Learning Forum

[Miller993]

I am currently running Ansys 2021 R1 on a work station equipped with a 12 Core – 24 thread AMD Ryzen 5900x CPU running @ 4.5 GHZ all core with 64 GB of DDR4 Ram at 3200 MHZ in Dual Channel and an M.2 SSD drive with near 3GB/sec Read & Write speeds . when I Start the solver with 8 Cores , A single iteration of the solution takes 9.7 seconds to finish , While when I Run with 12 cores it takes almost the same time if not slightly more . Given that all calculations are In-Core calculations ( No HDD or Page file is used ) , Why doesn't the time / iteration decrease when adding 4 real cores ? is it a problem with AMD systems compatibility with Ansys ? or is there something else wrong ?

The model is a Section of a Turbine can annular Combustor with Symmetric walls . Energy , Non-Premixed combustion models are used . Default K-Omega model is used with Default Coupled 2nd order Solver 

Everything in the fluent meshing watertight workflow related to mesh sizing is left as default except boundary layers are increased to 8 with Smooth transition method .   This geometry is just to benchmark my workstation performance, so I did not refine it but there are no issues with the quality

[Rob]

How many cells in the model? or can comment on the hardware side, but for parallel runs you need to ensure the data transfer doesn't outweigh the benefit of adding cpu. How does the speed compare for 4, 8, 10 & 12 cores? Do the comparison over 100 iterations and use the parallel timer from the top ribbon menu.