Random number generator + multithreading problem

Random number generator + multithreading problem

Space cpp

Member #16,322

May 2016

Hello.
I moved my AI evaluation function to run on a separated thread and now all random number generator calls returns the same number.
On the main thread everything is fine.

My RNG function:

#SelectExpand
 1int iRand(int min, int max)
 2{
 3  if (min >= max) return min;
 4  int i = rand()%(max - min + 1);
 5  return min + i;  
 6}

----------
My games

Mark Oates

Member #1,146

March 2001

Are you initializing the random numbers with srand?

--
Visit CLUBCATT.com for cat shirts, cat mugs, puzzles, art and more <-- coupon code ALLEGRO4LIFE at checkout and get $3 off any order of 3 or more items!

AllegroFlare • AllegroFlare Docs • AllegroFlare GitHub

bamccaig

Member #7,536

July 2006

Perhaps you need to reseed it on each thread?

void srand(unsigned);

Or seed it at all.

Most people pass time as the seed value so it changes each time you run the program.

srand(time(NULL));

--
I mean the best with what I say. It doesn't always sound that way.

Space cpp

Member #16,322

May 2016

I was calling srand(time(NULL)) at the beggining of the program. Calling it on the thread function solved the problem.

But... the program crashed on exit.
Do I have to use a mutex to ensure the safety of the srand/rand calls? I'm still kinda new to multithreaded programming.

----------
My games

amarillion

Member #940

January 2001

Hard to know what caused the crash without seeing the code. Can you share some?

Multithreaded programming is complex, and the question you should always ask is, do I really need it? You might be able to do without, and it will simplify things a lot.

--
Martijn van Iersel | My Blog | Sin & Cos | Food Chain Farm | TINS 2023 is Jun 2-5

Audric

Member #907

January 2001

Even if it's not strictly necessary in this case, it can still be a good occasion to learn/experiment.

In this case, I have a hard time imagining why having called rand() / srand() would lead to a crash later.
I would rather advise you double-check every freeing of resource that your child thread might still be using.

For example, if your AI thread uses any allegro function / data, the thread should already be stopped before allegro is shut down.
Note that even if you don't call al_uninstall_system(void) yourself, it will be called after you reach the end of main() (AFAIK it is using the atexit() system to do so)

In general, killing the thread should be one of the first things to do when you stop the program.

Space cpp

Member #16,322

May 2016

amarillion said:

Hard to know what caused the crash without seeing the code. Can you share some?

Sorry, the code is way too huge to post.

amarillion said:

the question you should always ask is, do I really need it? You might be able to do without, and it will simplify things a lot.

My game runs fine without it, but on really large maps the A.I. tends to take almost 1 second to calculate every move, making the whole game appears frozen. That's why I think using a separated thread for it became necessary.

Audric said:

In general, killing the thread should be one of the first things to do when you stop the program.

I call al_destroy_thread on my leave game routine, so it doesn't seem to be the case.

I'm testing the game again and not getting any chashes, it must have been a coincidence that something went wrong right after including the new srand call.

----------
My games