Why does inlet/outlet take a significantly longer time to simulate than piston?
Hi,
I have recently observed this in my simulations. I was running two irregular wave simulations, one with the piston and the other in inlet/outlet.
The counter-intuitive thing that I observed is that Inlet/Outlet is significantly slower than the piston case even when it has 30% lower number of particles compared to the piston case
Inlet/Outlet: 11.3 million
Piston: 15.97 million
Both have the same particle spacing (dp = 0.01m) and run on the same GPU (NVidia V100)
Here I have plotted the Time\sec from both the simulations. As you can see, the Piston case takes less than half the time to simulate, even with higher particles.
I was of the impression that Inlet/Outlet saves time. But this is contrary to that. Is there something incorrect that I am doing?
Comments
Inlet/outlet creates new particles and removes others... this has a significant impact on performance and memory management.
In addition, the new properties of the inflow particles like for example density should be extrapolated from vicinity so that SPH kernels and extra interactions are taking place... the same applies for some options of inlet and outlet (densty, velocity) so it depends on the options you are using the case can be even slower
In the case of the piston... particles are always the same.. and you only are adding a motion to a piston....
Regards