MeshGems Hex/Hybrid Meshing Technology Comes to MSC

Image courtesy of Distene.

MSC Software announced an extended partnership with Distene for their automatic mesh generation software, MeshGems. The meshes created by MeshGems are designed to work with various CAD and CAE applications, including FEA, CFD and electromagnetics.

The current partnership with MSC aims to add the automatic meshing of hexagonal (hex) and hybrid meshing in the MSC portfolio. MSC Actran will be one of the first to incorporate the hex technology to improve the speed of its automatic generated mesh.

Distene’s third-party technology targets CAE developers interested in automatic meshing. It includes a common API to support various other technologies. MeshGems is capable of performing surface meshes, mesh clean-up/modifications, mesh adaption and the aforementioned hex and hybrid meshing for volume meshing.

Recently, Distene announced the key features of the MeshGems 2.0 release, including boundary layer generation and improvements to parallel processing performance. With the growing use of HPC (high performance computing), GPU (graphics processing unit) computing and cloud computing, no doubt MSC is interested in the parallel processing capabilities.

In fact, according to Distene, their latest release can create a 1 billion element tetrahedral mesh 17.5 times faster when comparing one thread to 128 threads using HPC technology. As meshing takes considerable time, producing meshes 17.5 times faster should be quite appealing.

It would be interesting to see how MeshGems’ automatic meshing performance compares among HPC, GPU and cloud computing. Many simulation software organizations are starting to offer their customers cloud computing and GPU functionality to speed up their processes as they are cheaper than adding HPC infrastructure. GPUs have the advantage of being local to the host computer, and the cloud offers other benefits such as cooperative design and version control. It will be interesting to see how much focus there will be in HPC parallelization in the future compared to that of cloud and GPU.

How much time do you spend on your meshes? Would you use HPC to speed up the mesh time? Or would you prefer to speed up your operations using GPU or the cloud? Comment below.