High-performance computing (HPC) adds tremendous value to engineering simulation, providing high fidelity, accuracy and speed to simulations that offer insight into product performance. ANSYS Electronics HPC enables parallel processing for solution of the toughest and most challenging models — models with greater geometric detail, larger systems and more complex physics. Using ANSYS Electronics HPC to understand detailed product behavior gives you confidence in the design and helps to ensure that your product will be delivered first and succeed in the marketplace.
Adding an ANSYS Electronics HPC license to HFSS, SIwave, Q3D Extractor or Maxwell opens a world of bigger, faster and higher fidelity simulations. ANSYS goes well beyond simple hardware acceleration to deliver groundbreaking numerical solvers and HPC methodologies optimized for single multi-core machines and scalable to take advantage of full cluster power.
ANSYS Electronics HPC Packs
ANSYS Electronics HPC Packs are a flexible way to license parallel processing capability. For single users who want the ability to run simulations on a workstation, a single Electronics HPC Pack provides great value with increased throughput of up to eight times. For users with access to larger HPC resources, multiple HPC packs can be combined to enable parallel processing on hundreds or even thousands of processing cores. ANSYS HPC packs offer virtually unlimited parallel processing for the biggest jobs, combined with the ability to run multiple jobs using everyday parallel capacity.
ANSYS Electronics HPC Workgroup
ANSYS Electronics HPC Workgroup licenses enable you to use all the processing cores available for ANSYS applications. Each simulation can use any number of parallel processing instances enabled under the HPC Workgroup license. If there are multiple users running on workstations or users accessing a larger HPC resource for their most challenging simulations, HPC Workgroup licenses can be flexibly deployed where needed.
ANSYS Electronics HPC Enterprise
ANSYS Electronics HPC Enterprise licenses are designed to enable a distributed team to use the total number of processing cores that a company has available wherever that capacity is located. Each simulation can use any number of parallel processing instances enabled under the Electronics HPC Enterprise license. Wherever there are users, wherever there are HPC resources, the HPC Enterprise solution enables use of parallel processing.
Electromagnetic Solver Support
ANSYS Electronics HPC provides the parallel processing capability required to accelerate time to solution and solve the highest fidelity problems. The full portfolio of ANSYS electromagnetic solvers — including ANSYS HFSS, ANSYS Q3D Extractor, ANSYS SIwave and ANSYS Maxwell — use the same ANSYS Electronics HPC licenses to execute in parallel. Buy once, deploy once, and use parallel processing when and where it is needed.
ANSYS Electronics HPC takes advantage of multiple cores on a single computer to reduce solution time. Multithreading technology speeds up the initial mesh generation, direct and iterative matrix solves, and field recovery.
Spectral Decomposition Method
The majority of electromagnetic simulations require results such as field, far field and s-parameter data over a range of frequencies. The traditional method — running each frequency in sequence — can be time consuming. Spectral decomposition with ANSYS Electronics HPC distributes the multiple frequency solution in parallel over compute cores to accelerate frequency sweeps. You can use this method in tandem with multithreading. Multithreading speeds up extraction of each individual frequency point, while spectral decomposition performs many frequency points in parallel. The spectral decomposition method is scalable to large numbers of cores, offering significant computational acceleration.
Domain Decomposition Method
The domain decomposition method (DDM), available only for ANSYS HFSS, distributes a simulation across multiple, potentially networked cores to solve larger and more complex problems. This method is primarily designed to tackle larger problem size through the distributed memory nature of the solution. It can also be combined with multithreading and the spectral decomposition method to provide improvements in simulation throughput.
Finite Element Domain Decomposition
DDM generates a continuous finite element mesh over the entire structure, then subdivides that mesh and uses a distributed-memory parallel technique to distribute the solution for each mesh subdomain to a network of processors. This substantially increases simulation capacity. Domain decomposition is highly scalable to large numbers of processors and takes advantage of multithreading within the mesh subdomains to reduce solution times for individual subdomains.
Periodic Domain Decomposition
The periodic domain decomposition method available for HFSS uses a distributed-memory parallel technique for finite periodic geometries, such as antenna arrays. This method distributes unit cell mesh subdomains to a network of processors and RAM, while an industry-standard MPI maintains communications between domains. Simulation capacity and speed are substantially increased by re-using the adaptive mesh from a single unit cell for a large finite periodic structure, and processing the duplicated unit cells across a large number of processors. This method combines with multithreading to provide faster solves for each of the individual subdomains. The automated generation of domains makes this method easy to learn and implement.
Hybrid Domain Decomposition Method
The hybrid domain decomposition method uses the domain decomposition method on models consisting of finite element (FEM) and integral equation (IE) domains. The HFSS IE solver add-on makes it possible to create HFSS models that use a hybrid FEM–IE methodology to solve large EM problems. This methodology combines the virtues of two powerful techniques: the finite element method’s ability to handle complex geometries plus the method of moments’ (MoM) direct calculation of the free-space Green's function, which leads to accurate and efficient solutions for radiating and scattering problems. The hybrid domain decomposition method, when combined with electronics HPC, significantly increases simulation capacity by using the distributed memory parallel technique to distribute the matrix solution to a network of processors and memory.
Distributed Matrix Solver
The distributed matrix solver is a distributed memory parallel technique for HFSS and the HFSS integral equation (IE) solvers. With HFSS finite elements, the matrix solution is distributed across multiple cores or MPI-integrated computers to improve scalability and speed for direct matrix solver solutions. Analogously in HFSS-IE, which uses the method of moments (MoM) technique to solve for sources and currents on the surfaces of conducting and dielectric objects, the matrix solution can be similarly distributed. These distributed matrix solvers can be combined with multithreading and the spectral decomposition method to further increase simulation throughput.