Hello,
For a transient problem, {x}t={b}t ,the RHS {b}t is time-varying and depends on the previous {b}t-dt, but is constant. So during a complete analysis, we need to do factorization only once, and use this fac...
In this situation, multi-rhs is not suitable, and I tried to call the complete Pardiso (phase=11, 22, 33 and not execute phase=-1) at the 1st time step, then only executed phase=33 at later time steps. This way did not work however.
Do you have any suggestion?
I know there is a similar question, but the answers below don't seem to work either.
https://software.intel.com/en-us/forums/intel-math-kernel-library/topic/297075#
Thx!
Link Copied
Hello
As i understand, the Pardiso phase are designed for the cases , call phase 12 one time , then call phase =33 with different RHS in loop. Do you have test case to show the problem?
Best Regards,
Ying
For more complete information about compiler optimizations, see our Optimization Notice.