Memory Allocation Limitations on Windows for Large Arrays in C++

In summary, the conversation discusses a runtime error that occurs when declaring an array of size 24 mb in C++, and how it is related to the location of the declaration and the operating system being used. The conversation also includes a discussion about the limitations of memory allocation in Windows and the use of pointer arithmetic for faster processing.
  • #1
sid_galt
502
1
When I declare an array of size 24 mb in c++, I have discovered that the result is a runtime error if the declaration is made inside a function. If the declaration is made outside a function, then there is no problem.

The problem exists with both VC++ Express and Cygwin GCC (GCC gives a segmentation fault). There is no problem on Suse Linux.

Does windows have a limitation on the amount of memory that can be allocated using functions or something?

I have 1 GB Ram of which atleast 350 MB is free at all times and 3.2 GHz 2MB L2 cache Pentium IV.
 
Technology news on Phys.org
  • #2
It might help if you show some code.
Do you deallocate the memory when you exit the function?
 
  • #3
robphy said:
It might help if you show some code.
Do you deallocate the memory when you exit the function?

Actually the array declared is a static array. Here's the relevant code which generates error.
<code>
void LoadHistory(bool k)
{
//Declare and initialize variables
int matchUser[3000000][2] //The erring variable

fstream fbinary, fmovieratingcount;
fbinary.open(BINARY_FILE_PATH, ios::in | ios::binary);
fmovieratingcount.open(MOVIE_RATING_COUNT, ios::in | ios::binary);

if(!fbinary && !fmovieratingcount)
{
cout<<"FILE CAN'T BE OPENED!\a\n";
exit(1);
}

int tmp, movieId;

for(int i = 0; i < 3000000; i++)
{
matchUser[0] = 0;
}

//While loop to read movie ratings
while(!fmovieratingcount.eof() && !fbinary.eof() && net_movie_no < NO_MOVIES)
{
int custId, ratings_no;
int rating;

fmovieratingcount.read((char*) &tmp, sizeof(tmp));
fbinary.read((char*) &movieId, sizeof(movieId));
}

}
</code>

Moving matchUser declaration outside the function results in no error.

<code>
int matchUser[3000000][2] //No error now
void LoadHistory(bool k)
{
//Declare and initialize variables


fstream fbinary, fmovieratingcount;
fbinary.open(BINARY_FILE_PATH, ios::in | ios::binary);
fmovieratingcount.open(MOVIE_RATING_COUNT, ios::in | ios::binary);

if(!fbinary && !fmovieratingcount)
{
cout<<"FILE CAN'T BE OPENED!\a\n";
exit(1);
}

int tmp, movieId;

for(int i = 0; i < 3000000; i++)
{
matchUser[0] = 0;
}

//While loop to read movie ratings
while(!fmovieratingcount.eof() && !fbinary.eof() && net_movie_no < NO_MOVIES)
{
int custId, ratings_no;
int rating;

fmovieratingcount.read((char*) &tmp, sizeof(tmp));
fbinary.read((char*) &movieId, sizeof(movieId));
}

}
</code>
 
  • #5
Looks like you are trying to allocate a big array on the stack.
Stacks have finite sizes in C, usually you can increase the size but it really does not make any sense to create stacks that are that big.

By the way, what are you trying to do with this program? There are exceptions, so that's why I ask, but usually it is completely unnecessary and generally a bad idea to load something into an array that big.
 
Last edited:
  • #6
Thanks robphy.

MeJennifer said:
By the way, what are you trying to do with this program? There are exceptions, so that's why I ask, but usually it is completely unnecessary and generally a bad idea to load something into an array that big.

Playing around with this http://www.netflixprize.com

The 24 mb array is for quick mapping. This is actually a very small array. The actual data requires an array of size 200-300 mb minimum.
 
  • #7
sid_galt said:
Playing around with this http://www.netflixprize.com

The 24 mb array is for quick mapping. This is actually a very small array. The actual data requires an array of size 200-300 mb minimum.
I see! :smile:

If you want to be fast then don't use array indexing but, wherever possible, use pointer arithmetic.

For instance, in your array initialization routine, you require the program each time to calculate the address instead of just incrementing some pointer.

Keep that in mind when you run your statistical analysis, you could easily improve the speed of such a program by 50% if you minimize the use of array indexing and use pointer arithmetic.
 
  • #8
Thanks, I'll keep that in mind.
 

What is memory allocation on windows?

Memory allocation on windows refers to the process of assigning and managing portions of a computer's memory for programs and processes to use. It is a crucial aspect of computer science and is essential for the efficient functioning of a computer.

What are the different types of memory allocation on windows?

There are various types of memory allocation techniques on windows, including static allocation, dynamic allocation, stack-based allocation, heap-based allocation, and virtual memory allocation. Each technique has its advantages and disadvantages and is used for different purposes.

How does memory allocation work on windows?

The memory allocation process on windows involves several steps, including requesting memory space, reserving the space, and assigning it to the program or process. The operating system manages this process and ensures that each program or process has the required memory to run.

What are the common issues with memory allocation on windows?

Some common issues with memory allocation on windows include memory leaks, fragmentation, and out-of-memory errors. These issues can lead to system crashes, slow performance, and other problems. It is essential to understand these issues and how to prevent them for efficient memory allocation.

How can memory allocation be optimized on windows?

To optimize memory allocation on windows, one can use techniques such as pre-allocation, memory pooling, and recycling memory. It is also crucial to monitor and manage the memory usage of programs and processes regularly to prevent issues such as memory leaks and fragmentation.

Similar threads

  • Programming and Computer Science
Replies
17
Views
1K
  • Programming and Computer Science
Replies
4
Views
3K
  • Engineering and Comp Sci Homework Help
Replies
17
Views
1K
  • Programming and Computer Science
Replies
10
Views
6K
  • Programming and Computer Science
Replies
13
Views
7K
  • Engineering and Comp Sci Homework Help
Replies
7
Views
9K
  • Programming and Computer Science
Replies
5
Views
3K
  • MATLAB, Maple, Mathematica, LaTeX
Replies
4
Views
11K
  • Engineering and Comp Sci Homework Help
Replies
2
Views
2K
Back
Top