You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the basic MPI example (Send/Recv), we use the same seed to initialize random number generators (here) on all processes. It will cause the random number sequences generated on all MPI processes to be the same with each others, won't it?
If it will, then that means we are not using more random points to improve the accuracy when we use more MPI processes. However, the tutorial says:
This is a more accurate, because each node is running as many iterations as the singe node in the serial code ran.
which may not be true in this case, because each MPI process generates exactly the same set of random points.
The text was updated successfully, but these errors were encountered:
In the basic MPI example (Send/Recv), we use the same seed to initialize random number generators (here) on all processes. It will cause the random number sequences generated on all MPI processes to be the same with each others, won't it?
If it will, then that means we are not using more random points to improve the accuracy when we use more MPI processes. However, the tutorial says:
which may not be true in this case, because each MPI process generates exactly the same set of random points.
The text was updated successfully, but these errors were encountered: