[hpc-announce] [CALL FOR PARTICIPATION] HOT Interconnects 2020 - ALMOST HERE - WEDNESDAY, 8/19/2020 Advance Program August 19-21, 2020 - A Symposium on High-Performance Interconnects
Si, Min
msi at anl.gov
Tue Aug 18 11:56:24 CDT 2020
Virtual Conference HOT Interconnects 27
August 19-21, 2020
Advance Program
A Symposium on High-Performance Interconnects
Registration at HOT Interconnects 2020 is free!
Register now at https://www.123signup.com/register?id=rnsmq
Register for HOT Interconnects 2020!
==========================================================================
Welcome to HOT Interconnects 27, to be held August 19-21, 2020. Due to the spread of Covid-19,
there will not be an in-person event for Hot Interconnects 27, but it will be held as an on-line
virtual conference. All tutorials, keynotes, and sessions will be broadcast so that all registered
conference attendees will be able to access. Logon instructions will be emailed to all attendees
before the start of the conference. The broadcast will be in the Pacific Time Zone. Per-talk Slack
channels connect registered attendees to presenters for live Q&A and discussions with other attendees.
We hope that you enjoy this new format conference and all the interesting presentations described
here.
Don Draper, co-general chair.
==========================================================================
Stay current with Hot Interconnects with these social media accounts:
Facebook: https://www.facebook.com/hoti.org
Twitter: https://twitter.com/IEEEHOTI
--------------------------------------------------------------------------
Message from the TPC Chairs
--------------------------------------------------------------------------
Welcome to the 2020 IEEE 27th Annual Symposium on High-Performance Interconnects. This
conference brings together designers and architects of high-performance networks, with
presentations from universities and industrial research laboratories that focus on the
latest developments in leading-edge technologies. This year’s conference is entirely
virtual due to the COVID-19 pandemic, representing unique challenges and opportunities
for the conference program.
Our world-class team of 19 Technical Program Committee members returned insightful reviews
for 29 submissions. Each paper received between three and four reviews, thus providing
ample and valuable feedback to the authors, and ensuring high confidence on the outcome
of the review process. Out of the 29 submissions, 10 were accepted.
The technical program comprises five technical sessions that cover a number of hot topics
in high-performance interconnection networks. The paper sessions open with hardware related
papers, and are followed by a session on frameworks and architectures. Our second day has
three sessions: one session is dedicated to advances in machine learning for networks; a
second session centers on topologies and routing; and our last session concentrates on
containers and virtualization. In addition to the technical papers sessions we have two
distinguished keynote speakers, Michael Kagan (NVIDIA) and Brian Barrett (AWS EC2), in
addition to two invited talks on cutting-edge network design.
The organization of a major conference such as Hot Interconnects is a demanding and
time-consuming task. We would like to thank the TPC members for their thorough reviews
and active participation, despite the traditionally short review cycle of Hot Interconnects
(less than one month). Owing to their dedication and promptness, Hot Interconnects is able
to bring you high-quality, up-to-the-minute contributions.
We especially thank General Chairs Don Draper and Khaled Hamidouche for their leadership and
guidance and for their enormous contributions in organizing this meeting. In addition, we
would like to express our gratitude to the members of the Organizing Committee for their
contributions.
Finally, and most importantly, we thank the technical paper authors, keynote and tutorial
presenters, and panelists for their excellent contributions that make this event the premier
international conference for high-performance interconnects.
We hope you will enjoy the symposium!
Ryan Grant, Sandia National Laboratories
Manjunath Gorentla Venkata, NVIDIA
HOTI 2020 Technical Program Co-Chairs
--------------------------------------------------------------------------
HOT Interconnects 2020 Advance Program
--------------------------------------------------------------------------
Day 1: Wednesday, August 19, 2020
8:50 - 9:00 Introduction and Welcome - Co-General Chairs - Don Draper, Khaled Hamidouche
9:00 - 10:00 Keynote 1: The Data Center is a Computer
Michael Kagan, Chief Technology Officer, NVIDIA
10:00 - 10:10 Break
10:10 - 10:30 Diamond Sponsor: Intel
Ling Liao, Chief Architect of Photonic Integration, Intel Corp.
Session A: Hardware
10:30 - 11:00 A. 1. PCI Express® 6.0 at 64.0 GT/s with PAM-4 signaling: a low latency, high bandwidth,
high reliability, and cost-effective interconnect
Debendra Das Sharma, Intel
11:00 - 11:30 A. 2. Bunch of Wires: An Open Die-to-Die Interface
Shahab Ardalan (Ayar Labs), Halil Cirit (Facebook),
Ramin Farjad (Marvell), Mark Kuemerle (Marvell),
Ken Poulton (Keysight), Suresh Subramanian (Apex Semiconductor),
Bapi Vinnakota (Broadcom)
11:30- 12:30 Lunch
Session B: Frameworks and Architectures
12:30 - 13:00 B. 1. Configurable network Protocol Accelerator (COPA) - An Integrated
Networking/Accelerator Hardware/Software Framework
Venkata Krishnan, Olivier Serres and Michael Blocksome, Intel
13:00 -13:30 B. 2. The Open Domain-Specific Architecture
Kevin Drucker (Facebook), Dharmesh Jani (Facebook),
Ishwar Agarwal (Microsoft), Gary Miller (NXP Semiconductors),
Millind Mittal (Xilinx), Robert Wang (AnalogX)
and Bapi Vinnakota (Broadcom)
13:30-13:50 Break
Silver Sponsor: Lenovo
13:50 - 14:20 Invited Talk: TeraPHY: An O-band WDM Electro-optic Chiplet for Low Power,
Terabit/s Optical I/O
Mark Wade, Ayar Labs
14:20 - 14:30 Gold Sponsor: GiGaIO - The next step in computing infrastructure
Alan Benjamin, President and CEO
14:30 - 16:00 Panel: SmartNIC or DPU, Who Wins?
Moderator:
Scott Schweitzer, Xilinx
Panelists:
Andrew Gospodarek, Broadcom
Michael Kagan, NVIDIA
Gordon Brebner, Xilinx
Pradeep Sindhu, Fungible
Vipin Jain, Pensando
--------------------------------------------------------------------------
Day 2: Thursday, August 20, 2020
9:00 - 10:00 Keynote 2: Scalable Reliable Datagrams: Combining High Performance Computing
and Cloud-Scale Networks
Brian Barrett, AWS EC2
10:00 - 10:20 Break
10:20 - 10:30 Gold Sponsor: TidalScale - Run your Largest Workloads at Half the Cost
Eric Fitzpatrick
Session C: Machine Learning
10:30 - 11:00 C. 1. Scalable Distributed Training of Recommendation Models: An ASTRA-SIM +
NS3 case-study with TCP/IP transport
Saeed Rashidi (Georgia Tech), Pallavi Shurpali (Facebook),
Srinivas Sridharan (Facebook), Naader Hassani (Facebook),
Dheevatsa Mudigere (Facebook), Krishnakumar Nair (Facebook),
Misha Smelyanskiy (Facebook), Tushar Krishna (Georgia Tech)
11:00 - 11:30 C. 2. Communication-Efficient Distributed Deep Learning with GPU-FPGA
Heterogeneous Computing
Kenji Tanaka, Yuki Arikawa, Ito Tsuyoshi, Kazutaka Morita,
Naru Nemoto, Fumiaki Miura, Kazuhiko Terada, Junji Teramoto
and Takeshi (NTT Corporation)
11:30 - 11:35 Silver Sponsor: AlgoLogic - A quick Update on Algorithms and Logic
John Lockwood
11:35 - 12:30 Lunch
Session D: Topologies and Routing
12:30 - 13:00 D. 1. Optimizing Packet Dropping by Efficient Congesting-Flow Isolation in
Lossy Data-Center Networks
Cristina Olmedilla (Universidad de Castilla-La Mancha),
Jesús Escudero-Sahuquillo (Universidad de Castilla-La Mancha),
Pedro Javier García (Universidad de Castilla-La Mancha),
Francisco Alfaro Cortés (Universidad de Castilla-La Mancha),
Francisco J. Quiles (Universidad de Castilla-La Mancha),
José L. Sánchez (Universidad de Castilla-La Mancha),
Wenhao Sun (Huawei), Xiang Yu (Huawei), Xu Yonghui (Huawei),
Jose Duato (Universitat Politècnica de València)
13:00 - 13:30 D. 2. Isolated Trees in Multi-Tenant Fat Tree Datacenters for In-Network Computing
Ori Rottenstreich (Technion), Jose Yallouz
and Lion Levi (Mellanox Technologies)
Silver Sponsor: Ayar Labs
13:30 - 13:45 Break
Session E: Containers and Virtualization
13:45 - 14:15 E. 1. Kano: Efficient Container Network Policy Verification
Yifan Li (Tsinghua University), Chengjun Jia (Tsinghua University), Xiaohe Hu (Tsinghua University), Jun Li (RITT)
14:15 - 14:45 E. 2. Reliable and Time-efficient Virtualized Function Placement
Roi Ben Haim and Ori Rottenstreich (Technion)
14:45 - 14:55 Gold Sponsor: IBM - Power10 Chipset
14:55 - 15:10 Break
15:10 - 15:40 Invited Talk 2: Constructing a Computationally-Efficient Distributed GPU Appliance
In a PCI Express-based Network
Scott Taylor, GigaIO
15:40 - 15:45 Close Don Draper, Co-General Chair
--------------------------------------------------------------------------
Day 3: Friday, August 21, 2020
Tutorials:
9:00 - 12:30 Exploiting HPC for distributed Deep Learning
Dhabaleswar K. (DK) Panda, Hari Subramoni, Arpan Jain, and Ammar Ahmad Awan - OSU
12:30 - 13:30 Lunch
13:30 - 17:00 InfiniBand, High-speed Ethernet, RoCE, Omni-Path, EFA and Slingshot for Beginners
Dhabaleswar K. (DK) Panda, Hari Subramoni - OSU
17:00 - 17:05 Close
--------------------------------------------------------------------------
Send questions relating to the program to the
program chairs Ryan Grant and Manjunath Gorentla Venkata at:
program at hoti.org
and questions relating to conference operation
to the general chairs Don Draper and Khaled Hamidouche at:
info at hoti.org
Check the HOT INTERCONNECTS 27 web page for updates:
www.hoti.org
Send questions relating to registration to
registration at hoti.org
Sponsor Hot Interconnects! Email: sponsor at hoti.org
Sister conferences:
www.hotchips.org
http://www.coolchips.org/
http://www.mcsoc-forum.org/
Go to the Hot Interconnects web site for updates: http://www.hoti.org
=========================================================================
More information about the hpc-announce
mailing list