Hello guys
I want to count the time needed for the factorization of a 1000 x 1000 matrix with the parallel routine PDGEQRF... so I use the following
T1=MPI_WTIME()
CALL PDGEQRF(.....)
T2=MPI_WTIME()
and then I call the T2-T1.. however it returns 0, while if I call only the T2 it returns a satisfying result... the problem is that T2 includes the time for all the preparation , the matrix distribution over the process grid etc etc..
I want to count only the time for the factorization of the routine and avoid the time for distributing the 1000 x 1000 matrix.. any advises ?
THANKS

