Problem declaring large arrays

Hi,

I declare

Code:

---

        float found[100000][2];        // Stores found.txt
        float actual[10000][2];        // Stores actual.txt
        float mov[1000000][3];                // Stores moving.txt
        int mapCount[100];                // Stores mapCount.txt

---

which i plan to store a lot of numbers from text files and the program always crashes.

I've narrowed it down the the large declaration because program works fine with smaller arrays.

I would have thought the computer would have enough memory to be able to do this? Can anyone shed any light on why this can't be done or some alternatives? (I'm already thinking I'm going to have to read data in iteratively but it would be nice to have it all there all at once.

Thanks

Matt.

Matt, check your compiler help. It should give you info on if, how, and how much, may be possible.

I agree with you that you'll probably have to work with this data in "chunks", smaller than the BIG arrays you were hoping for. And don't forget what your OS can do for you, as well. I was amazed to discover that Windows2000 has special resources available for system sorting - and it leaves my best Quicksort, in the dust for that reason. (and has no such limitations on size like I would have, since it will automatically can use disk drive space as a stand-in for RAM.

Very sweet. :)

Adak

You're looking at over 12mb of static memory there if you assume a 32-bit float, so I wouldn't be surprised if you're getting a stack overflow. With that much data, I'd suspect you won't need to access it all at once so your easiest bet would be just to write what you don't immediately need to some sort of swap file and put it into memory dynamically when you need it and delete that memory when you don't.

As Adak said, your OS might use it's own swap file for your program, but it's not likely to do that if it's all static memory. Try allocating the memory dynamically.

They're probably using Turbo C++ 3 anyway. :rolleyes:


Quzah.

Assuming that you actually have a 32-bit compiler, and these are local variables in say main, then

Code:

---

static float found[100000][2];

---

should get you out of the immediate hole.

Why in the world do you need an array that size? It would take eons to fill it up regardless of the algo employed.