[wsh]

Hi,
is it possible to check solving time in Fluent? I mean time beetwen clicking "calculate" button and "calculation complete" statement. Without using barbarian-style method like stopwatch, of course

[LuckyTran]

I don't know how to do it in serial mode but in parallel mode from the GUI you can click on Parallel>Timer>Usage and it will print the wall time and cpu time. In TUI you just use /parallel/timer/usage

[wsh]

Thanks Lucky,

i don't know much about using parallel mode - how does it affect my simulation? I am using Fluent to simple transient simulation and in serial mode after ~10 iterations solution is converged but in parallel mode Fluent is not stopping calculation even after 200 iterations... I am running it in 1 process - how much should I use?

Performance Timer for 450 iterations on 1 compute node
Average wall-clock time per iteration: 0.010 sec
Global reductions per iteration: 36 ops
Global reductions time per iteration: 0.000 sec (0.0%)
Message count per iteration: 4 messages
Data transfer per iteration: 0.000 MB
LE solves per iteration: 4 solves
LE wall-clock time per iteration: 0.004 sec (37.1%)
LE global solves per iteration: 0 solves
LE global wall-clock time per iteration: 0.000 sec (0.0%)
LE global matrix maximum size: 0
AMG cycles per iteration: 6.033 cycles
Relaxation sweeps per iteration: 157 sweeps
Relaxation exchanges per iteration: 0 exchanges
Time-step updates per iteration: 0.02 updates
Time-step wall-clock time per iteration: 0.000 sec (0.0%)

Total wall-clock time: 4.706 sec

I can't see here anything which is simillar to calculation time, "Total wall-clocl time" has too small value.

EDIT: I have tried command "solver-cpu-time" also, but Fluent is returning "invalid command".

[LuckyTran]

Total wall clock time should be what you are looking for.

There's almost no difference between serial or parallel if you are only running on 1 process, so I have no clue why you solution doesn't converge when running in parallel mode. Algorithmically there is no difference, except that the communication between the head node and compute node is now done via mpi rather than direct communication.

Mesh partitioning occurs if you run on more than 1 processes (by default it generates one partition per process). The partitioning causes some differences in discretization at the partition boundary, but the effect should be minor.

[miladghasemi]

The right format of this command is " (solver-cpu-time) ", you must type ().