfile.close();
   
    cudaFree(d_x);
    cudaFree(d_solutionA);
    cudaFree(d_solutionB);
 
    cudaDeviceReset(); // Important
}


Using Shared Memory

Assignment 2 Code
__global__ void kernel(double* solutionNew, double* solutionLast, double* x, int n, float dt, float dx, float K, int j){
        int k = blockIdx.x * blockDim.x + threadIdx.x;

        if( k < n){
                k += 1;
                solutionNew[k] = solutionLast[k]+dt*(K/dx/dx*
                                (solutionLast[k-1]+solutionLast[k+1]-2.*solutionLast[k]
                                 +(x[k]*j*dt*1000))
                                );
        }
}
Improved Code
__global__ void kernel(double* solutionNew, double* solutionLast, double* x, int n, float dt, float dx, float K, int j){
        int k = blockIdx.x * blockDim.x + threadIdx.x;
       
        double *t_c;
        __shared__ double s_c[1024];
   
        if ( n < 1024 && k < (n + 1))
            s_c[k] = solutionLast[k];
        __syncthreads();
               
        if( n < 1024) t_c = s_c;
        else t_c = solutionLast;

                       
        if( k == 0){
            solutionNew[k] = 10;
        } else if( k == n){
            solutionNew[k] = 120;
        }
                       
        if( k < n){
                k += 1;
                solutionNew[k] = t_c[k]+dt*(K/dx/dx*
                                (t_c[k-1]+t_c[k+1]-2.*t_c[k]
                                 +(x[k]*j*dt*1000))
                                );
        }               
}


Profile Result

Sample Data = 500

As3v1-500.png

Sample Data = 1000

File:As3v1-2500.png


Sample Data = 5000

As3v1-5000.png


Sample Data = 10000

As3v1-10000.png

Chart

A3 duration chart.png


Assignment 3 v2

Description of v2

Changed Double precision to single precision float

Sample Date = 500

As3v2-500.png

Sample Date = 1000

As3v2-1000.png

Sample Date = 5000

As3v2-5000.png

Sample Date = 10000

As3v2-10000.png

Char

Table.png


Conclusions / Problems Encountered

Using CUDA, Our team achieved around 8000% speed up in total run time compare to original project and final result. We were certain that by implementing kernel to the original project will result in huge speed up because a calculation was done in a loop(serial) to get each heat points in specific time. To accomplish the project, first we focused on understanding the original project to find out the "hot spot" in the code and different variables and their uses. Working with heat equation problem was challenging, because we had to understand how heat points are calculated with a equation and had to find out which part/variables are needed and not to develop a kernel. The original project did not have any dependent variables given from user input, so we had to decide which part of the equation we want to be dependent. As we develop kernel and work with different resources in the program we had to decide how to work with and approach different memories in the program(accessing global memory, using shared memory..etc).One of the principle logic the program used is, to get a new heat value a last heat value is needed for calculation. So, when developing kernel our team had to figure out how many times memory copies are need and what parts can be done in GPU side(shared,register memories...) In result, we managed to minimize the number of memory copies from CPU to GPU and use more shared and register memories.

Retrieved from "https://wiki.cdot.senecapolytechnic.ca/w/index.php?title=GPU610/TeamKCM&oldid=107114"