Machine Learning in HPC: Workloads, Frameworks, Data Types, Configurations, Cloud Usage, Business Outcomes

September 29, 2023|Addison Snell, Dan Olds, Steve Conway

EXECUTIVE SUMMARY

Intersect360 Research studied machine learning (ML) initiatives across HPC-using organizations worldwide. This study of 152 qualified respondents in academic, commercial and government organizations was conducted from Q4 2022 to Q1 2023. Our goal was to capture the dynamics of ML initiatives within organizations that employ HPC, with ML treated as a broad category that includes deep learning and other AI methods. The study analyzes a wide spectrum of factors: budgets, workflows, data types and resources, configurations, cloud usage, and organizational/business outcomes.

A large majority (89%) of the surveyed sites now use or have plans in place to use ML, and 93% of the survey respondents directly influence ML budgets and/or purchasing. Annual budgets for combined HPC and ML range from under $10,000 to more than $10 million. Slightly more than half of the average budget is spent on standalone HPC, with one-fifth going toward pure AI and the remaining one-quarter toward mixed HPC and ML. Half of the sites have difficulty finding enough ML developers.

The sites represent a wide range of application domains. Only about one in five ML applications are full production mode. ML is most heavily used for image recognition and pre/post-processing of HPC (simulation) applications. Text, numbers, video, audio and video data are all well represented. ML data sizes range from 500GB to more than 1PB. Supervised learning is much more common than unsupervised. Open-source software is only slightly more common than software from other sources. Most sites use multiple frameworks, with PyTorch and TensorFlow most popular. The highest levels of scalability are associated with cloud-based or cloud-provided frameworks.

Download Report - Advisory Clients

Purchase Report

Apply for Academic Citation

Back to All Reports

Access License

Clear

Purchase Reports

Single User Access

Single-User Access ($490): Report is for the sole use of one person, such as for forecasting or product planning. Redistribution or sharing of the reports is prohibited, even internally within the purchasing organization or department. Citation of quotes or data, such as on a web site, in a press release, or for enterprise use, requires approval from Intersect360 Research.

Enterprise Usage

Enterprise-Wide Usage ($2,400): Report can be shared within the purchasing organization—for example, across product, strategy, sales, or forecasting teams. This does not extend to partner organizations, such as suppliers or resellers, but does include contractors working for the purchasing organization. Public citation of quotes or data, such as on a web site or in a press release, requires approval from Intersect360 Research. Advisory service clients are granted enterprise-wide usage rights for all reports in the subscribed service (HPC Advisory Service or Hyperscale Advisory Service), but not reprint rights unless purchased separately.

Reprints

Reprints ($4,800 per report): Full redistribution rights for some or all of a report, such as for PR purposes, including email, print, web, or other distribution. The reports may not be edited in any way except as approved by Intersect360 Research (such as to excerpt a section or to highlight particular text). An organization must have enterprise usage of a report—either through individual report purchase or advisory service access—to purchase reprints. Contact Intersect360 Research to purchase.