HPC Trends for 2017
3 likes•1,495 views
The document discusses trends in high-performance computing (HPC) for 2017, highlighting architecture specialization, the rise of HPC in the cloud, and the growing influence of machine learning and artificial intelligence. Market analyses reveal the HPC market's revenue, the distinctions between hyperscale and HPC markets, and the anticipated growth rates in areas like deep learning. Key findings point to the predominance of open-source applications in academic contexts and emphasize the transformative effect of the hyperscale market on HPC technologies and resource utilization.
HPC Trends for 2017
- 1. HPC Trends for 2017 HPCAC Lugano Michael Feldman, Managing Editor, TOP500 News
- 2. Intersect360 Research in 2017 10 year history of HPC analyst business Covering “high performance data center” markets, including traditional HPC and supercomputing, high-performance enterprise, cloud, big data, and hyperscale New hyperscale advisory service Weekly podcast, “This Week in HPC” Contributing editors to The Next Platform, market research partners of TOP500.org, media partnerships with HPCwire, insideHPC Sponsors of HPC Advisory Council
- 3. Key Trends • Architecture specialization • HPC in the Cloud • HPC vs. Hyperscale • Machine learning and AI
- 4. HPC User Forward-Looking Views FAVORABLEUNFAVORABLE Intersect360 Research special study, 2016
- 5. Plans for POWER/OpenPOWER Will Definitely Use 13% Will Evaluate 40% No plans to evaluate or use 47% Plans to Use or EvaluatePower/OpenPOWER for HPC "over the next few years" Intersect360Research special study, 2016
- 6. Plans for ARM Will Definitely Use 12% Will Evaluate 47% No plans to evaluate or use 41% Plans to Use or EvaluateARM for HPC "over the next few years" Intersect360Research special study, 2016
- 7. Top Accelerator* Supplier • 38% of the systems reported have accelerators installed • Nvidia was installed in the majority of systems with accelerators and accounted for the majority of accelerators installed. • Intel Xeon Phi was installed in less than 20% of systems. • FPGAs are reported, yet account for less than 2% of systems and less than 1% of accelerators installed. Top Suppliers by Share of Systems (N=316 systems) Top Suppliers by Share of Total Accelerators (N=54,375 accelerators) * Accelerators includes GPUs, co-processors, and FPGAs
- 8. Top Interconnects/Network Suppliers • In all locales, two vendors account for over 50% of the mentions, with Mellanox being the only supplier in all three locales. Entries not specifying supplier were not included in total. • InfiniBand continues to be the performance system network of choice for system interconnects, with about two-thirds of the mentions. Ethernet 10G+ and InfiniBand have about equal shares for storage systems networks. • Intel’s Omni-Path was reported for a few systems and storage systems, with expectations for increased share in 2017. Top System Interconnect Suppliers Top Storage Interconnect Suppliers Top LAN Suppliers
- 9. • About 76% of applications reported were Open Source or ISV software (generally-available applications). • ISV and in-house packages are still the majority at commercial sites, representing about 84% of the packages reported. • Open source accounts for more than 50% of the packages reported at academic and government sites. Top Generally Available Applications Top Open Source Applications Top ISV Applications
- 10. Total HPC Market (Combined HPTC and HPBC) Revenue by Product Class ($M)
- 11. The Hyperscale Market • Included since our 2007 HPC methodology as “ultrascale internet” • In 2015, recognized “Hyperscale” as distinct market in proximity to HPC; began process of separating it from HPC • Defined hyperscale methodology with market research and feedback from interested clients • Published definitions and methodology report in April 2016
- 12. Hyperscale Market Size and Tiers • Intersect360 Research has a tiered model for the hyperscale market, as follows: Tier Annual IT Orgs Market Size Tier 1 > $1B 8 ~$20B Tier 2 > $100M ~40 ~$10B Tier 3 > $1M hundreds ~$5B Total: ~$35B DRAFT, Intersect360 Research, 2016 • We anticipate publication of market size very soon, possibly by end of year
- 13. Deep Learning • Intersect360 Research tracks AI (including deep learning, machine learning, cognitive computing, etc.) as part of the hyperscale market – Similar to but distinct from HPC – Low precision, intensely parallel, strong affinity to public cloud • DL is not a vertical market. It is more akin to an algorithm or method of computation, like an FFT • Cloud providers and end users are in early stages of investment for their applications
- 14. Deep Learning Size and Growth • According to our analysis, the deep learning market opportunity is $2.0B to $2.5B for 2016 • This represents about 6.5% of the hyperscale market, which we track separately from HPC • We expect growth rate to be very high for the next two years, 2017-18, close to 100% Y/Y • After that, growth will moderate, and possibly even contract by end of five-year period
- 15. The Next Ten Years in HPC (Write These Down; Grade Us in 2027) 1. Fundamental drivers of HPC remain strong. Forever. 2. Specialization proliferates again. Custom architectures will reemerge in the market. 3. Public cloud (utility computing) remains well below 10% of the HPC market through the next 10 years. 4. Object storage will take off in commercial markets. 5. The hyperscale market will have a tremendous effect: – Consuming HPC-associated technologies like GPU, InfiniBand – Defining system configurations and product specifications – Influencing software and middleware – Introducing new revolutionary applications based on AI (including deep learning) and augmented reality. These are hyperscale applications that will predominant be run on cloud- based resources.
- 17. HPC Trends for 2017 HPCAC Lugano Thank you!