Good for multi-thread program?

[dekeenfrance]

Hi,

I am running a heavy simulation which takes time, its main structure looks like this:

for(int i=0; i<1000; i++) { // repeat what's inside 1000 times -- A

   for(int j=0; j<2000; j++) { // sequential computation from 1 to 2000 --B
      // create and run a big object of vectors: myV
      //......
   }

   // some similar loops running on the vector created in part <B>
   for(auto iter = myV.being(); iter != myV.end(); iter++) { // --C
      // ......
   }
   for(auto iter = myV.being(); iter != myV.end(); iter++) { // --D
      // ......
   }

   // ......
   for(auto iter = myV.being(); iter != myV.end(); iter++) { // --K
      // ......
   }

}

Should I use multi-thread programming to speed up? for part <A>? or maybe for part <C> to <K>?
Could someone expert in this area give me some opinion, please?

Thanks in advance.

[majidkamali1370]

If each for loop's data is independent of others, you can put each to a function and then call each function in a thread. If loops need to execute in series, I don't think you have any chances.
In threads you need to care about data races, shared data, thread finish etc.

Good Luck

[dekeenfrance]

Thanks for your opinion. Any other comments please?

[jim80y]

Are your loops C .. K going to be altering the content of myV?

If not then you'll be fine with threads doing the work.

Jim

[zb0S216C]

Do each of the loops need to run in sync with each other?
Is there any communication between the loops?
Do each of the loops have a specified time-frame in which they need to finish?
Does one loop have a priority over another?
Do your loops access a common resource?
Should each thread execute two or more of the loops rather than one?
Should each thread remain active after it has finished?

_Wazzak