Which Renders Faster?

Wed, 21 Jul 2010 18:40:52 -0500

Well, in the ultimate goal of speed optimization, I have to ask: Which renders faster?

or elements?

Of course, each would be styled (sometimes heavily), with the

s being styled to act like tables.

If nobody knows, is there a way to benchmark them both?

Inactivity

Thu, 15 Jul 2010 15:02:29 -0500

I apologize for my inactivity for those of you that still check this site faithfully.

Between my business and my blog I've been quite busy these days - but I haven't forgot about this site.

I hope to be more active in the future, thank you to all of you who have stayed.

Calculator

Fri, 19 Mar 2010 11:07:12 -0500

I made a very simple calculator program based on the CodeIgniter framework, as a test figuring out how to work with the framework.

If you find any bugs please tell me.

Check it out!

AnLib

Thu, 11 Mar 2010 19:08:24 -0600

AnLib is a library of PHP classes I have created to help create and deploy PHP applications swiftly. It is still in alpha stages.

Get it on Sourceforge.

PHPLiveX

Sun, 07 Mar 2010 06:47:17 -0600

Just stumbled across this awesome tool for PHP developers. PHPLiveX allows you to call PHP functions through JavaScript.

To find out more and download it, visit the website.

Just some questions.

Sat, 20 Feb 2010 22:50:31 -0600

Hello, I briefly mentioned my dad's business in another post of mine. Well he runs a water business, Clear Water Plus, where he provides purified water and the likes. He has people purchase pre-paid water cards which he keeps the physical cards and marks them off as people use the gallons. In my college course, we were just getting into inputing and storing phone book data, phone numbers and names etc and then searching and referencing them back. I was going to take that knowledge and tweak it into a program for him to store the water card information and look it up on his computer instead of digging through hundreds of cards on a daily basis. I didn't get that far however. I was wondering if C++ is my best viable option for that type of program. I was going to have him use an excel sheet, but he is very computer illiterate and he kept somehow managing to not save/delete stuff while attempting to search and the spread sheet. So a program is a more viable option considering input control and user friendliness.

I just wanted to know, if anybody could give me a head start link to the basics I would need to create the simple program and get it functioning in a user-friendly format for him. I'm sure I could find plenty of guides for this type of stuff if I ran over to google, but I like this site and am trying to encourage it's activity. One major reason also, is I'm not sure C++ is my best choice, because I would like to eventually adapt the program into a web interface for him, so people could purchase/charge their water cards online. Databasing, security, and encryption can be terribly nasty though, so I'm assuming the best solution to this would be to just find a free or relatively cheap shopping cart system that has the functionality for a water card set up. Amount of gallons someone has and how to recharge them, I know some of those shopping cart systems have something like a variable item of quantity. The only issue then would be my dad easily accessing this information, and deducting the gallons as they're used in the store.

If any of you have any thoughts/suggestions, I would really really appreciate it. Thanks for your time.

[guide] Optimizations in your C++ Program

Sat, 20 Feb 2010 22:29:14 -0600

Hi everyone.
This is my very first real article on anything programming related, so bare with me.

I'm not actually going to go too far into this, just going to bring up some small things that you can do to make your programs more efficient (sometimes slightly, other times greatly).

Now, for those compiling with gcc, it's a pretty smart compiler, and it will most likely assume what you're trying to do, and correct it accordingly. Though the same cannot be said about every compiler, or every version of every compiler; such as the while vs for (I'll get to that in a moment); I think microsoft and borland's compile it literally (optimization, aside).
So it's better to be safe than sorry, especially if you don't know if your code will compiler on various platforms.

while() vs for()
NOTE:My compiler ended up auto-optimizing this, so I couldn't create some assembly code to match my argument (though you can google it up, and you'll find some code generated on other compilers which support this argument), but I can generate some assembly that would look similar, if you want.

There can be abit of controversy when it comes to for vs while for an infinite loop.
In a literal compilation, while() and for() will not do the same thing.
Consider the following two:

Code:

while(1)

  std::cout<<"oh noes!"<
for(;;)

  std::cout<<"oh noes!"<


What the program may compile to in the first example (via pseudo-assembly) is:

cpp Code:
label:
Check if '1'
print "oh noes!" endline
jump to 'label'


Here's the second code:

cpp Code:
<empty statement -- do nothing>
label:
<empty statement -- do nothing>
print "oh noes!" endline
<empty statement -- do nothing>
jump to 'label'


Where you see "", that's actually where the two semi-colons for the for loop are.

So in short..

A while() loop, since it's conditional, will validate to make sure that the condition '1' is true, if so, it will continue the loop, whereas a for() statement would have no statement to condition, so it simply loops.

One fun thing I like to do in my code, is this:

cpp Code:
#define EVER ;;
for(EVER) //muhaha!




Postfix vs Prefix operators

One thing I see alot in code, is the use of 'var++' over '++var'.

Protip: Unless you need the value of 'var' before incrementing it, use pre-incremental.

You may think, "what harm can ++i do?".  Honestly, not that much, if it's just an integer.

See, when you use ++i, it increments i, and returns it's value.  When you're using i++, it stores the value of i, increments it, and returns it's old value.

In assembly, this is just switching of three commands, but what about with objects?

A great example I found here, is the following:

cpp Code:
class MyClass
{
public:
    MyClass& operator++()    //Prefix increment operator  (++x)
    {
        //Perform increment operation
        return *this;
    }
 
    MyClass  operator++(int) //Postfix increment operator (x++)
    {
        MyClass temp = *this;
        //Perform increment operation
        return temp;
    }
};


As you see, a new object is initialized, holding the old object, and returns that object.

This means that you're taking a performance shot to the stack every time you use postfix.



Dynamic allocation vs static (for lack of a better term) allocation.

First off, let me define dynamic and static:

cpp Code:
{
  //static:
  object obj;
  int foo;
  foo = 0;
  obj.setValue(foo);
}
{
  //dynamic:
  object *obj = new object;
  int *foo = new int;
  *foo = 0;
  obj->setValue(*foo);
  delete foo;
  delete obj;
}
//Please excuse my dereferencing if it's wrong.


Now, the primary difference between the dynamic and static, is that when you dynamically allocate data, it is saved in the heap; whereas if you statically allocate it, it is saved on the stack.

So for large objects, it's a good idea to dynamically allocate it, to preserve the stack.

Another benefit of dynamically allocated objects, is that they don't have to follow the law of scope, and in a sense, they never go out of scope until they are manually deleted.



Two-Dimensional Array Access

Access wouldn't be the word -- I should say traverse a two-dimensional array.

But the question is, which is better to use?

ary[x][y] or ary[y][x]?



It may not seem like there's any difference, but one is actually much slower than the other.

Here's an example of what the array foo[3][3] would look like in memory (don't question my epic pant skillz ;D):



Putting more into perspective, consider the two:

cpp Code:
for(x=0; x<3; ++x)
  for(y=0; y<3; ++y)
    foo[x][y] = 0;
for(x=0; x<3; ++x)
  for(y=0; y<3; ++y)
    foo[y][x] = 0;


Here is what the first five steps of the first for/for loop is doing in memory:

(the red dot represents where in memory the program is pointing to, and the arrow represents where that pointer is pointing to).



As you can see, it's using the pointer foo[0] to reach each location of memory that is next to it.

Next, we look at the second part:





As you can see, the second way, the data pointer is throwing itself all over memory, instead of moving though it one block at a time.

To prove it's length, here is a program I threw together (well, I modified it from the original C program my professor used last year) that creates an array that's 0xF00*0xF00 in size, and traverses it:

http://pastebin.com/f45636265

Here are the results for when I ran it:

x->y: 0.0500

y->x: 0.1000

As expected, the latter was about half as slow.



switch() vs else if

I generated some assembly output for this one (I generated assembly output for the rest as well, but those aren't worth looking at).

One thing I try to do, is use switch() as much as possible, as opposed to else if.

Consider the following two programs:

cpp Code:
int main()
{
  int myVar = 0;
  int foo;
  if(myVar == 1)
    ++myVar;
  else if(myVar == 2)
    --myVar;
  else if(myVar ==  3)
    myVar++;
  else if(myVar ==  4)
    myVar--;
  else if(myVar ==  5)
    --myVar;
  else if(myVar ==  6)
    myVar=0x3;
  else if(myVar ==  7)
    myVar=0xff;
  else if(myVar ==  <img src="http://lastcode.net/images/smilies/CrystalXP/glasses.gif" style="vertical-align: middle;" border="0" alt="Glasses" title="Glasses" />
    myVar=0xff;
  else if(myVar ==  9)
    myVar=0x2a;
  else if(myVar ==  10)
    myVar=0x801;
  else if(myVar ==  11)
    myVar=0x6;
  else if(myVar ==  12)
    myVar=0xfa;
  else if(myVar ==  13)
    myVar=0x0354;
  else if(myVar ==  14)
    myVar=myVar+8;
  else if(myVar ==  15)
    myVar=415;
  else
    myVar=0xF;
  foo = myVar;
return 0;
}


cpp Code:
int main()
{
  int myVar = 0;
  int foo;
  switch(myVar)
  {
  case 1:
      ++myVar;
      break;
    case 2:
    --myVar;
    break;
  case 3:
    myVar++;
    break;
  case 4:
    myVar--;
    break;
  case 5:
    --myVar;
    break;
  case 6:
    myVar=0x3;
    break;
  case 7:
    myVar=0xff;
    break;
  case 8:
    myVar=0xff;
    break;
  case 9:
    myVar=0x2a;
    break;
  case 10:
    myVar=0x801;
    break;
  case 11:
    myVar=0x6;
    break;
  case 12:
    myVar=0xfa;
    break;
  case 13:
    myVar=0x0354;
    break;
  case 14:
    myVar=myVar+8;
    break;
  case 15:
    myVar=415;
    break;
  }
  foo = myVar;
  return 0;
}


The brackets are left out, as it would make the else if too long.

Now, when this is compiled, the first program would end up compiling into a mess of cmp and jmp commands, one-by-one, working its way down.

If myVar happened to be 15, it would end up doing 15 conditionals.  The worst-case scenario here, would be that myVar isn't a number listed, and the result would be a test of every possible condition, until it reaches the else.



As for the switch(), that actually manages to be pretty efficient (well... VERY efficient).

This is because the second will compile into jump tables -- or hash maps.  This means that there is only one test done, and it knows which to jump to.

You can find the assembly output for the first program here: http://pastebin.com/f511c7140 and the second here: http://pastebin.com/f654d9226



We now have a situation though.  And that's the question of "what should we do with strings?".

I ran into this exact problem when I was writing my virtual machine, well, more specifically, my assembler.

In my first version, I was using a mess of else if and strcmp; here's a snippet of my parser, so you can see what I mean:

cpp Code:
char* first = new char[256];
  //Get first 4 characters
  while(line[j]!='\0' && line[j]!=' ' && line[j] != '\n' && line[j]!=';' && line[j]!='\r')
    first[j] = line[j++]; //Get instruction
  first[j++]='\0';
/*Flow*/
  if(!(strcmp("jmp", first)))
  {
    if(jumps(line, second, j, result, table, size))
    {
      result.dword = 1;
      return result;
    }
    else
      result.byte[3] = 0x20;//Set the highest byte
  }
  else if(!(strcmp("jz", first)))
  {
    if(jumps(line, second, j, result, table, size))
    {
      result.dword = 1;
      return result;
    }
    else
      result.byte[3] = 0x24;//Set the highest byte
  }
  else if(!(strcmp("jg", first)))
  //etc


This honestly was killing me!

The reason for this issue, is because a string (whether it's the std string, or a char*) is pretty much an array of characters, and you can't treat an array as one value.



My ultimate solution, was to hash the string, throw the solution keys into an enumeration, and run a switch() on the key -- it greatly improved the efficiency, and the num helped make things more sensible.



There are other solutions though; one, is to use nested switch() statements -- they have been used professionally before, but at the cost of making your wonderful code look horrible, also, the efficiency drops the more nested switches you use; here's an example that does something for the following strings: "jmp", "je", "cat"

cpp Code:
switch(str[0])
{
  case 'j':
    switch(str[1])
    {
      case 'm':
      //do stuff for jmp
      break;
      case 'e':
      //do stuff for je
      break;
    }
    break;
  case 'c':
    switch(str[1])
    {
      case 'a':
      //do stuff for "cat"
      break;
    }
    break;
}




The second, is to use the STL Container classes, and use the strings as the key, and just have the entries point to functions like in a hash table.



Other than that, I can't really think of any other ways of doing this.



Inline vs Non-Inline

Just a quickie on here.

If you call a certain function many times (esepcially a small function, such as void "addXY(int& x, int& y, int& z) {z = x+y;}", you may want to consider making it inline.

Use inline sparingly though, because it doesn't always result in faster code, and usually results in longer code



Preprocessors

Another controversy is whether to use #define (and macros) or use typedef (or functions).

Since the preprocessors are made during compile-time rather than run-time, they will be faster, but I personally use typedef over #define due to the fact I feel it's safer to work to work with type-safe objects then constants (you can fund much discussion on this subject).



Compiler Flags

For those using GCC, there are afew flags you should be aware of when compiling your code.

First, the -O (not -o) flag, you can choose the level of compiler-size optimization for your code:

-O0

This option turns all optimizations off.

-O1

This option will try to reduce the code's size and efficiency while trying not to slow down the compiling process.

-O2

Level two optimization will enable as just about all the compiler optimizations that shouldn't hurt your program.  This one may show down the actual time it will take to compile, but it's worth it.

I'd always recommend -O2

-O3

This turns all the optimizations on; there's a chance it will break your code though, I don't touch it.

-mtune and -march

mtune will tune your program to your CPU type, and march will generate assembly for your specific CPU (it will assume that mtune=march).

If you don't mind your program only working on your computer, you can use:

-march=native

Which will detect your CPU type, and act accordingly.

If you want your binary to execute on any machine (well, any x86), you can use -march=generic (or -mtune=generic); it will use all of those slow legacy instructions for the 8086 

You can see the rest of the -m optimizations here:

http://gcc.gnu.org/onlinedocs/gcc/i386-a...tions.html







Well, I hope this helped someone -- I enjoyed writing it atleat



[QT]I would like some help with cross compiling
Sat, 20 Feb 2010 19:42:12 -0600
I am using Ubuntu>Debian>Linux>Unix>Linus Torvalds? 

And am wondering if I can compile with my linux system for win32?

and if so how?

sorry if the question is trivial I am very new to using QT and any programming language that requires it to be compiled.


Hello Everyone
Sat, 20 Feb 2010 15:48:11 -0600
Hi,

I'm Admiral Refuge.



Mickcy sent me a link here abit ago, so I figured I'd join.



I know abit about C/C++ (and afew others), so I suppose a programming forum will be nice.


WEB HOSTS
Fri, 19 Feb 2010 15:16:53 -0600
Here's a list of several web hosts sorted into convenient categories.



FREE SHARED
Freehostia
000webhost
ByetHost
XtreemHost
MocaHost


SHARED
The Forum Forge
Freehostia
XtreemHost


VPS
Slicehost
Linode
PhotonVPS
VPS.NET
SoftLayer


DEDICATED
Steadfast
SoftLayer
ThePlanet


ALL OF ABOVE (except free)
ByetHost
HostGator
MocaHost




If you'd like me to add another host please post a link in this thread.


STARCRAFT 2 BETA
Wed, 17 Feb 2010 18:33:18 -0600
YES. IT IS OUT TODAY - AWESOME!



TORRENTING NOW! FOR THE WIN! SORRY FOR SHO-



Oh god rand out of breath. Sweet, going to see if I can play tonight.



http://starcraft.incgamers.com/blog/comm...-started1/


Just some suggestions.
Wed, 17 Feb 2010 00:02:50 -0600
I thought I'd throw some input out there.



First off, let me start by posting things I like


Well founded
Organized forum board
Clean forum and nice clean logo!




Now, I would like to throw some suggestions out there.


Getting Started Guides(A basic introduction guide for each section)
An Off Topic Section to replace the current General Section, then you can make General Section for Introductions and general topics related to the site(while not necessarily code related, and let General Discussions raise post counts
Some kind of Thank You Button or Karma System.  That really helps on sites like this, it helps set an impression on the time of feedback a user is going to give.
A more thorough Announcment of what this site is about(just go ahead and put rules inside that thread) Also could explain user groups and list all the administrators/moderators.
A section dedicated to projects, that don't necessarily have to relate to one side of code.  Something like "I'm starting such and such project who wants to be invovled"






Thanks for the time.  


O, hai :3
Tue, 16 Feb 2010 18:20:34 -0600
Hello, my name is Christian, I have been coding for about 3 years now (C, C++, PHP, some Python and learning Java atm), I'll be lurking arround seeking for help or trying to help others if posible.



I'm from Colombia, South America (Thats why my english is sucky), I was refered by Anman on the Tremulous forums (My nick there is Snake),  I hope to enjoy and learn a lot on these forums.



Thank you for reading and your attention.





:3 <3


Questions on Parsing
Thu, 11 Feb 2010 15:08:17 -0600
Does anyone have any useful information on parsing text, based on the contents of a .xml file? 


My Website
Sun, 07 Feb 2010 21:32:24 -0600
What do you think of it? http://marksiforum.com


BCS - Bad Code Spotter
Sat, 30 Jan 2010 20:33:09 -0600
Bad Code Spotter (BCS) is a lightweight utility written in CSS to spot deprecated tags and attributes in HTML 4.1.



Using it is simple, download the BCS file at the bottom of this post. Then put it in the same directory as your HTML file. Link the HTML file to BCS by putting this into the  section of the HTML document:


Code:


Save the HTML file and open it in Mozilla Firefox or Google Chrome.



Key

Red Background + Solid Red Border = Bad inline element

Solid Dashed Border = Bad block element

Solid Orange Border = Bad attribute on an element



After you fixed the problems don't forget to remove the code!



Live Demo: http://lastcode.net/dev/bcs (Credits to Mickcy for the terrible code)


Unknown Error
Mon, 25 Jan 2010 16:02:05 -0600
My GeSHi script keeps throwing me a specific error now (I don't need help on making it now, just figuring out this damn error). 



The error: "The page you are trying to view cannot be shown because it uses an invalid or unsupported form of compression."



The code:



PHP Code:

 $language_data = array (
 
        'LANG_NAME' => 'jass',
        'COMMENT_SINGLE' => array(1 => '//'),
        'COMMENT_MULTI' => array('/*' => '*/'),
        'CASE_KEYWORDS' => GESHI_CAPS_NO_CHANGE,
        'QUOTEMARKS' => array('"', '"'),
        'ESCAPE_CHAR' => '\\',
        
        'KEYWORDS' => array(
            1 => array (
                        'native', 'constant', 'exitwhen', 'takes', 'returns', 'private', 'public', 'keyword', 'static', 'delegate', 'initializer',
                        'uses', 'needs', 'requires', 'defaults', 'interface', 'endinterface', 'globals', 'endglobals', 'function', 'endfunction', 
                        'scope', 'endscope', 'library', 'library_once', 'endlibrary', 'struct', 'endstruct', 'method',
                        'endmethod', 'loop', 'endloop', 'if', 'then', 'else', 'elseif', 'endif', 'operator', 'readonly', 'stub',
                        'module', 'endmodule', 'implement', 'optional', 'true', 'false','null','super','this'
                        ),
                'STYLES' => array(
                'KEYWORDS' => array(
                    1 => 'color: #050505;'
                    )
                )
                )
            )
                
?>


Does anyone see anything that could cause this error?


[VB.net] Dope Wars
Thu, 21 Jan 2010 11:34:12 -0600
The original text based game is back, and being ported to VB.net by me!



Currently, the game has a database system of drugs & locations that is easily modified via text files, as well as a fully managed loan system, safehouse storage, buying/selling of drugs and more!



The source code (when the game is done) will be available to anyone out there who wants it, and is relatively small right now (1,263 lines when I last counted).



The port is also a console application for simplicity (and not having to deal with a GUI), and there are very few graphics (if any are there, there's only ASCII...and very little).



The source code will be posted later on when I get the game to a stable release.

LastCode.net - All Forums

[MyBB Plugin] HTML for Admins UPDATE!

Pushing an element into associative array?

Which Renders Faster?

Inactivity

Calculator

AnLib

PHPLiveX

Just some questions.

[guide] Optimizations in your C++ Program

[QT]I would like some help with cross compiling

Hello Everyone

WEB HOSTS

STARCRAFT 2 BETA

Just some suggestions.

O, hai :3

Questions on Parsing

My Website

BCS - Bad Code Spotter

Unknown Error

[VB.net] Dope Wars