×

The future of computing beyond Moore’s law. (English) Zbl 1462.68001

Summary: Moore’s Law is a techno-economic model that has enabled the information technology industry to double the performance and functionality of digital electronics roughly every 2 years within a fixed cost, power and area. Advances in silicon lithography have enabled this exponential miniaturization of electronics, but, as transistors reach atomic scale and fabrication costs continue to rise, the classical technological driver that has underpinned Moore’s Law for 50 years is failing and is anticipated to flatten by 2025. This article provides an updated view of what a post-exascale system will look like and the challenges ahead, based on our most recent understanding of technology roadmaps. It also discusses the tapering of historical improvements, and how it affects options available to continue scaling of successors to the first exascale machine. Lastly, this article covers the many different opportunities and strategies available to continue computing performance improvements in the absence of historical technology drivers.

MSC:

68-03 History of computer science
01A67 Future perspectives in mathematics
68Mxx Computer system organization
91B99 Mathematical economics
PDFBibTeX XMLCite
Full Text: DOI Link

References:

[1] Moore GE. 1965 Cramming more components onto integrated circuits. Electronics 38, 33-35. (doi:10.1109/N-SSC.2006.4785860) · doi:10.1109/N-SSC.2006.4785860
[2] Mack C. 2015 The multiple lives of Moore’s law. IEEE Spectrum 52, 31-31. (doi:10.1109/MSPEC.2015.7065415) · doi:10.1109/MSPEC.2015.7065415
[3] Markov IL. 2014 Limits on fundamental limits to computation. Nature 512, 147-154. (doi:10.1038/nature13570) · doi:10.1038/nature13570
[4] Shalf JM, Leland R. 2015 Computing beyond Moore’s law. IEEE Computer 48, 14-23. (doi:10.1109/MC.2015.374) · doi:10.1109/MC.2015.374
[5] Law M, Colwell RC. 2013 The chip design game at the end of Moore’s Law. Hot Chips Symposium. Keynote, pp. 1-16. See www.hotchips.org/wp-content/uploads/hc_archives/hc25/HC25.15-keynote1-Chipdesign-epub/HC25.26.190-Keynote1-ChipDesignGame-Colwell-DARPA.pdf.
[6] Thompson N, Spanuth S. 2018 The decline of computers as a general purpose technology: why deep learning and the end of Moore’s Law are fragmenting computing. SSRN abstract 3287769. (doi:10.2139/ssrn.3287769) · doi:10.2139/ssrn.3287769
[7] Jouppi NP et al. 2017 In-datacenter performance analysis of a tensor processing unit. In Newslett. ACM SIGARCH Computer Architecture News - ISCA’17, vol. 45 (2), May, pp. 1-12. New York, NY: ACM. (doi:10.1145/3079856.3080246) · doi:10.1145/3079856.3080246
[8] Hsu J. 2016 Nervana systems: turning neural networks into a service. IEEE Spectrum 53, 19. (doi:10.1109/MSPEC.2016.7473141) · doi:10.1109/MSPEC.2016.7473141
[9] Facebook Inc. 2017 Introducing Big Basin: our next-generation AI hardware. See https://code.facebook.com/posts/1835166200089399/introducing-big-basin-our-next-generation-ai-hardware/.
[10] Caulfield A. 2016 A cloud-scale acceleration architecture. In 2016 49th Annu. IEEE/ACM Int. Symp. on Microarchitecture (MICRO-49), Taipei, Taiwan, 15-19 October, 13pp. New York, NY: IEEE. (doi:10.1109/MICRO.2016.7783710) · doi:10.1109/MICRO.2016.7783710
[11] Shao YS, Xi SL, Srinivisan V, Wei GY, Brooks D. 2016 Co-designing accelerators and SoC interfaces using gem5-Aladdin. In 2016 49th Annu. IEEE/ACM Int. Symp. on Microarchitecture (MICRO-49), Taipei, Taiwan, 15-19 October, 12pp. New York, NY: IEEE. (doi:10.1109/MICRO.2016.7783751) · doi:10.1109/MICRO.2016.7783751
[12] Shaw DE et al. 2014 Anton 2: raising the bar for performance and programmability in a special-purpose molecular dynamics supercomputer. In SC’14: Proc. Int. Conf. for High Performance Computing, Networking, Storage and Analysis, New Orleans, LA, 16-21 November, pp. 41-53. New York, NY: IEEE. (doi:10.1109/SC.2014.9) · doi:10.1109/SC.2014.9
[13] Ohmura I, Morimoto G, Ohno Y, Hasegawa A, Taiji M. 2014 MDGRAPE-4: a special-purpose computer system for molecular dynamics simulations. Phil. Trans. R. Soc. A 372, 20130387. (doi:10.1098/rsta.2013.0387) · doi:10.1098/rsta.2013.0387
[14] Prabhakar R, Zhang Y, Koeplinger D, Feldman M, Zhao T, Hadjis S, Pedram A, Kozyrakis C, Olukotun K. 2018 Plasticine: a reconfigurable accelerator for parallel patterns. IEEE Micro 38, 20-31. (doi:10.1109/MM.2018.032271058) · doi:10.1109/MM.2018.032271058
[15] Johansen H et al. 2014 Software productivity for extreme-scale science. Report on DOE Workshop. See www.orau.gov/swproductivity2014/SoftwareProductivityWorkshopReport2014.pdf.
[16] Asanovic K et al. 2006 The landscape of parallel computing research: a view from Berkeley. EECS Department, UC Berkeley. Technical Report No. UCB/EECS-2006-183. See www2.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-183.pdf.
[17] Miller DAB, Ozaktas HM. 1997 Limit to the bit-rate capacity of electrical interconnects from the aspect ratio of the system architecture. J. Parallel Distrib. Comput. 41, 42-52. (doi:10.1006/jpdc.1996.1285) · doi:10.1006/jpdc.1996.1285
[18] Miller DAB. 2000 Rationale and challenges for optical interconnects to electronic chips. Proc. IEEE 88, 728-749. (doi:10.1109/5.867687) · doi:10.1109/5.867687
[19] Horowitz M, Yang CKK, Sidiropoulos S. 1998 High-speed electrical signaling: overview and limitations. IEEE Micro 18, 12-24. (doi:10.1109/40.653013) · doi:10.1109/40.653013
[20] Kogge P, Shalf J. 2013 Exascale computing trends: adjusting to the ‘new normal’ for computer architecture. Comput. Sci. Eng. 15, 16-26. (doi:10.1109/MCSE.2013.95) · doi:10.1109/MCSE.2013.95
[21] Unat D et al. 2017 Trends in data locality abstractions for HPC systems. IEEE Trans. Parallel Distrib. Syst. 28, 3007-3020. (doi:10.1109/TPDS.2017.2703149) · doi:10.1109/TPDS.2017.2703149
[22] Unat D, Shalf J, Hoefler T, Dubey A, Schulthess T. 2014 PADAL: Programming Abstractions for Data Locality Workshop Series. See www.padalworkshop.org/.
[23] Meyer H, Sancho JC, Quiroga JV, Zyulkyarov F, Roca D, Nemirovsky M. 2017 Disaggregated computing. An evaluation of current trends for datacentres. Procedia Comput. Sci. 108, 685-694. (doi:10.1016/j.procs.2017.05.129) · doi:10.1016/j.procs.2017.05.129
[24] Taylor J. 2015 Facebook’s data center infrastructure: open compute, disaggregated rack, and beyond. In 2015 Optical Fiber Communications Conf. and Exhibition (OFC), Los Angeles, CA, 22-26 March, 1p. New York, NY: IEEE. See https://ieeexplore.ieee.org/abstract/document/7121902.
[25] Tokunari M, Hsu HH, Toriyama K, Noma H, Nakagawa S. 2014 High-bandwidth density and low-power optical MCM using waveguide-integrated organic substrate. J. Lightwave Technol. 32, 1207-1212. (doi:10.1109/JLT.2013.2292703) · doi:10.1109/JLT.2013.2292703
[26] Bergman K. 2018 Empowering flexible and scalable high performance architectures with embedded photonics. In 2018 IEEE Int. Parallel and Distributed Processing Symp. (IPDPS), Vancouver, BC, Canada, 21-25 May, p. 378. New York, NY: IEEE. (doi:10.1109/IPDPS.2018.00047) · doi:10.1109/IPDPS.2018.00047
[27] Michelogiannakis G, Wilke J, Teh MY, Glick M, Shalf J, Bergman K. 2019 Challenges and opportunities in system-level evaluation of photonics. In Metro and Data Center Optical Networks and Short-Reach Links II, SPIE OPTO, San Francisco, CA, 2-7 February, Proc. SPIE 10946. (doi:10.1117/12.2510443) · doi:10.1117/12.2510443
[28] Nikonov DE, Young IA. 2016 Overview of beyond-CMOS devices and a uniform methodology for their benchmarking. Proc. IEEE 101, 2498-2533. (doi:10.1109/JPROC.2013.2252317) · doi:10.1109/JPROC.2013.2252317
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.