Emerging Technology Trends Chris Green Accelerator Controls - PowerPoint PPT Presentation

Emerging Technology Trends Chris Green Accelerator Controls Modernization Workshop Friday September 28, 2018

Outline 2 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Focus and motivation • Triggering / filtering within time constraints will continue to be a demanding challenge, but much HPC-caliber hardware is already commodity. • Event-building on commodity hardware is already happening: high-speed, low latency local networking is already here, affordable and only getting better. • Integrated, heterogeneous systems improving rapidly – PFLOP racks. • Moore’s Law is failing (Intel’s 10nm delayed to ~2020, 5nm limit in ~2024) and Dennard scaling has been dead for >10y. Many novel uses for all those transistors which will require planning and ingenuity to take advantage of: need to start now! • What will be available in 10+ years and how to plan to use it? 3 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

What is driving changes in computing architecture? What does the Future Hold: Strategic Vision for ASCR’s Research Program • DOE’s ASCR program looks to be driving a good Emerging trends are pointing to a future that is increasingly Applications 1. Instrumented : Sensors, satellites, drones, offline repositories amount of future 2. Interconnected : Internet of Things, composable infrastructure, heterogeneous Compilers, Libraries, resources technology Debuggers 3. Automated : Complexity, real ‐ time, machine learning System Software 4. Accelerated : Faster &flexible research pathways for science & research insights • This slide provides a good OS, Runtime What is the role of ASCR’s Research Program in transforming the way we carry out Neuromorphic summary of computing Quantum Others energy & science research? FPGA GPU CPU 1. Post ‐ Moore technologies : Need basic research in new algorithms, software drivers stacks, and programming tools for quantum and neuromorphic systems Digital Non ‐ Digital 2. Extreme Heterogeneity : Need new software stacks, programming models to support the heterogeneous systems of the future 3. Adaptive Machine Learning, Modeling, & Simulation for Complex Systems : Need algorithms and tools that support automated decision making from Prediction Observation intelligent operating systems, in situ workflow management, improved resilience and better computational models. 4. Uncertainty Quantification: Need basic research in uncertainty quantification and artificial intelligence to enable statistically and mathematically rigorous foundations for advances in science domain ‐ specific areas. Modeling Hypothesis 5. Data Tsunami: Need to develop the software and coordinated infrastructure to accelerate scientific discovery by addressing challenges and opportunities associated with research data management, analysis, and reuse. Helland - ASCAC Presentation 9/26/2017 4 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Post-Moore directions Computing Beyond Moore’s Law TABLE 1. Summary of techology options for extending digital electronics. Not that Improvement Class Technology Timescale Complexity Risk Opportunity post-Moore … Medium Low Low Architecture and Advanced energy management Near-Term software advances Advanced circuit design Near-Term High Low Medium System-on-chip specialization Near-Term Low Low Medium Logic specialization/dark silicon Mid-Term High High High Near threshold voltage (NTV) operation Near-Term Medium High High 3D integration and Chip stacking in 3D using thru-silicon vias (TSVs) Near-Term Medium Low Medium packaging Numerous # Opportuni-es#to#Con-nue#Moore’s#Law#Technology!## Metal layers Mid-Term Medium Medium Medium (but$winning$soluAon$is$unclear)$ Active layers (epitaxial or other) Mid-Term High Medium High Revolu-onary# Resistance reduction Superconductors Far-Term High Medium High Heterogeneous#HPC# Crystaline metals Far-Term Unknown Low Medium architectures#&# New#Materials#and#Efficient# solware## y Millivolt switches (a Tunnel field-e f ect transistors (TFETs) Mid-Term Medium Medium High Neuromorphic Spintronics better transistor) Quantum Heterogeneous semiconductors/strained silicon Mid-Term Medium Medium Medium 10+%years%(10%year%lead% Carbon Analog nanotubes z New models of Carbon nanotubes and graphene Far-Term High High High and computaion graphene Adabiatic New devices and materials PETs Piezo-electric transistors (PFETs) Far-Term High High High reversible Data fm ow Beyond transistors Spintronics Far-Term Medium High High Photonic TFETs (new logic Approximate ICs Topological insulators Far-Term Medium High High computing paradigms) Superconducting Devices# Nanophotonics Near/Far-Term Medium Medium High 3D stacking, =me)% adv. packaging Recon f gurable Biological and chemical computing Far-Term High High High CMOS computing Systems General Dark NTV on chip purpose silicon x New architectures and packaging Slide%courtesy%of%John%Shalf% More#Efficient#Architectures#an#Packaging % 10%years%scaling%amer%2025% Nowell – SSDBM, June 29, 2017 Slide%courtesy%of%John%Shalf% 5 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

What is our trajectory? Here is the story. • Seems that given the deluge of data and the end of Moore’s Law, that a major Machine computing problem is imminent. takeover • Thanks in part to AI and post-Moore technological advances, there is hope. Saved AI Likely final state Post- The Singularity Moore Exascale 6 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Xeon Scalable Processor (SP) platform 7 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Xeon SP roadmap • Moving slowly ahead, significantly behind desktops for CPU generation. Gen. Family Process PC Date Server CPU Server Date Lag (mo.) • Main-line 10nm delayed until 1 Nehalem 45nm Nov-08 Xeon-5500 Mar-09 4 likely 2020—tick-tock is dead (or Westemere 32nm Jul-10 Xeon-5600 Feb-11 7 2 Sandy Bridge 32nm Jan-11 E5 Mar-12 14 at least on sabbatical). 3 Ivy Bridge 22nm Apr-12 E5-v2 Sep-13 17 • Not many surprises here: 4 Haswell 22nm Jun-13 E5-v3 Sep-14 15 architecture changes unlikely to 5 Broadwell 14nm Jan-15 E5-v4 Mar-16 14 6 Skylake 14nm Sep-15 Xeon Au/Pt Jul-17 22 be earth-shattering. 7 Kaby Lake 14nm+ Jan-17 • Bronze / Silver / Gold / Platinum 8 Kaby Lake R 14nm+ Aug-17 Coffee Lake 14nm++ Oct-17 variants, up to 28 HT cores, 48 Coffee Lake R 14nm++? Oct-18 PCIE-3 (!) lanes, 6 DDR4 links 9? Cannonlake 10nm? per CPU (Skylake). Icelake 10nm+? Tigerlake 10nm++? • Xeon incorporating Xeon Phi innovations (e.g. AVX512). 8 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Intel Xeon / FPGA hybrid • https://www.nextplatform.com/2018/05/24/a-peek-inside-that-intel-xeon-fpga- hybrid-chip/ • Xeon Skylake SP-6138P / Arria 10 GX 1150 hybrid CPU. 9 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

Intel Configurable Spatial Accelerator • Dataflow engine: non-Von Neumann. • Aimed at exascale: Aurora? • Intel keeping this relatively quiet for now. • Uncertainty over what it means to be “configurable:” might involve silicon customized for an application (FPGA?). Von Neumann 10 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

AMD Zen • “The Epyc 7000s are system-on-chip designs, which means they do not require an external chipset to operate in single-socket mode or two expand out to two-sockets with NUMA interconnects. All the necessary I/O for linking the sockets is on the chips, as are all of the controllers to link out to memory and peripherals.” – https://www.nextplatform.com/2017/06/20/competition-returns-x86-servers-epyc-fashion/ 11 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

AMD Zen 12 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

IBM Power • “Having launched both the scale-out and scale-up Power9s, IBM is now working on a third P9 variant with ‘advanced I/O:’” https://www.hpcwire.com/2018/08/23/ibm-at-hot-chips-whats-next-for-power/ 13 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.

IBM Power 14 Chris Green, Accelerator Controls Modernization Workshop, 2018-09-28, FNAL.