A Balancing Act
Computer hardware continues to evolve at a rapid pace, leading to faster cores, smaller parts, and a huge boost in the availability of computational resources. In the world of CFD, high-performance computing (HPC) offers some significant advantages. You can run larger, more complex cases, which improve the accuracy of your results and provide deeper insight into your system. HPC also drives productivity by reducing simulation runtime and increasing throughput.
To take advantage of these benefits, software must keep pace with improvements in hardware. At Convergent Science, we hug the cusp of innovative technologies while ensuring the reliability and stability of our CONVERGE CFD software.
Leveraging the Cutting Edge
Convergent Science employs a team of HPC experts dedicated to improving the performance of CONVERGE on the latest computing hardware. In addition, we collaborate with several national laboratories and industry partners, including Argonne National Laboratory, Hewlett Packard Enterprise, Intel, Cray, and NVIDIA, to ensure CONVERGE scales well, to implement support for various MPI packages and CPU architectures, and to test CONVERGE on different HPC systems.
CONVERGE 3.0, the latest major release of our software, demonstrates the success of these efforts. With innovative strategies to make load balancing as efficient as possible, CONVERGE 3.0 scales linearly on thousands of cores.
It’s clear the CFD industry is moving toward a many-core architecture, and simulations are moving from a handful of cores to tens of thousands of cores. For clients who do not have the architecture to maintain and run a cluster themselves, we actively test and partner with cloud service providers. Cloud computing platforms offer affordable access to advanced computational resources that can be tailored to meet your simulation needs. CONVERGE is compatible with all major cloud providers for running highly parallel CFD simulations.
Increased Capability, Increased Capacity
The greater availability of computational resources allows you to run larger simulations and capture more complex phenomena, which offers benefits across the spectrum of application areas. For example, gas turbine simulations often contain hundreds of millions of cells. Running these cases on a low number of cores translates to extremely long runtimes, which are often unacceptable for fast-paced industry timelines. Using thousands of cores, however, you can obtain results in a reasonable amount of time, even for massive cases.
For internal combustion engines, with sufficient computational resources, you can run multiple engine cycles in a day for quick turnaround on geometry optimization, emissions studies, cyclic variability analysis, and other engine simulations. Running CONVERGE 3.0 on hundreds of cores, you can simulate more than 12 engine cycles in a single day!
|Cores||Time (h)||Speedup||Efficiency||Cells per core||Engine cycles per day|
Some applications require you to use a very fine mesh to, for example, resolve flame structures with large eddy simulations (LES) or resolve small gaps in pumps and compressors, which means a large overall cell count. Pairing CONVERGE 3.0 with the power of HPC accelerates these simulations, so that you don’t have to sacrifice accuracy for speed.