search

Par4All / par4all

Par4All is an automatic parallelizing and optimizing compiler (workbench) for C and Fortran sequential programs
www.par4all.org
Other
51 stars 11 forks source link

Parallelization Issues #5

Open Premas opened 7 years ago

Premas commented 7 years ago

I tried to parallelize the below code using Par4all and I am able to get the parallelized code. But you can see that in the parallelized code, "n" is assigned with value 1410065407, instead of 1.0E10. Pls help me to solve the issue.

/ input code fragment /

define n 10000000000

for(i = 0 ; i < n ; i++)
      A[i] = i;
 for(i = 0 ; i < n ; i++)
      sum += A[i];

/Parallelized code/

pragma omp parallel for

for(i = 0; i <= 1410065407; i += 1) A[i] = i;

pragma omp parallel for reduction(+:sum)

for(i = 0; i <= 1410065407; i += 1) sum += A[i];

keryell commented 7 years ago

10000000000-2^33 1410065408

OK, so this code by itself is not really portable since you guess you have at least 34-bit integers to store this constant...

Unfortunately PIPS assumes 32-bit integers, so this constant... I hope there is a warning during compilation.

What about:

const long long int n = 10000000000LL;
for(i = 0 ; i < n ; i++)
A[i] = i;
for(i = 0 ; i < n ; i++)
sum += A[i];

?

Premas commented 7 years ago

Sorry for the delay. Thank you for your reply. No, I didn't get any user warning related to memory. Thank you, this method works. Instead of defining 'n' as a macro, I defined 'n' as variable and assigned '10000000000'. Now it works.

/input code / long long int n= 10000000000; for(i = 0 ; i < n ; i++) sum += A[i];

/parallelized code / long long int n= 10000000000;

pragma omp parallel for reduction(+:sum)

for(i = 0; i <= n-1; i += 1) sum += A[i];