Energy efficient real-time computing for extremely large telescopes with GPU

Size: px
Start display at page:

Download "Energy efficient real-time computing for extremely large telescopes with GPU"

Transcription

1 Energy efficient real-time computing for extremely large telescopes with GPU Florian Ferreira & Damien Gratadour Observatoire de Paris & Université Paris Diderot 1 Project # funded by European Commission under program H2020-EU coordinated in H2020-FETHPC-2014

2 Observing stars from the ground Atmospheric turbulence Modify the trajectory of light rays when they cross the atmosphere Reduces astronomical images quality Similar to the effect of aberrations in an optical system Adaptive optics Compensate in real-time for the effect of optical aberrations on image quality Already in use on most 5-10m astronomical telescope to provide nominal image quality whatever the turbulence conditions 2

3 Adaptive optics Disturbed wavefront Compensate in real-time the wavefront perturbations Using a wavefront sensor to measure them Using a deformable mirror to reshape the wavefront Deformable mirror Real-time controller Beamsplitter Corrected wavefront Wavefront sensor High resolution camera Commands to the mirror must be computed in real-time (1ms rate) 3 Loop closed Loop open

4 Adaptive optics Example with observations of the moon using a 8m telescope Without AO With AO 4

5 European Extremely Large Telescope (E-ELT) 39m diameter telescope! Credit : ESO Primary mirror will be made of ~800 segments of 1.4m diameter Theoretical resolution in the near-infrared : 10 milliarcseconds, i.e. 2x105 smaller than the full moon (~30 arcminutes) About 1000 m2 of collecting area, i.e. 15 times more sensitive than the largest state of the art professional telescopes currently in operation 5

6 European Extremely Large Telescope 39m diameter telescope : 100m dome, 2800 tones structure 360, seismic safe (Chile) 1.2 G project, first light foreseen in 2024 European project led by ESO involving public labs and private companies AO system complexity scales as the square of telescope diameter Credits : ESO x25 more complex AO systems 6 Credits : ESO

7 Adaptive optics Disturbed wavefront Compensate in real-time the wavefront perturbations Using a wavefront sensor to measure them Using a deformable mirror to reshape the wavefront Deformable mirror Real-time controller Beamsplitter Corrected wavefront Wavefront sensor High resolution camera Commands to the mirror must be computed in real-time (1ms rate) 7 Loop closed Loop open

8 AO real-time controller Highly heterogeneous HPC facility 8

9 The Green Flash project Large European initiative Goal : prototype a generic RTC for the next generation of AO on extremely large telescopes 4 partners in Europe (2 academic partners + 2 SMEs), project lead : Observatoire de Paris, 3.8 M investment funded under the H2020 program (FET-HPC, project #671662) Assess various technologies (CPUs, GPUs, FPGAs) and find the best trade-off Assemble a full featured prototype in the lab by

10 GPUs for real-time computing Naive implementation : several copies and multiple kernel launches 10

11 Low latency data transfer to GPUs Critical issue: high latency data acquisition from the camera using an off-the-shelf frame grabber Typical approach : Double-copy mechanism Multiple kernel launches DDR Mem. DDR Mem. GPU FPGA DMA engine PCIe bus CPU High jitter in performance not compatible with time deterministic constraint 11 DDR Mem. 10 Gbe Frame-grabber Serial interface Pixel data

12 Low latency data transfer to GPUs Solution : using GPUdirect coupled to a persistent kernel strategy DDR Mem. DDR Mem. GPU FPGA DMA engine PCIe bus No more jitter in the execution CPU Efficient data distribution 12 DDR Mem. 10 Gbe Frame-grabber Serial interface Pixel data

13 Low latency data transfer to GPUs Enabled through a smart interconnect strategy Based on FPGA boards Using dedicated dev. tools (a.k.a. QuickPlay from Accelize) FPGA design made easy 13

14 GPUs for real-time computing Our solution : using GPUdirect coupled to persistent kernels 14

15 Full scale prototyping Relies on NVIDIA DGX-1 Being assessed in the lab today. First performance estimates are consistent with system specifications NVIDIA DGX-1 WFS pixels (4x10GbE) WFS pixels (4x10GbE) WFS pixels (4x10GbE) Fast intra-cluster com. (40 GbE) X86 X86 NIC GPU GPU NIC WFS pixels (4x10GbE) NIC GPU GPU NIC NIC GPU GPU NIC NIC GPU GPU NIC WFS pixels (4x10GbE) WFS pixels (4x10GbE) Fast intra-cluster com. (40 GbE) NVlink PCIe 15

16 Full scale prototyping Relies on NVIDIA DGX-1 Master / slaves strategy implemented on DGX-1 Very low jitter introduced by sync / comm. 16

17 Tomographic adaptive optics Small patches of interest in a large field of view 17

18 Tomographic adaptive optics Multiple guide stars using Lasers and multiple correctors 18

19 Tomographic Adaptive Optics New concept of Multi-Object AO Multiplexed-AO (several WFS, several DM) Credits : E. Marchetti, ESO 19 Credits : E. Marchetti, ESO

20 AO loop supervision High throughput HPC facility Optimize loop performance by regular control matrix update Control matrix is built through statistical analysis of real-time data 20

21 AO loop supervision AO loop supervisor pipeline : computing a tomographic reconstructor (ToR) for AO Ongoing collaboration with KAUST 21

22 AO loop supervision Computing the tomographic reconstructor on DGX-1 25 sec to compute the scale is well within specs! and 8 P100 GPUs perform almost 20x better than a single KNL P100 is more than 2x more efficient than KNL including Nvlink communications. 22

23 AO simulation ToR is also at the core of our AO simulation pipeline (based on pure linear algebra to simulate system behavior) 23

24 AO simulation Producing AO performance map for a given patch of the sky and given turbulence conditions 24

25 AO simulation Portable SW stack, studying performance scaling over several generations of HW : GPUs always win Regular performance increase at each new HW release Credits : Hatem Ltaief 25

26 Conclusions and future work GPUs help to design the largest telescopes critical systems Provide the required throughput for efficient simulations of large scale AO systems GPUs proposed to be at the core of telescope operations : As compute units in the real-time controller, coupled to a smart interconnect strategy with sensors Assembled in a standard cluster for regular tomographic matrix update What's next? Large scale simulations for system design trade-off and sky coverage studies Full scale prototyping of GPU-based real-time controller Study new supervision strategy with optimized linear algebra (H-matrix formalism) 26

27 That s it for today Thank you! Credits : ESO 27

Driving the next generation of Extremely Large Telescopes using Adaptive Optics with GPUs

Driving the next generation of Extremely Large Telescopes using Adaptive Optics with GPUs Driving the next generation of Extremely Large Telescopes using Adaptive Optics with GPUs Damien Gratadour LESIA, Observatoire de Paris Université Paris Diderot LESIA, Observatoire de Paris ANR grant ANR-12-MONU-0022

More information

GTC 2017 Green Flash Persistent Kernel : Real-Time, Low-Latency and HighPerformance Computation on Pascal. Julien BERNARD

GTC 2017 Green Flash Persistent Kernel : Real-Time, Low-Latency and HighPerformance Computation on Pascal. Julien BERNARD Green Flash Persistent Kernel : Real-Time, Low-Latency and HighPerformance Computation on Pascal Julien BERNARD Project #671662 funded by European Commission under program H2020-EU.1.2.2 coordinated in

More information

Efficient Observations Forecast for the World s Biggest Eye Using DGX-1

Efficient Observations Forecast for the World s Biggest Eye Using DGX-1 Efficient Observations Forecast for the World s Biggest Eye Using DGX-1 Damien Gratadour 1 and Hatem Ltaief 2 1 LESIA, Observatoire de Paris and Université Paris Diderot, France 2 Extreme Computing Research

More information

A Real Time Controller for E-ELT

A Real Time Controller for E-ELT A Real Time Controller for E-ELT Addressing the jitter/latency constraints Maxime Lainé, Denis Perret LESIA / Observatoire de Paris Project #671662 funded by European Commission under program H2020-EU.1.2.2

More information

A Real Time Controller for E-ELT

A Real Time Controller for E-ELT A Real Time Controller for E-ELT Addressing the jitter/latency constraints Maxime Lainé, Denis Perret LESIA / Observatoire de Paris Project #671662 funded by European Commission under program H2020-EU.1.2.2

More information

FPGA based microserver for high performance real-time computing in Adaptive Optics

FPGA based microserver for high performance real-time computing in Adaptive Optics FPGA based microserver for high performance real-time computing in Adaptive Optics C. Patauner a, R. Biasi a, M. Andrighettoni a, G. Angerer a, D. Pescoller a, F. Porta a, D. Gratadour b a Microgate Srl,

More information

Computing Challenges in Adaptive Optics for the Thirty Meter Telescope. Corinne Boyer ICALEPCS Grenoble, France October 10, 2011

Computing Challenges in Adaptive Optics for the Thirty Meter Telescope. Corinne Boyer ICALEPCS Grenoble, France October 10, 2011 Computing Challenges in Adaptive Optics for the Thirty Meter Telescope Corinne Boyer ICALEPCS Grenoble, France October 10, 2011 1 This Talk Introduction to the Thirty Meter Telescope (TMT) Adaptive Optics

More information

Approach to Enable Real-Time HPC. President, CEO and Cofounder National Instruments

Approach to Enable Real-Time HPC. President, CEO and Cofounder National Instruments Applying a Graphical System Design Approach to Enable Real-Time HPC Dr. James Truchard President, CEO and Cofounder National Instruments National Instruments Leaders for 30 years in Computer-Based Measurement

More information

Wavefront Prediction using Artificial Neural Networks. By: Steve Weddell

Wavefront Prediction using Artificial Neural Networks. By: Steve Weddell Wavefront Prediction using Artificial Neural Networks By: Steve Weddell 1 Motivation for Research (1 of 2) Turbulence layer #1 Turbulence layer #2 Ground Turbulence Layer Science Object Reference Guide

More information

The SPARTA Platform: Design, Status and. Adaptive Optics Systems (ESO)

The SPARTA Platform: Design, Status and. Adaptive Optics Systems (ESO) The SPARTA Platform: Design, Status and Perspectives Marcos acossuárez Valles aes Adaptive Optics Systems (ESO) msuarez@eso.orgorg SPARTA Platform Targets ESO Standard Platform for Adaptive Optics Real-Time

More information

The AOLI low-order non-linear curvature wavefront sensor

The AOLI low-order non-linear curvature wavefront sensor The AOLI low-order non-linear curvature wavefront sensor A method for high sensitivity wavefront reconstruction Jonathan Crass Institute of Astronomy, University of Cambridge SPIE Astronomical Telescopes

More information

Gemini Observatory. Multi-Conjugate Adaptive Optics Control System. The Gemini MCAO System (ICALEPCS, Geneva, October 2005) 1

Gemini Observatory. Multi-Conjugate Adaptive Optics Control System. The Gemini MCAO System (ICALEPCS, Geneva, October 2005) 1 Gemini Observatory Multi-Conjugate Adaptive Optics Control System The Gemini MCAO System (ICALEPCS, Geneva, October 2005) 1 The Gemini MCAO System Andy Foster Observatory Sciences Ltd William James House,

More information

Studying GPU based RTC for TMT NFIRAOS

Studying GPU based RTC for TMT NFIRAOS Studying GPU based RTC for TMT NFIRAOS Lianqi Wang Thirty Meter Telescope Project RTC Workshop Dec 04, 2012 1 Outline Tomography with iterative algorithms on GPUs Matri vector multiply approach Assembling

More information

DESIGN AND TESTING OF GPU BASED RTC FOR TMT NFIRAOS

DESIGN AND TESTING OF GPU BASED RTC FOR TMT NFIRAOS Florence, Italy. Adaptive May 2013 Optics for Extremely Large Telescopes III ISBN: 978-88-908876-0-4 DOI: 10.12839/AO4ELT3.13172 DESIGN AND TESTING OF GPU BASED RTC FOR TMT NFIRAOS Lianqi Wang 1,a, 1 Thirty

More information

European Organization for Astronomical Research in the Southern

European Organization for Astronomical Research in the Southern State-of-the-art detector controller for ESO instruments Leander H. Mehrgan, Domingo Alvarez, Dietrich Baade, Claudio Cumani, Siegfried Eschbaumer, Gert Finger, Christoph Geimer, Derek Ives, Manfred Meyer,

More information

Fast End-to-End Multi-Conjugate AO Simulations Using Graphical Processing Units and the MAOS Simulation Code

Fast End-to-End Multi-Conjugate AO Simulations Using Graphical Processing Units and the MAOS Simulation Code Fast End-to-End Multi-Conjugate AO Simulations Using Graphical Processing Units and the MAOS Simulation Code Lianqi Wang 1a and Brent Ellerbroek 1 TMT Observatory Corportaion, 1111 South Arroyo Pkwy Suite

More information

Quasi-real-time end-to-end adaptive optics simulations at the E-ELT scale

Quasi-real-time end-to-end adaptive optics simulations at the E-ELT scale Quasi-real-time end-to-end adaptive optics simulations at the E-ELT scale Damien Gratadour 1a, Arnaud Sevin 1, Eric Gendron 1, and Gerard Rousset 1 Laboratoire d Etudes Spatiales et d Instrumentation en

More information

NEW WAVEFRONT SENSING CONCEPTS FOR ADAPTIVE OPTICS INSTRUMENTATION

NEW WAVEFRONT SENSING CONCEPTS FOR ADAPTIVE OPTICS INSTRUMENTATION NEW WAVEFRONT SENSING CONCEPTS FOR ADAPTIVE OPTICS INSTRUMENTATION K. El Hadi* a, M. Gray a, T. Fusco b, a, B. Le Roux a. a Aix-Marseille Université, CNRS, LAM (Laboratoire d Astrophysique de Marseille)

More information

ELT-scale real-time control on Intel Xeon Phi and many core CPUs

ELT-scale real-time control on Intel Xeon Phi and many core CPUs ELT-scale real-time control on Intel Xeon Phi and many core CPUs David R. Jenkins, Alastair G. Basden, and Richard M. Myers CfAI, Department of Physics, Durham University, DH1 3LE, UK ABSTRACT The next

More information

An FPGA-based High Speed Parallel Signal Processing System for Adaptive Optics Testbed

An FPGA-based High Speed Parallel Signal Processing System for Adaptive Optics Testbed An FPGA-based High Speed Parallel Signal Processing System for Adaptive Optics Testbed Hong Bong Kim 1 Hanwha Thales. Co., Ltd. Republic of Korea Young Soo Choi and Yu Kyung Yang Agency for Defense Development,

More information

Maximizing heterogeneous system performance with ARM interconnect and CCIX

Maximizing heterogeneous system performance with ARM interconnect and CCIX Maximizing heterogeneous system performance with ARM interconnect and CCIX Neil Parris, Director of product marketing Systems and software group, ARM Teratec June 2017 Intelligent flexible cloud to enable

More information

40Gbps+ Full Line Rate, Programmable Network Accelerators for Low Latency Applications SAAHPC 19 th July 2011

40Gbps+ Full Line Rate, Programmable Network Accelerators for Low Latency Applications SAAHPC 19 th July 2011 40Gbps+ Full Line Rate, Programmable Network Accelerators for Low Latency Applications SAAHPC 19 th July 2011 Allan Cantle President & Founder www.nallatech.com Company Overview ISI + Nallatech + Innovative

More information

Israel, 24 November Reaching New Heights in Astronomy

Israel, 24 November Reaching New Heights in Astronomy Reaching New Heights in Astronomy ESO and Industry Construction of world-class observatories Buildings, telescopes, instruments Only possible with strong involvement of industry Procurement by competitive

More information

UCLA Adaptive Optics for Extremely Large Telescopes 4 Conference Proceedings

UCLA Adaptive Optics for Extremely Large Telescopes 4 Conference Proceedings UCLA Adaptive Optics for Extremely Large Telescopes 4 Conference Proceedings Title Simulations of AO for the E-ELT and its instruments Permalink https://escholarship.org/uc/item/7kh262xf Journal Adaptive

More information

STATE OF THE ART ADAPTIVE OPTICS. Philippe Feautrier WAVEFRONT SENSOR CAMERAS AT FIRST LIGHT IMAGING.

STATE OF THE ART ADAPTIVE OPTICS. Philippe Feautrier WAVEFRONT SENSOR CAMERAS AT FIRST LIGHT IMAGING. STATE OF THE ART ADAPTIVE OPTICS WAVEFRONT SENSOR CAMERAS AT FIRST LIGHT IMAGING Philippe Feautrier philippe.feautrier@firstlight.fr LBTO UM Vis and IR WFS cameras at FLI 1 First Light Imaging: our origins

More information

Down selecting suitable manycore technologies for the ELT AO RTC. David Barr, Alastair Basden, Nigel Dipper and Noah Schwartz

Down selecting suitable manycore technologies for the ELT AO RTC. David Barr, Alastair Basden, Nigel Dipper and Noah Schwartz Down selecting suitable manycore technologies for the ELT AO RTC David Barr, Alastair Basden, Nigel Dipper and Noah Schwartz GFLOPS RTC for AO workshop 27/01/2016 AO RTC Complexity 1.E+05 1.E+04 E-ELT

More information

DCS-ctrl: A Fast and Flexible Device-Control Mechanism for Device-Centric Server Architecture

DCS-ctrl: A Fast and Flexible Device-Control Mechanism for Device-Centric Server Architecture DCS-ctrl: A Fast and Flexible ice-control Mechanism for ice-centric Server Architecture Dongup Kwon 1, Jaehyung Ahn 2, Dongju Chae 2, Mohammadamin Ajdari 2, Jaewon Lee 1, Suheon Bae 1, Youngsok Kim 1,

More information

27 March 2018 Mikael Arguedas and Morgan Quigley

27 March 2018 Mikael Arguedas and Morgan Quigley 27 March 2018 Mikael Arguedas and Morgan Quigley Separate devices: (prototypes 0-3) Unified camera: (prototypes 4-5) Unified system: (prototypes 6+) USB3 USB Host USB3 USB2 USB3 USB Host PCIe root

More information

The CAFADIS camera: a new tomographic wavefront sensor for Adaptive Optics.

The CAFADIS camera: a new tomographic wavefront sensor for Adaptive Optics. 1st AO4ELT conference, 05011 (2010) DOI:10.1051/ao4elt/201005011 Owned by the authors, published by EDP Sciences, 2010 The CAFADIS camera: a new tomographic wavefront sensor for Adaptive Optics. J.M. Rodríguez-Ramos

More information

Performance of Monte-Carlo simulation of adaptive optics systems of the EAGLE multi-ifu Instrument for E-ELT

Performance of Monte-Carlo simulation of adaptive optics systems of the EAGLE multi-ifu Instrument for E-ELT Performance of Monte-Carlo simulation of adaptive optics systems of the EAGLE multi-ifu Instrument for E-ELT Alastair G. Basden a, Timothy Butterley a, Mark A. Harrison a, Timothy J. Morris a, Richard

More information

Closed loop Optical Integrated Modeling

Closed loop Optical Integrated Modeling Closed loop Optical Integrated Modeling R. Conan a, G. Angeli a, A. Bouchez a, K. Das a, B. Irrarazaval a, B. McLeod b, F. Quiros-Pacheco a, and D. Schwartz a a GMTO, 465 N. Halstead Avenue, Pasadana,

More information

Enyx soft-hardware design services and development framework for FPGA & SoC

Enyx soft-hardware design services and development framework for FPGA & SoC soft-hardware design services and development framework for FPGA & SoC Smart NIC Smart Switch Your custom hardware hardware acceleration experts 3rd party IP Cores AXI ARM DMA CPU Your own soft-hardware

More information

OCP Engineering Workshop - Telco

OCP Engineering Workshop - Telco OCP Engineering Workshop - Telco Low Latency Mobile Edge Computing Trevor Hiatt Product Management, IDT IDT Company Overview Founded 1980 Workforce Approximately 1,800 employees Headquarters San Jose,

More information

RDMA in Embedded Fabrics

RDMA in Embedded Fabrics RDMA in Embedded Fabrics Ken Cain, kcain@mc.com Mercury Computer Systems 06 April 2011 www.openfabrics.org 2011 Mercury Computer Systems, Inc. www.mc.com Uncontrolled for Export Purposes 1 Outline Embedded

More information

Shack-Hartmann tomographic wavefront reconstruction using LGS: analysis of spot elongation and fratricide effect

Shack-Hartmann tomographic wavefront reconstruction using LGS: analysis of spot elongation and fratricide effect 1st AO4ELT conference, 05010 (2010) DOI:10.1051/ao4elt/201005010 Owned by the authors, published by EDP Sciences, 2010 Shack-Hartmann tomographic wavefront reconstruction using LGS: analysis of spot elongation

More information

Building the Most Efficient Machine Learning System

Building the Most Efficient Machine Learning System Building the Most Efficient Machine Learning System Mellanox The Artificial Intelligence Interconnect Company June 2017 Mellanox Overview Company Headquarters Yokneam, Israel Sunnyvale, California Worldwide

More information

Gen-Z Memory-Driven Computing

Gen-Z Memory-Driven Computing Gen-Z Memory-Driven Computing Our vision for the future of computing Patrick Demichel Distinguished Technologist Explosive growth of data More Data Need answers FAST! Value of Analyzed Data 2005 0.1ZB

More information

Computer simulations and real-time control of ELT AO systems using graphical processing units

Computer simulations and real-time control of ELT AO systems using graphical processing units Invited Paper Computer simulations and real-time control of ELT AO systems using graphical processing units Lianqi Wang, Brent Ellerbroek Thirty Meter Telescope Project, 1111 S. Arroyo Pkwy, Suite 200,

More information

A real-time simulation facility for astronomical adaptive optics

A real-time simulation facility for astronomical adaptive optics Advance Access publication 2014 February 18 doi:10.1093/mnras/stu143 A real-time simulation facility for astronomical adaptive optics Alastair Basden Department of Physics, South Road, Durham DH1 3LE,

More information

Simulations of Laser Guide Star Adaptive Optics Systems for the European Extremely Large Telescope

Simulations of Laser Guide Star Adaptive Optics Systems for the European Extremely Large Telescope st AO4ELT conference, 03005 (200) DOI:0.05/ao4elt/20003005 Owned by the authors, published by EDP Sciences, 200 Simulations of Laser Guide Star Adaptive Optics Systems for the European Extremely Large

More information

VST Project. Test Procedure in Europe. Guiding / ADC. Doc. no. : VST-PRO-OAC-AG-001. Date: Issue: 1.0 DRAFT VERSION

VST Project. Test Procedure in Europe. Guiding / ADC. Doc. no. : VST-PRO-OAC-AG-001. Date: Issue: 1.0 DRAFT VERSION Pag. 1 of 18 VST Project Test Procedure in Europe Guiding / ADC Doc. no. : VST-PRO-OAC-AG-001 Date: 2006-10-09 Issue: 1.0 DRAFT VERSION Name Date Signature Written by M. Brescia, P. Schipani 2006-10-09

More information

FPGA Implementation of RDMA-Based Data Acquisition System Over 100 GbE

FPGA Implementation of RDMA-Based Data Acquisition System Over 100 GbE 1 FPGA Implementation of RDMA-Based Data Acquisition System Over 100 GbE Wassim Mansour, Member, IEEE, Nicolas Janvier, Member, IEEE, and Pablo Fajardo Abstract This paper presents an RDMA over Ethernet

More information

ICD 2.3/4.3. Wavefront Correction Control System to Data Handling System

ICD 2.3/4.3. Wavefront Correction Control System to Data Handling System ICD 2.3/4.3 Wavefront Correction Control System to Data Handling System Version: Issued By: Erik Johansson, Keith Cummings, Kit Richards, Luke Johnson Draft 19 December 2014 Wavefront Correction Group

More information

Status of PSF Reconstruction at Lick

Status of PSF Reconstruction at Lick Status of PSF Reconstruction at Lick Mike Fitzgerald Workshop on AO PSF Reconstruction May 10-12, 2004 Quick Outline Recap Lick AO system's features Reconstruction approach Implementation issues Calibration

More information

Computing Infrastructure for Online Monitoring and Control of High-throughput DAQ Electronics

Computing Infrastructure for Online Monitoring and Control of High-throughput DAQ Electronics Computing Infrastructure for Online Monitoring and Control of High-throughput DAQ S. Chilingaryan, M. Caselle, T. Dritschler, T. Farago, A. Kopmann, U. Stevanovic, M. Vogelgesang Hardware, Software, and

More information

Chelsio Communications. Meeting Today s Datacenter Challenges. Produced by Tabor Custom Publishing in conjunction with: CUSTOM PUBLISHING

Chelsio Communications. Meeting Today s Datacenter Challenges. Produced by Tabor Custom Publishing in conjunction with: CUSTOM PUBLISHING Meeting Today s Datacenter Challenges Produced by Tabor Custom Publishing in conjunction with: 1 Introduction In this era of Big Data, today s HPC systems are faced with unprecedented growth in the complexity

More information

Paving the Road to Exascale

Paving the Road to Exascale Paving the Road to Exascale Gilad Shainer August 2015, MVAPICH User Group (MUG) Meeting The Ever Growing Demand for Performance Performance Terascale Petascale Exascale 1 st Roadrunner 2000 2005 2010 2015

More information

Interconnect Your Future

Interconnect Your Future #OpenPOWERSummit Interconnect Your Future Scot Schultz, Director HPC / Technical Computing Mellanox Technologies OpenPOWER Summit, San Jose CA March 2015 One-Generation Lead over the Competition Mellanox

More information

Intel Research mote. Ralph Kling Intel Corporation Research Santa Clara, CA

Intel Research mote. Ralph Kling Intel Corporation Research Santa Clara, CA Intel Research mote Ralph Kling Intel Corporation Research Santa Clara, CA Overview Intel mote project goals Project status and direction Intel mote hardware Intel mote software Summary and outlook Intel

More information

INDUSTRY-LED COLLABORATION

INDUSTRY-LED COLLABORATION INDUSTRY-LED COLLABORATION EUREKA Instruments EUREKA Innovation Days 24 May 2018 Zeynep Sarılar InterCluster spokesperson & ITEA Chairwoman Full members 41 full members (40 countries + European Commission)

More information

Realizing the Next Generation of Exabyte-scale Persistent Memory-Centric Architectures and Memory Fabrics

Realizing the Next Generation of Exabyte-scale Persistent Memory-Centric Architectures and Memory Fabrics Realizing the Next Generation of Exabyte-scale Persistent Memory-Centric Architectures and Memory Fabrics Zvonimir Z. Bandic, Sr. Director, Next Generation Platform Technologies Western Digital Corporation

More information

Matrox Imaging White Paper

Matrox Imaging White Paper Reliable high bandwidth video capture with Matrox Radient Abstract The constant drive for greater analysis resolution and higher system throughput results in the design of vision systems with multiple

More information

Designing, developing, debugging ARM Cortex-A and Cortex-M heterogeneous multi-processor systems

Designing, developing, debugging ARM Cortex-A and Cortex-M heterogeneous multi-processor systems Designing, developing, debugging ARM and heterogeneous multi-processor systems Kinjal Dave Senior Product Manager, ARM ARM Tech Symposia India December 7 th 2016 Topics Introduction System design Software

More information

A framework for optimizing OpenVX Applications on Embedded Many Core Accelerators

A framework for optimizing OpenVX Applications on Embedded Many Core Accelerators A framework for optimizing OpenVX Applications on Embedded Many Core Accelerators Giuseppe Tagliavini, DEI University of Bologna Germain Haugou, IIS ETHZ Andrea Marongiu, DEI University of Bologna & IIS

More information

Introduction Technology Equipment Performance Current developments Conclusions. White Rabbit. A quick introduction. Javier Serrano

Introduction Technology Equipment Performance Current developments Conclusions. White Rabbit. A quick introduction. Javier Serrano White Rabbit A quick introduction Javier Serrano CERN BE-CO Hardware and Timing section ICALEPCS pre-conference workshop Barcelona, 7 October 2017 Javier Serrano Introduction to White Rabbit 1/29 Outline

More information

EXPLOITING ACCELERATOR-BASED HPC FOR ARMY APPLICATIONS

EXPLOITING ACCELERATOR-BASED HPC FOR ARMY APPLICATIONS EXPLOITING ACCELERATOR-BASED HPC FOR ARMY APPLICATIONS James Ross High Performance Technologies, Inc (HPTi) Computational Scientist Edward Carmack David Richie Song Park, Brian Henz and Dale Shires HPTi

More information

Price list. November Price non EU (exvat) / EU in Euros

Price list. November Price non EU (exvat) / EU in Euros Price list November 2017 Price (exvat) / EU in Euros 1 Astronomical instruments metrology services The price included optical alignment and a detailed report with the following information: - PTV and RMS

More information

Building the Most Efficient Machine Learning System

Building the Most Efficient Machine Learning System Building the Most Efficient Machine Learning System Mellanox The Artificial Intelligence Interconnect Company June 2017 Mellanox Overview Company Headquarters Yokneam, Israel Sunnyvale, California Worldwide

More information

Toward a Memory-centric Architecture

Toward a Memory-centric Architecture Toward a Memory-centric Architecture Martin Fink EVP & Chief Technology Officer Western Digital Corporation August 8, 2017 1 SAFE HARBOR DISCLAIMERS Forward-Looking Statements This presentation contains

More information

Parallel Stochastic Gradient Descent: The case for native GPU-side GPI

Parallel Stochastic Gradient Descent: The case for native GPU-side GPI Parallel Stochastic Gradient Descent: The case for native GPU-side GPI J. Keuper Competence Center High Performance Computing Fraunhofer ITWM, Kaiserslautern, Germany Mark Silberstein Accelerated Computer

More information

A Wave Optics Propagation Code for Multi-Conjugate Adaptive Optics. B. L. Ellerbroek Gemini Observatory, 670 N. A'ohoku Place, Hilo HI USA

A Wave Optics Propagation Code for Multi-Conjugate Adaptive Optics. B. L. Ellerbroek Gemini Observatory, 670 N. A'ohoku Place, Hilo HI USA A Wave Optics Propagation Code for Multi-Conjugate Adaptive Optics B. L. Ellerbroek Gemini Observatory, 67 N. A'ohoku Place, Hilo HI 9672 USA Gemini Preprint #69 A Wave Optics Propagation Code for Multi-Conjugate

More information

Trajectory handling for LINC-NIRVANA s Field Derotators

Trajectory handling for LINC-NIRVANA s Field Derotators Trajectory handling for LINC-NIRVANA s Field Derotators Frank Kittmann 1a, Thomas Bertram 1, Matthew Horrobin 2, Albert Conrad 1, Jan Trowitzsch 1, Florian Briegel 1, Jürgen Berwein 1, and Lars Mohr 1

More information

TECHNICAL OVERVIEW ACCELERATED COMPUTING AND THE DEMOCRATIZATION OF SUPERCOMPUTING

TECHNICAL OVERVIEW ACCELERATED COMPUTING AND THE DEMOCRATIZATION OF SUPERCOMPUTING TECHNICAL OVERVIEW ACCELERATED COMPUTING AND THE DEMOCRATIZATION OF SUPERCOMPUTING Table of Contents: The Accelerated Data Center Optimizing Data Center Productivity Same Throughput with Fewer Server Nodes

More information

SysML for Telescope System Modeling

SysML for Telescope System Modeling by the INCOSE MBSE Challenge Team SE^2 Presented to the LA chapter of INCOSE, February 2 nd 2010 page 1 Agenda What is SE^2 What is ESO? What is the Challenge project about? The deliverables What have

More information

EMBEDDED VISION AND 3D SENSORS: WHAT IT MEANS TO BE SMART

EMBEDDED VISION AND 3D SENSORS: WHAT IT MEANS TO BE SMART EMBEDDED VISION AND 3D SENSORS: WHAT IT MEANS TO BE SMART INTRODUCTION Adding embedded processing to simple sensors can make them smart but that is just the beginning of the story. Fixed Sensor Design

More information

Superior AO Technology Solving Today s Most Demanding Optical Problems

Superior AO Technology Solving Today s Most Demanding Optical Problems Superior AO Technology Solving Today s Most Demanding Optical Problems AFFORDABLE ADAPTIVE OPTICS Clarifi A closed loop adaptive optics system known as Clarifi senses your beam conditions and adjusts the

More information

Evaluating On-Node GPU Interconnects for Deep Learning Workloads

Evaluating On-Node GPU Interconnects for Deep Learning Workloads Evaluating On-Node GPU Interconnects for Deep Learning Workloads NATHAN TALLENT, NITIN GAWANDE, CHARLES SIEGEL ABHINAV VISHNU, ADOLFY HOISIE Pacific Northwest National Lab PMBS 217 (@ SC) November 13,

More information

Telescope Wavefront Errors

Telescope Wavefront Errors Telescope Wavefront Errors Henri Bonnet 1 Tasks of WFC at E-ELT Help System Engineering develop and maintain the technical budgets Develop Control Strategy Define WFC I/F to instruments. 2 How we do it

More information

Winter College on Optics in Environmental Science February Adaptive Optics: Introduction, and Wavefront Correction

Winter College on Optics in Environmental Science February Adaptive Optics: Introduction, and Wavefront Correction 2018-23 Winter College on Optics in Environmental Science 2-18 February 2009 Adaptive Optics: Introduction, and Wavefront Correction Love G. University of Durham U.K. Adaptive Optics: Gordon D. Love Durham

More information

Summary of Data Management Principles

Summary of Data Management Principles Large Synoptic Survey Telescope (LSST) Summary of Data Management Principles Steven M. Kahn LPM-151 Latest Revision: June 30, 2015 Change Record Version Date Description Owner name 1 6/30/2015 Initial

More information

arxiv: v1 [astro-ph.im] 2 Jul 2018

arxiv: v1 [astro-ph.im] 2 Jul 2018 Scalable Platform for Adaptive optics Real-time Control (SPARC) Part 2: Field Programmable Gate Array (FPGA) implementation and performance Avinash Surendran a,*, Mahesh P. Burse b, A. N. Ramaprakash b,

More information

Reducing adaptive optics latency using Xeon Phi many-core processors

Reducing adaptive optics latency using Xeon Phi many-core processors doi:10.1093/mnras/stv1813 Reducing adaptive optics latency using Xeon Phi many-core processors David Barr, 1,2 Alastair Basden, 3 Nigel Dipper 3 andnoahschwartz 1 1 UK Astronomy Technology Centre, Royal

More information

OpenCAPI Technology. Myron Slota Speaker name, Title OpenCAPI Consortium Company/Organization Name. Join the Conversation #OpenPOWERSummit

OpenCAPI Technology. Myron Slota Speaker name, Title OpenCAPI Consortium Company/Organization Name. Join the Conversation #OpenPOWERSummit OpenCAPI Technology Myron Slota Speaker name, Title OpenCAPI Consortium Company/Organization Name Join the Conversation #OpenPOWERSummit Industry Collaboration and Innovation OpenCAPI Topics Computation

More information

RapidIO.org Update. Mar RapidIO.org 1

RapidIO.org Update. Mar RapidIO.org 1 RapidIO.org Update rickoco@rapidio.org Mar 2015 2015 RapidIO.org 1 Outline RapidIO Overview & Markets Data Center & HPC Communications Infrastructure Industrial Automation Military & Aerospace RapidIO.org

More information

CafeGPI. Single-Sided Communication for Scalable Deep Learning

CafeGPI. Single-Sided Communication for Scalable Deep Learning CafeGPI Single-Sided Communication for Scalable Deep Learning Janis Keuper itwm.fraunhofer.de/ml Competence Center High Performance Computing Fraunhofer ITWM, Kaiserslautern, Germany Deep Neural Networks

More information

Building NVLink for Developers

Building NVLink for Developers Building NVLink for Developers Unleashing programmatic, architectural and performance capabilities for accelerated computing Why NVLink TM? Simpler, Better and Faster Simplified Programming No specialized

More information

Functional Specification

Functional Specification EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH ORGANISATION EUROPEENE POUR LA RECHERCHE NUCLEAIRE White Rabbit Switch Functional Specification Version: 0.c Date: September 6 of 2010. Author: J. Gabriel Ramírez

More information

Parallel waveform extraction algorithms for the Cherenkov Telescope Array Real-Time Analysis

Parallel waveform extraction algorithms for the Cherenkov Telescope Array Real-Time Analysis Parallel waveform extraction algorithms for the Cherenkov Telescope Array Real-Time Analysis, a, Andrea Bulgarelli a, Adriano De Rosa a, Alessio Aboudan a, Valentina Fioretti a, Giovanni De Cesare a, Ramin

More information

Embedded Hardware and Software

Embedded Hardware and Software Embedded Hardware and Software Saved by a Common Language? Nithya A. Ruff, Director, Product Marketing 10/11/2012, Toronto Synopsys 2012 1 Synopsys Industry Leadership $1,800 $1,600 $1,400 $1,200 $1,000

More information

Computer and Machine Vision

Computer and Machine Vision Computer and Machine Vision Lecture Week 12 Part-2 Additional 3D Scene Considerations March 29, 2014 Sam Siewert Outline of Week 12 Computer Vision APIs and Languages Alternatives to C++ and OpenCV API

More information

Progress Towards Low-Cost Compact Metric Adaptive Optics Systems

Progress Towards Low-Cost Compact Metric Adaptive Optics Systems Progress Towards Low-Cost Compact Metric Adaptive Optics Systems Justin D. Mansell, Brian Henderson, Brennen Wiesner, Robert Praus, and Steve Coy Active Optical Systems, LLC www.aos-llc.com 1 Outline Introduction

More information

New ARMv8-R technology for real-time control in safetyrelated

New ARMv8-R technology for real-time control in safetyrelated New ARMv8-R technology for real-time control in safetyrelated applications James Scobie Product manager ARM Technical Symposium China: Automotive, Industrial & Functional Safety October 31 st 2016 November

More information

Comparison of Reconstruction and Control algorithms on the ESO end-to-end simulator OCTOPUS

Comparison of Reconstruction and Control algorithms on the ESO end-to-end simulator OCTOPUS st AO4ELT conference, 32 (2) DOI:.5/ao4elt/232 Owned by the authors, published by EDP Sciences, 2 Comparison of Reconstruction and Control algorithms on the ESO end-to-end simulator OCTOPUS I. Montilla,a,

More information

Computer Vision on Tegra K1. Chen Sagiv SagivTech Ltd.

Computer Vision on Tegra K1. Chen Sagiv SagivTech Ltd. Computer Vision on Tegra K1 Chen Sagiv SagivTech Ltd. Established in 2009 and headquartered in Israel Core domain expertise: GPU Computing and Computer Vision What we do: - Technology - Solutions - Projects

More information

MODERN DIMENSIONAL MEASURING TECHNIQUES BASED ON OPTICAL PRINCIPLES

MODERN DIMENSIONAL MEASURING TECHNIQUES BASED ON OPTICAL PRINCIPLES MODERN DIMENSIONAL MEASURING TECHNIQUES BASED ON OPTICAL PRINCIPLES J. Reichweger 1, J. Enzendorfer 1 and E. Müller 2 1 Steyr Daimler Puch Engineering Center Steyr GmbH Schönauerstrasse 5, A-4400 Steyr,

More information

Interconnection Network for Tightly Coupled Accelerators Architecture

Interconnection Network for Tightly Coupled Accelerators Architecture Interconnection Network for Tightly Coupled Accelerators Architecture Toshihiro Hanawa, Yuetsu Kodama, Taisuke Boku, Mitsuhisa Sato Center for Computational Sciences University of Tsukuba, Japan 1 What

More information

VPI / InfiniBand. Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability

VPI / InfiniBand. Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability VPI / InfiniBand Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability Mellanox enables the highest data center performance with its

More information

The Plenoptic Camera as Wavefront Sensor for the VTT Solar Telescope

The Plenoptic Camera as Wavefront Sensor for the VTT Solar Telescope The Plenoptic Camera as Wavefront Sensor for the VTT Solar Telescope Noelia Martínez a, Luis Fernando Rodríguez Ramos a, Luz María Montoya a, Iciar Montilla a, and Manuel Collados a a Instituto de Astrofísica

More information

HA-PACS/TCA: Tightly Coupled Accelerators for Low-Latency Communication between GPUs

HA-PACS/TCA: Tightly Coupled Accelerators for Low-Latency Communication between GPUs HA-PACS/TCA: Tightly Coupled Accelerators for Low-Latency Communication between GPUs Yuetsu Kodama Division of High Performance Computing Systems Center for Computational Sciences University of Tsukuba,

More information

Exploring System Coherency and Maximizing Performance of Mobile Memory Systems

Exploring System Coherency and Maximizing Performance of Mobile Memory Systems Exploring System Coherency and Maximizing Performance of Mobile Memory Systems Shanghai: William Orme, Strategic Marketing Manager of SSG Beijing & Shenzhen: Mayank Sharma, Product Manager of SSG ARM Tech

More information

Are you looking for ultrafast and extremely precise stereovision technology for industrial applications? Learn about

Are you looking for ultrafast and extremely precise stereovision technology for industrial applications? Learn about Edition November 2017 Image sensors and vision systems, Smart Industries, imec.engineering Are you looking for ultrafast and extremely precise stereovision technology for industrial applications? Learn

More information

Optimizing Out-of-Core Nearest Neighbor Problems on Multi-GPU Systems Using NVLink

Optimizing Out-of-Core Nearest Neighbor Problems on Multi-GPU Systems Using NVLink Optimizing Out-of-Core Nearest Neighbor Problems on Multi-GPU Systems Using NVLink Rajesh Bordawekar IBM T. J. Watson Research Center bordaw@us.ibm.com Pidad D Souza IBM Systems pidsouza@in.ibm.com 1 Outline

More information

The EuroHPC strategic initiative

The EuroHPC strategic initiative Amsterdam, 12 December 2017 The EuroHPC strategic initiative Thomas Skordas Director, DG CONNECT-C, European Commission The European HPC strategy in Horizon 2020 Infrastructure Capacity of acquiring leadership-class

More information

SILECS Super Infrastructure for Large-scale Experimental Computer Science

SILECS Super Infrastructure for Large-scale Experimental Computer Science Super Infrastructure for Large-scale Experimental Computer Science Serge Fdida (UPMC) Frédéric Desprez (Inria) Christian Perez (Inria) INRIA, CNRS, RENATER, CEA, CPU, CDEFI, IMT, Sorbonne Universite, Universite

More information

in Action Fujitsu High Performance Computing Ecosystem Human Centric Innovation Innovation Flexibility Simplicity

in Action Fujitsu High Performance Computing Ecosystem Human Centric Innovation Innovation Flexibility Simplicity Fujitsu High Performance Computing Ecosystem Human Centric Innovation in Action Dr. Pierre Lagier Chief Technology Officer Fujitsu Systems Europe Innovation Flexibility Simplicity INTERNAL USE ONLY 0 Copyright

More information

Path to Exascale? Intel in Research and HPC 2012

Path to Exascale? Intel in Research and HPC 2012 Path to Exascale? Intel in Research and HPC 2012 Intel s Investment in Manufacturing New Capacity for 14nm and Beyond D1X Oregon Development Fab Fab 42 Arizona High Volume Fab 22nm Fab Upgrades D1D Oregon

More information

OpenCAPI and its Roadmap

OpenCAPI and its Roadmap OpenCAPI and its Roadmap Myron Slota, President OpenCAPI Speaker name, Consortium Title Company/Organization Name Join the Conversation #OpenPOWERSummit Industry Collaboration and Innovation OpenCAPI and

More information

vs. GPU Performance Without the Answer University of Virginia Computer Engineering g Labs

vs. GPU Performance Without the Answer University of Virginia Computer Engineering g Labs Where is the Data? Why you Cannot Debate CPU vs. GPU Performance Without the Answer Chris Gregg and Kim Hazelwood University of Virginia Computer Engineering g Labs 1 GPUs and Data Transfer GPU computing

More information

Laser Beacon Tracking for High-Accuracy Attitude Determination

Laser Beacon Tracking for High-Accuracy Attitude Determination Laser Beacon Tracking for High-Accuracy Attitude Determination Tam Nguyen Massachusetts Institute of Technology 29 th AIAA/USU Conference on Small Satellites SSC15-VIII-2 08/12/2015 Outline Motivation

More information

HPC future trends from a science perspective

HPC future trends from a science perspective HPC future trends from a science perspective Simon McIntosh-Smith University of Bristol HPC Research Group simonm@cs.bris.ac.uk 1 Business as usual? We've all got used to new machines being relatively

More information