The International HPC Certification Forum and AU-NZ

Lev Lafayette

Senior HPC Support and Training OfficereResearchAustralasia, Melbourne, 2018

1

The Problem at Hand

Datasets size and complexity is increasing faster than personal computers are improving, an issue which has been known for many decades [1, 2, 3] (e.g., SKA, vehicle automation etc); expected 10x increase between 2013 and 2020.

High Performance Computing (HPC) is the most effective and efficient means to process large and complex datasets with excellent RoI [4, 5]. Some nascent sociology that correlates research output with HPC provision [6].

Therefore, availability and usage of HPC systems is a critical issue (i.e., an existential risk) for research organisations [6]. However, many researchers come to need HPC systems but have minimal skills (if any) in using such systems. Uptake is improving, usage is improving, but still lower than demand.

2

Current Approaches

There are very few courses in formal curriculum (e.g., one course UoM, used to be a Grad Cert at UWA). What does exist is on the postgraduate level. Training of researchers is therefore typically left to individual HPC sites (e.g., Pawsey, NCI, UniMelb, NeSI et al) as demand increases.

There is inconsistency in training approach and content, despite a high level of homogeneity in HPC skills; individual sites differ but core concepts are similar (Linux skills, scheduling requests, environment modules etc).

Most HPC engineers do not have education experience or knowledge e.g., proximal development, andragogy (adult) vs pedagogy (child) continuum. disciplinary learning techniques versus “learning styles” (fake news!), structured and integrated knowledge, formative vs summative assessment (right answer versus understanding), self-efficacy etc.

Existing training programmes are suboptimal in efficiency and effectiveness; desire for better collaboration.

3

The International HPC Certification Forum

The International HPC Certification Forum ("the Forum") had its inaugural meeting at ISC in 2018. Developed from the Performance Conscious HPC (PeCoH) project in 2017 with the Hamburg HPC Competence Center (HHCC) as a virtual institution [8].

Developing skill tree with different competency branches (HPC Knowledge, Performance Engineering, HPC Environment etc), and skill leaves [9]. The leaves make up over 100 skills. Most recent version hasAdministration and Big Data branches as well.

Forum separates Content Delivery from Certification; will offer examination, "endorsed training" material to build an ecosystem of participants. 4

Skill Tree, Skill Branches, Skill Leaves

5

Australian and New Zealand Contributions

Australia has been involved from the first AGM of the International Forum and now has with two topic chairs on the governing Board (HPC Knowledge, Software Development). Lightning talk at subsequent follow-up Zoom meeting held in 2018 following eResearch for interested parties.

Opportunity for local development of resources, both for teaching, and for contributions to the International Forum. Resource can include; HPC content, teaching delivery, assessment methods.

High priority of higher degree skilltraining noted by Commonwealth [10]

Overall objective: increase regional research ouput. 6

Future Steps

The International Forum needs to complete the skill tree, develop a summative certification exam, and establish PGP key-signing for verification of certificates.

Content development and membership for the Australia & New Zealand HPC Educators repository.https://github.com/australia-new-zealand-hpc-educators

Australia & New Zealand HPC Educators workshop and hackathon? Plan to have 20 or so. HPC educators from AU and NZ spend a daydiscussing techniques, content, and assessment, anda day smashing out content for the repositories.

7

References

[1] Martin Hilbert, Priscila López. The World’s Technological Capacity to Store, Communicate, and Compute Information, Science Vol. 332 no. 6025 1 April 2011, pp. 60-65[2] Guo, Huadong, et al. "Scientific big data and digital earth." Chinese Science Bulletin 59.35 (2014): 5066-5073.[3] Markov, Igor L. "Limits on fundamental limits to computation." Nature 512.7513 (2014): 147.[4] Fox, Geoffrey, et al. "Big data, simulations and hpc convergence." Big Data Benchmarking. Springer, Cham, 2015. 3-17.[5] Joseph, Earl., et al., Creating Economic Models Showing the Relationship Between Investments in HPC and the Resulting Financial ROI and Innovation — and How It Can Impact a Nation's Competitiveness and Innovation, IDC Special Study, October 2013 [6] Apon, Amy., et al., High Performance Coputing Instrumentation and Research Productivity in U.S. Universities, Journal of Information Technology Impact, Vol 10, No 2, pp87-98, 2010[7] Lafayette, Lev. "Critical Issues in the Teaching of High Performance Computing to Postgraduate Scientists.", Proceedings of the International Conference on Computational Science, 2014[8] Himstedt, Kai., et al., An HPC Certification Program Proposal Meeting HPC Users' Varied Backgrounds, 2018 [9] Kunkel, Julian., et al,, Towards an HPC Certification Program, Journal of Computer Science Education, Volume 10, Issue 1, 2019, p88-89[10] Review of Australia's Resarch Training System, Commonwealth of Australia, 2017

8

Acknowledgements

Thanks to Ann Backhaus (Pawsey Supercompter Centre), Roger Edberg (National Computional Infrastructure), and Waseem Kamleh (University of Adelaide) for their assistance and advice in reviewing this presentation.

Thanks to the University of Melbourne for providing time and financial support for the project.

The International HPC Certification Forum and PeCoH were supported by the German Research Foundation (DFG).

9