Benchmarks for comparison in multiple HPC applications

At the annual Supercomputing Conference (SC), there has always been an unofficial theme that draws attention. In recent years, the focus has shifted toward Machine Learning and Deep Learning, following a period of interest in data-intensive computing and the potential of cloud computing to reshape the future of supercomputing. These topics, while diverse, share a common thread: they are not centered on traditional CPU processors. Instead, they revolve around performance improvements and ecosystem development under the X86 architecture. However, at this year’s SC17, the conversation once again turned to the core processor. This time, it was the ARM-based hardware and software ecosystem that took center stage. Cray demonstrated comprehensive system integration, showcasing benchmarks that could rival Intel’s most advanced offerings. The ARM-based "Isambard" supercomputer, set to be deployed at the University of Bristol next year, features 10,000 cores powered by Cavium’s ThunderX2 ARM processors. It represents significant progress in ARM-based HPC systems, with systematic research and development efforts ongoing for years. One of the most well-known ARM-based HPC projects is the Mont Blanc system at the Barcelona Supercomputing Center, which initially used dual Cortex-A15 ARM chips and now runs on the ThunderX2. McIntosh-Smith and his team have published compelling benchmark results comparing a Cray 8-node cluster with a 32-core ThunderX2 against Intel’s Skylake and Broadwell platforms across multiple HPC applications. Overall, the results show that memory-intensive applications perform exceptionally well on ThunderX2, often outperforming Skylake. For floating-point intensive workloads, Skylake still holds an edge due to its wider vector units, but ThunderX2 performs comparably to Broadwell. With increased high-bandwidth memory, the performance gap may further narrow—something worth watching closely. OpenFOAM, an open-source CFD application widely used in scientific and industrial HPC, clearly demonstrates the memory bandwidth advantage of ThunderX2. Similarly, weather and climate simulations, such as those using Nemo, also benefit from improved memory performance. While compute-intensive applications like GROMACS, CP2K, and VASP show smaller performance gaps between processors, memory-bandwidth-heavy tasks reveal more distinct differences. Cray’s XC50 system, now featuring ARM-based components, promises even more real-world performance data at future SC conferences. The Isambard project architecture highlights how Cray has integrated ThunderX2 with Aries interconnect technology, along with full Cray software support—showcasing a strong commitment to ARM in the HPC space. McIntosh-Smith believes that future ARM processors will match or exceed the vector capabilities of other CPU manufacturers, bringing them closer to parity in HPC performance. These results, achieved with minimal optimization and just a few hours of fine-tuning, reflect years of dedicated work on ARM-based HPC systems. As hardware matures, the software ecosystem for ARM in HPC will continue to evolve rapidly. With the official launch of ARM processors in supercomputers, the industry is entering a new era of architectural transformation. The HPC field is opening up to a wider range of processor choices, and the ARM architecture is poised to play a major role in shaping the future of high-performance computing.

EI 28 Transformer

Ei28 Transformer,15V Smd Transformer,28 Volt Transformer,28V Transformer,variac transformer

Guang Er Zhong(Zhaoqing)Electronics Co., Ltd , https://www.geztransformer.com