ST. PAUL, Minn., Feb. 23, 2021 — Hyperion Research today announced that the new, sixth edition of its worldwide study of high performance computing (HPC) buyers and users is now available for purchase.
The study details the resources, practices and purchasing intent of a large, representative group of 194 government, academic and industrial HPC data centers in 26 countries that together host nearly 2,000 HPC systems (28% of the centers manage more than 15 systems each). As always, Hyperion Research placed special emphasis on the private sector, to enable drilling down into specific industries.
“Our latest study confirms that HPC centers plan to expand their resources and adjust priorities and practices, to perform simulations at larger scale and exploit AI, cloud computing, massive data storage and the promise of quantum computing,” said Hyperion Research CEO Dr. Earl Joseph. “A growing majority of these centers are already using AI and cloud computing today.”
According to Steve Conway, senior advisor for HPC market dynamics, “HPC is in a high-innovation period worldwide, especially regarding processing elements, methodologies and the environments where HPC happens. The new study shows that what used to happen almost exclusively on premises is increasingly extending out to clouds, edge locations and the enterprise world.”
The extensive study is divided into seven topical reports that can be purchased separately or as a set:
- Trends in HPC servers, processors and accelerators/coprocessors
- HPC verticals and application workload areas
- System software and middleware
- HPC storage and interconnects
- Using public clouds for HPC workloads
- AI and other HPDA (high performance data analysis)
- Evolving HPC budgets for hardware, software, storage, and clouds
Sample Findings
- 51% of the surveyed end-user applications are still run on a single node.
- 75% of the sites now employ accelerators/coprocessors. NVIDIA GPUs have a sizeable lead in this category, but AMD is gaining market share and a growing number of the HPC sites (18%) exploit FPGAs.
- C++ and Python are the leading parallel programming languages, but nearly half of the sites still use Fortran.
- In our 2017 edition of this study, only 4% of the HPC sites ran more than half their HPC workloads in public/external clouds; in the new study, that figure grew to 12%.
- Data locality surpassed data security as the most-named barrier to exploiting cloud computing.
- In the 2017 study, peak performance for the sites’ largest HPC systems averaged 3.9PF; in the new study, the average almost quadrupled to 15.4PF.
- Infiniband continues to be the leading storage system backbone protocol, but Ethernet is rising quickly.
These reports are designed to help HPC vendors and users track best practices and benchmark HPC installations based on real-world usage and trends. HPC OEMs and other vendors will better understand the changing market dynamics in order to better position their products, assist in designing future products, and size the potential for future sales.
Click here for more information and view report summaries, or to discuss further with a Hyperion Research analyst: https://hyperionresearch.com/hpc-technology-services/hpc-mcs-study-2020/
About Hyperion Research, LLC
Hyperion Research provides data-driven research, analysis and recommendations for technologies, applications, and markets in high performance computing and emerging technology areas to help organizations worldwide make effective decisions and seize growth opportunities. Research includes market sizing and forecasting, share tracking, segmentation, technology and related trend analysis, and both user & vendor analysis for multi-user technical server technology used for HPC and HPDA (high performance data analysis). We provide thought leadership and practical guidance for users, vendors and other members of the HPC community by focusing on key market and technology trends across government, industry, commerce, and academia.
Source: Hyperion Research, LLC