Date: May 20, 2019
Venue: RIKEN Center for Computational Science (R-CCS), Seminar room (1st floor)
(7-1-26 Minatojima-minami-machi, Chuo-ku, Kobe, Hyogo 650-0047, Japan)
Area Map: Area Map around R-CCS
This workshop is hosted by RIKEN Center for Computational Science and cosponsored by "General Computer Architecture and System Software that Accomplish 100x Performance on 2028” in NEDO exploratory research program "Development of efficient high performance AI chips and next-generation computing technology"
|9:50 - 10:00||Opening|
|10:00 - 10:50||Keynote: “Architecture Specialization for HPC to Extend Technology Scaling”
(John Shalf, Lawrence Berkeley National Laboratory)
The tapering of traditional technology changing has substantially changed requirements for future architectures for large-scale datacenters and HPC systems. Architectural specialization is creating new datacenter requirements such as emerging accelerator technologies for machine learning workloads and rack disaggregation strategies will push the limits of current technologies. This talk provides an updated view of what a 2021-2023 system might look like and the challenges ahead, based on our most recent understanding of technology roadmaps. It also will discuss the tapering of historical improvements in lithography, and how it affects options available to continue scaling of successors to the first exascale machines. Many of these emerging hardware specializations can offer dramatic performance improvements for targeted applications, but will require a reconsideration of how to co-design hardware and algorithms for future generations of systems.
|10:50 - 11:25||
Invited Talk 1: “Silicon Photonics and Optical Hub for Computing System”
(Tatsuya Usuki, Kenji Mizutani and Yutaka Urino, PETRA)
|11:25 - 13:00||Lunch|
|13:00 - 13:35||Invited Talk 2: “Bandwidth Steering in HPC using Silicon Nanophotonics”
(George Michelogiannakis, Lawrence Berkeley National Laboratory)
|13:35 - 14:10||Invited Talk 3: “qFirm: Digital Firmware for Classical Control of Qubits”
(Farzad Fatollahi-Fard, Lawrence Berkeley National Laboratory)
|14:10 - 14:45||Invited Talk 4: Extending Classical Processors to Support Future Large Scale Quantum Accelerators
(Butko Anastasiia, Lawrence Berkeley National Laboratory)
|14:45 - 15:05||Break (Cafe Time)|
|15:05 - 15:40||Invited Talk 5: “How open source designs will drive the next generation of HPC Systems”
(David Daniel Donofrio, Lawrence Berkeley National Laboratory)
|15:40 - 16:15|| Invited Talk 6: “PARADISE: Modeling and Simulation of Emerging Post-CMOS Devices and Architectures”
(Dilip Vasudevan, Lawrence Berkeley National Laboratory)
|16:15 - 16:35||Break (Cafe Time)|
|16:35 - 17:30||Panel discussion
(Moderator: Prof. Matsuoka, Panelist: all the speakers)
|17:30 - 17:40||Closing|
Note: The afternoon sessions will be held as part of R-CCS Cafe (R-CCS seminar).
Abstracts of Talks 2-6 are posted on the R-CCS Cafe web page.
For questions please send an email to: p-moore[at]ml.riken.jp. (Replace [at] with @ to send an email.)