C++ faster execution [closed] - c++

Question

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 8 years ago.
Improve this question
I am using a while loop to execute some algorithms and what not, but it's not very fast, how would I allow my program to use more RAM? (Which I assumes is what limits it) It's currently sitting steadily at 504kB.
I am using
C::B 13.12
Windows 7 64bit
mingw32-g++.exe (I don't think I need the 64bit version unless I want to go over 4GB ram right?)
I apologize if this question has been asked and answered before, but I can't seem to find it if it has.
Edit: So this will scan 100 pixels, what is causing this to take 2.2 seconds?
#include <windows.h>
#include <iostream>
#include <wingdi.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
using namespace std;
void scan(HDC dc, int x, int y, int &r, int &g, int &b) {
COLORREF color = GetPixel(dc, x, y);
r = GetRValue(color);
g = GetGValue(color);
b = GetBValue(color);
}
int main() {
HDC dc = GetDC(NULL);
int r,g,b;
for(int i = 0; i < 100; i++) {
scan(dc,100,100 + i,r,g,b);
}
ReleaseDC(NULL, dc);
return 0;
}
2:nd edit: Is it possible to reduce this time without editing the code? I mean, it has to be limited by either my RAM or my CPU right?

Answer 1

Your program isn't limited to that small amount of memory. Since it's most likely compiled as a 32 bit application, it will be able to allocate up to 2 GB of ram by default.
So, no, memory is most likely not your issue, unless you constantly request and free it again (but even then it really depends on your code).
If your program is too slow, you might be able to use parallelization to get faster processing, but once again this really depends on your actual code.
You might also be able to use templates and instanziation to optimize your algorithm(s) at compile time, but yet again without knowing the code... no.
Since the edit:
The bottleneck is - as mentioned already - the repeated calls to GetPixel() which in itself is rather expensive, since there's work to be done that isn't cached etc.
Instead, you should copy the window contents to your own memory area and read the pixels (as bytes) directly.
You can follow this MSDN example. They're writing the bitmap contents/pixels to a file, but you'll essentially want to do the same, just read that data directly. Just look for the use of the variable lpbitmap to find the related lines.
In short you'll want to create a screenshot with BitBlt() to a bitmap and then copy those pixels utilizing GetDIBits():
// memory dc for the window
hdcMemDC = CreateCompatibleDC(hdcWindow);
// bitmap for the screenshot
hbmScreen = CreateCompatibleBitmap(hdcWindow, rcClient.right-rcClient.left, rcClient.bottom-rcClient.top);
// connect both
SelectObject(hdcMemDC,hbmScreen);
// copy the window contents
BitBlt(hdcMemDC, 0,0, rcClient.right-rcClient.left, rcClient.bottom-rcClient.top, hdcWindow, 0, 0, SRCCOPY);
// get the bitmap object
GetObject(hbmScreen, sizeof(BITMAP), &bmpScreen);
// access the bitmap
HANDLE hDIB = GlobalAlloc(GHND,dwBmpSize);
// lock the bitmap
char *lpbitmap = (char *)GlobalLock(hDIB);
// copy the pixel data
GetDIBits(hdcWindow, hbmScreen, 0, (UINT)bmpScreen.bmHeight, lpbitmap, (BITMAPINFO *)&bi, DIB_RGB_COLORS);
// now access lpbitmap inside your loop and later on clean up everything

Answer 2

The problem is with GetPixel. It is an extremely slow API. An alternate approach is to copy the screen to a memory buffer and then access the memory buffer.