Skip to main content

Patch notes 1.0.30

Features/Improvements/Changes:​

  • New core SDK classes: MeshObj and SimulationMeshInfoCollection
    • Simulation mesh related functions (plot, etc) moved to new MeshObj class.
  • Added support to not join layers with same name in join_models (becomes layer_name and layer_name_1,...)
  • Ability to set GPU count on a created simulation.
  • Simulation.estimate now returns a range of support gpu counts for simulation.
  • In Device list, order by Name and then Created.
  • For multi gpu tasks the SDK now only supports even number of GPUs (f.ex. 2, 4, 6, 8,...).
  • SimulationMeshInfoCollection object that can be used to access simulation meshes, as well as monitor the meshing process.
  • When using local mesh sizing the max allowed value of mesh_sizing_m is bound to the simulation crossover frequency with the formula: speed_of_sound / crossover_frequency / 1.585

Bug fixes:​

  • Fix bug where simulation.estimate would pick any mesh for a simulation that requires custom mesher settings.
  • Add SDK missing package dependency: meshio.
  • Fix issue where table columns would be cut of in jupyter notebooks.
  • Fix bug in plot_mesh where layer index ended up as float.
  • Remove source position validation check for boundary sources.
  • Fix geometry library id connection not working for meshes after Boundary Source changes.

Breaking changes​

  • Simulation GpuAllocationStrategy is now deprecated, by default the Utilization_optimized strategy is choose, but the user can now set the gpu count manually by using the simulation.set_gpu_count function.
  • simulation.get_mesh_info now returns a SimulationMeshInfoCollection object which contains all mesh objects connected to simulation.
  • simulation.plot_mesh moved to MeshObject.plot