Code dumps core, why?

Question

I've created code which splits into several threads, when I try to join them together, code returns that it dumped core ( means it tried to access nonallocated memory )

code: https://code.sololearn.com/ca2a7A23a9A1

any nuber bigger than 99 causes error

Answer 1

If it is working with lower numbers then it may be that you are exceeding the memory limit on Sololearn. You can try your code with an offline IDE to confirm if it’s Sololearn specific.

Answer 2

the error caused by your code was overflow. Did you write the code yourself? anyway, you could try this solution (I adjusted your code):
https://code.sololearn.com/cA7a0a7a20A1

N: B. I tested it on MinGW with C11.

Output on my machine:
r = 1000
PI = 3.138552000000000
Without threads: 0.004000
PI = 3.138552000000000
With threads: 0.003000

Answer 3

MinGW is for windows, isn't it? by the way, no compiler is able to find atomic

Answer 4

i said C11, compile it with C11, it should work on GCC (Linux as well)

Answer 5

when i pasted in your code that you posted above on my ide and it exited with an overflow error. the data type in your loop was int, and I think your sq() is not doing what it should be doing. I used sqrtl() from math.h and long long as data types. In addition, your thread arg struct is messy. and no need to create an array of thread args, you could use just one.

also compare your "for" loop in with thread and without thread parts of your solution.

Answer 6

and run your code and see if you get the same or almost the same result as mine, cause I didn't when I ran your code.

Answer 7

i think you should study a bit more threading in C. anyway, I think it's enough for this thread. Good luck!

Answer 8

i don't need to wait for the result, that's why my thread func returns 0 and my resa and resb are atomic,

pthread_join() means wait for the thread to finish.

Answer 9

if you want to use another loop, just do it, still, it won't fix your problem.

Answer 10

this will fix my problem, got it.

for (int i = 0; i < NUM_OF_THREADS; i++)
{
pthread_create(threads + i, NULL, threadFunction, (void*)(&args));
}
for (int i = 0; i < NUM_OF_THREADS; i++)
{
pthread_join(threads[i], NULL);
}

anyway, if my solution seems useless to you better wait for other responses.

Answer 11

no, that doesn't help, it's reason why I have tried it on sololearn, both ends same at input > 99

Answer 12

did you run it on your machine? and do you understand threading?

Answer 13

my bad, it will run on sololearn, try now

Answer 14

threading should be something like "splitting" program into multiple processes - not useful on low core cpu, but with 4 will be maximally used with at least 4 threads... so my splitting code into squere of input is not really good... but there should be no problem with it

so I gues I understand threading enough to know how to use it, by the way, I had lot of issues when I first time were learning threading in c++ when I were accessing memory at same location because sometimes some threads were "ignored" (they tried to overwrite same place in memory)

Answer 15

your version works fine until 10,000

Answer 16

btw, in these kinds of calculations, threading may not be a faster option.

Answer 17

yeah, it's slower, I setted my cores to static, it works but this is output:
r = 10000
PI = 3.141191
Without threads: 0.002468
PI = 3.100882
With threads: 0.003102

Answer 18

and here is mine:

r = 10000
PI = 3.141292960000000
Without threads: 0.006000
PI = 3.141292960000000
With threads: 0.006000

Process returned 0 (0x0) execution time : 3.180 s
Press any key to continue.

Answer 19

and it's propable that sololearn's compiler is not good for threading because sometimes it works, sometimes returns weird stuff (something around correct result or really much lower) and other times it just dumps core - I'm not sure if threading should be used

Answer 20

sololearn didn't invent any compiler btw, it uses GCC on its Linux server. and in your code you had some errors, check them out.