This poster presents a parallel implementation of the Hybrid Total FETI (HTFETI) method. HTFETI i... more This poster presents a parallel implementation of the Hybrid Total FETI (HTFETI) method. HTFETI is a multilevel domain decomposition method in which a small number of neighboring subdomains is aggregated into clusters. The presented HTFETI solver is able to solve 3D heat transfer problems of size up to 124 billion unknowns on 17,576 nodes of the Titan supercomputer. A superlinear strong scalability of the iterative solver is presented for 20 billion unknown heat transfer problems and 11 billion DOF structural mechanics problem executed on 2,744 to 17,576 nodes. In addition, a super-linear strong scaling is presented for 300 million DOF real world problem. The acceleration of the HTFETI method using Local Schur Complement method delivers 7.8 speedup (heat transfer) and 2.7 speedup (structural mechanics) of the iterative solver when executed on the Intel Xeon Phi and 3.4 speedup for structural mechanics delivered by the GPUs on the Titan machine.
This paper presents an algorithm for highly-parallel loading and processing of unstructured mesh ... more This paper presents an algorithm for highly-parallel loading and processing of unstructured mesh databases in a distributed memory environment of large HPC clusters without collecting data into a single process. The algorithm is proved effective, having linear speedup in the large dataset limit. Demonstrated on Ansys CDB, EnSight, VTK Legacy, and XDMF databases, we show that it is possible to efficiently reconstruct meshes with 800 million nodes and 500 million elements in several seconds on thousands of processors, even from databases that were not designed to be read in parallel. The algorithm is implemented in our MESIO library that can be used as (i) an efficient parallel loader (e.g. for numerical physical solvers) or as (ii) a high performing parallel converter between mesh databases.
This poster presents a parallel implementation of the Hybrid Total FETI (HTFETI) method. HTFETI i... more This poster presents a parallel implementation of the Hybrid Total FETI (HTFETI) method. HTFETI is a multilevel domain decomposition method in which a small number of neighboring subdomains is aggregated into clusters. The presented HTFETI solver is able to solve 3D heat transfer problems of size up to 124 billion unknowns on 17,576 nodes of the Titan supercomputer. A superlinear strong scalability of the iterative solver is presented for 20 billion unknown heat transfer problems and 11 billion DOF structural mechanics problem executed on 2,744 to 17,576 nodes. In addition, a super-linear strong scaling is presented for 300 million DOF real world problem. The acceleration of the HTFETI method using Local Schur Complement method delivers 7.8 speedup (heat transfer) and 2.7 speedup (structural mechanics) of the iterative solver when executed on the Intel Xeon Phi and 3.4 speedup for structural mechanics delivered by the GPUs on the Titan machine.
This paper presents an algorithm for highly-parallel loading and processing of unstructured mesh ... more This paper presents an algorithm for highly-parallel loading and processing of unstructured mesh databases in a distributed memory environment of large HPC clusters without collecting data into a single process. The algorithm is proved effective, having linear speedup in the large dataset limit. Demonstrated on Ansys CDB, EnSight, VTK Legacy, and XDMF databases, we show that it is possible to efficiently reconstruct meshes with 800 million nodes and 500 million elements in several seconds on thousands of processors, even from databases that were not designed to be read in parallel. The algorithm is implemented in our MESIO library that can be used as (i) an efficient parallel loader (e.g. for numerical physical solvers) or as (ii) a high performing parallel converter between mesh databases.
Uploads
Papers by Ondřej Meca