Followers 0

# Which of these two options is best?

## 9 posts in this topic

I am using SDL alongside C. Before, I was doing this:

dest.x-=campos.x;
dest.y-=campos.y;
SDL_BlitSurface(surface,NULL,screen,&dest);

I wrote a function and now use this instead of the above code:

draw(NULL,dest,surface);

In both instances, NULL is the texture clipping. Here is the function:

void draw(SDL_Rect clip, SDL_Rect ofs, SDL_Surface *s) {
ofs.x-=campos.x;
ofs.y-=campos.y;
SDL_BlitSurface(s,&clip,screen,&ofs);
}

Drawing graphics this way has saved me 0x109C bytes (4.1 KB) so far. I know this is more effective code-wise, but which method is best in terms of processing? I want to make the system requirements be as low as possible. Even if one method is as much as 1% more effective, please share. If you have numbers, statistics, or anything like that to back up your statement, please supply those. Is there a way I can improve the code I have shared? Something that bugs me is that I have to pass dest instead of &dest, as well as the whole surface, into the function instead of drawing directly.

Thank you! I look forward to your responses!

0

##### Share on other sites
Sorry, it has saved you 4.1 kb of what?
0

##### Share on other sites

Ultramailman: 4.1 KB of space when the project is compiled.

Servant of the Lord: You have taught me a few things in your post! Thanks! I see exactly what you are saying, but the lines of code I am using are a trivial matter and will never need to be changed. I had space and system requirement optimization in mind, but 4.1 KB isn't a whole lot of space, so if the new code requires more out of the system, it doesn't seem like a fair trade. I would like a way to test it and see which is more processor-efficient. When it comes to major pieces of code such as making something follow a path, using a function is far more necessary for organization, debugging, and so on, but when it all boils down to choosing between 1.) typing one line instead of three and 2.) having code that isn't as harsh on the processor, I feel it is necessary to go with the latter, especially when the code already works properly and does not need any improvement. Are you able to tell me which method I shared is more processor-effective? Thank you for your descriptive response, and a Reputation++; for you!

0

##### Share on other sites

Don't worry about picky little details for now. Develop your game, and then if you have performance problems fix them then. Unless you're running a 8080, you really have no need to worry about issues like this.

0

##### Share on other sites

More Reputation++; for you guys! I totally forgot about the compiler interpreting it and making it lower level either way. I am actually getting better results in my project by doing it the way I was before (without a function), so I will stick with that. Code can almost always be improved, so as long as it runs and there are not any memory leaks or anything, it is good enough for me, and definitely good enough for those who run it. Thank you very much for your descriptive responses and good luck with your own projects as well! :D

0

##### Share on other sites

Turns out you write programs for other people to read, not for compilers to read.  Oh, sure, most people write write-only code and rather than maintaining it or fixing the bugs, they just ignore the bugs or throw their work away and start again.  Then, there are productive successful projects.

The most important thing is to make your code understandable.  Clarity of purpose.  Clarity of meaning.  A function or class should have a single, clearly defined purpose and it should be named to describe that purpose.  70% to 80% of the cost of software is maintenance, and new code becomes old code in maintenance the moment after it's written.  You're likely the next programmer who has to understand what you wrote, so focus on clarity rather than premature picooptimization.

First make it work, then make it work fast.  You won't get to the second step if you can't understand what you wrote.

1

##### Share on other sites
I see. I think if you will only use that snippet once, then it's fine to just use the three lines of code directly and not write a function for it. But even if you do make a function for those three lines, compilers can notice that it is used once and inline it to the same thing.

If you use that snippet many times in many different places, then it is better to write a function for it, because it is better to repeat less code. If you absolutely must have the performance benefit of not using a function, you can consider making the function "static inline". This will allow your compiler to decide whether to inline it or not. If static inline is not possible with your compiler, you can define a function-like macro. That way you will have good looking and (possibly) faster code. Also note that inlining doesn't always make the code faster.
0

## Create an account

Register a new account