TetriSched: global rescheduling with adaptive plan-ahead in dynamic heterogeneous clusters
|
|
|
- Barbara Shelton
- 5 years ago
- Views:
Transcription
1 TetriSched: global rescheduling with adaptive plan-ahead in dynamic heterogeneous clusters Alexey Tumanov Timothy Zhu, Jun Woo Park, Michael Kozuch, Mor Harchol-Balter, Gregory R. Ganger PARALLEL DATA LABORATORY Carnegie Mellon University
2 Motivation Spark Analytics (by 5pm) 100xCPU1 80min -OR- 100xCPU2 40min Deep Learning (by 3pm) 10xGPU 20min -OR- 100xCPU2 40min Long Running 10Gbps backbone FPGA FPGA Heterogeneous Cluster 2 FPGA FPGA
3 Heterogeneity Amplifies Options rack1 rack2 m4 m3 m2 m1 time Anti-Affinity(BE) MPI (by t=3) GPU (by t=3) rack1 rack2 m4 m3 m2 m1 rack1 rack2 m4 m3 m2 m1 3
4 Exploiting Time Flexibility (Plan-ahead) Previously: just look at current state rack1 rack2 time 4
5 Exploiting Time Flexibility (Plan-ahead) Previously: just look at current state Plan-ahead: estimate runtimes and choices rack1 rack2 time 5
6 Exploiting Time Flexibility (Plan-ahead) Previously: just look at current state Plan-ahead: estimate runtimes and choices should this job wait for better placement? rack1 rack2 time 6
7 Key Challenges Leverage runtime estimates robustly Express combinatorially many options succinctly Including quantifying their relative merit Exploit this knowledge to improve allocation All in a practical manner 7
8 Quantify: Internal Utility Functions SLO jobs: zero value after deadline BE jobs: lower value for longer completion time Higher value for more important jobs (U > u) U SLO job deadline u best-effort job utility completion time 8
9 Express: Space-Time Request Language Utility u(p,t): placement à scalar value u n Choose k (nck) k=2, s=1, d=2, u nck ( m rack1, ) i n à refers to a group of nodes to choose from k à how many nodes to choose nck ( m, ) i k=2, s=0, d=4, u/2 rack1 rack2 m4 m3 m2 m1 time 9
10 Express: Space-Time Request Language Utility u(p,t): placement à scalar value u nck ( m rack1, k=2, s=1, d=2, u ) i rack1 rack2 m4 m3 m2 m1 time value 10 u 3
11 Express: Space-Time Request Language Utility u(p,t): placement à scalar value u nck ( m rack1, ) i k=2, s=1, d=2, u nck ( m, ) i k=2, s=0, d=4, u/2 rack1 rack2 m4 m3 m2 m1 time value 11 u/2 4
12 Express: Space-Time Request Language Utility u(p,t): placement à scalar value u nck ( m rack1, ) i k=2, s=1, d=2, u nck ( m, ) i k=2, s=0, d=4, u/2 rack1 rack2 m4 m3 m2 m1 time value 12 u u/2 3 4
13 STRL Expression Composition nck ( m rack1, k=2, s=1, d=2, u ) i max OR nck ( m m rack2,, ) i i k=2, s=1, s=0, d=2, d=4, u/2u nck ( m, k=2, s=0, d=4, u/2) i 13
14 TetriSched System Architecture resources jobs framework framework framework Perforator Reservation Reservation System YARN ResAlloc YARN ProxyScheduler CapacityScheduler Job schedule and placement Resource request Runtime, deadline STRL Generator Framework Plugins TetriSched Scheduler Core objective function supply/demand constraints MPI GPU HA... nck nck min max nck nck Resources MILP solver Time MILP STRL Compiler MILP max sum min 14
15 TetriSched System Architecture resources jobs framework framework framework Perforator Reservation Reservation System ResAlloc YARN ProxyScheduler CapacityScheduler Job schedule and placement Resource request Runtime, deadline STRL Generator Framework Plugins TetriSched Scheduler Core objective function supply/demand constraints MPI GPU HA... nck nck min max nck nck Resources MILP solver Time MILP STRL Compiler MILP max sum min 15
16 TetriSched Scheduler Core Compile to MILP Aggregate sum max min Global scheduling Adaptive plan-ahead Solve MILP Cache Results Interpret Results 16
17 Experimental Results Real cluster: 256 nodes (8 racks, 32 nodes/rack) Workloads: Production derived: Facebook SLO + Yahoo BE job mix Systems compared: Rayon / CapacityScheduler Rayon / TetriSched Takeaway: With global rescheduling and plan-ahead, TetriSched outperforms Rayon/CapacityScheduler stack 17
18 Leverage runtime estimates robustly 6/2 AttaLnment(%) ayRn/C6 7etrL6ched (stlmate (rrrr(%) 18
19 Leverage runtime estimates robustly 6/2 AttaLnment(%) 5ayRn/C6 7etrL6ched Achieves high SLO attainment with 2x error (stlmate (rrrr(%) 19
20 Leverage runtime estimates robustly 0ean /atency(s) ayRn/C6 3x TetrL6ched 5x (stlmate (rrrr(%) 20
21 Leverage runtime estimates robustly 0ean /atency(s) 5ayRn/C6 TetrL6ched x x Achieves lower best-effort latency (stlmate (rrrr(%) 21
22 Leverage runtime estimates robustly 6/2 AttaLnment(%) ayRn/C6 7etrL6ched 0ean /atency(s) (stlmate (rrrr(%) 5ayRn/C6 TetrL6ched (stlmate (rrrr(%) Exploits runtime estimates better & more robustly 22
23 Benefit from plan-ahead and global 6/2 AttaLnPent(%) ayRn/C6 7etrL6cheG 7etrL6cheG-1G Oan-aheaG(s) 23
24 Benefit from plan-ahead and global 6/2 AttaLnPent(%) ayRn/C6 7etrL6cheG Oan-aheaG(s) 7etrL6cheG-1G Plan-ahead adds 2.5x in SLO attainment Global scheduling further improves perf. 24
25 Takeaway Results Each primary TetriSched feature is needed: Soft constraint support yields 2x over baseline Plan-ahead support yields 2.5x over baseline Global scheduling yields 40% over baseline Scales to sizeable clusters 256 node real cluster 1000-node simulated cluster 25
26 Conclusions Modern clusters induce complex scheduling tradeoffs Scheduling is hard must schedule harder! TetriSched: General support for space-time preferences (STRL) Leveraging runtime estimates robustly (plan-ahead) Global scheduling (STRL+MILP) Key takeaway result: Significantly higher SLO attainment, lower latency 26
Asymmetry-aware execution placement on manycore chips
Asymmetry-aware execution placement on manycore chips Alexey Tumanov Joshua Wise, Onur Mutlu, Greg Ganger CARNEGIE MELLON UNIVERSITY Introduction: Core Scaling? Moore s Law continues: can still fit more
Dynamic Resource Allocation for Distributed Dataflows. Lauritz Thamsen Technische Universität Berlin
Dynamic Resource Allocation for Distributed Dataflows Lauritz Thamsen Technische Universität Berlin 04.05.2018 Distributed Dataflows E.g. MapReduce, SCOPE, Spark, and Flink Used for scalable processing
S-Store: Streaming Meets Transaction Processing
S-Store: Streaming Meets Transaction Processing H-Store is an experimental database management system (DBMS) designed for online transaction processing applications Manasa Vallamkondu Motivation Reducing
Building A Data-center Scale Analytics Platform. Sriram Rao Scientist/Manager, CISL
Building A Data-center Scale Analytics Platform Sriram Rao Scientist/Manager, CISL CISL: Cloud and Information Services Lab Started in May 2012 Mission Statement: Applied research lab working on Systems
Warehouse- Scale Computing and the BDAS Stack
Warehouse- Scale Computing and the BDAS Stack Ion Stoica UC Berkeley UC BERKELEY Overview Workloads Hardware trends and implications in modern datacenters BDAS stack What is Big Data used For? Reports,
RobinHood: Tail Latency-Aware Caching Dynamically Reallocating from Cache-Rich to Cache-Poor
RobinHood: Tail Latency-Aware Caching Dynamically Reallocating from -Rich to -Poor Daniel S. Berger (CMU) Joint work with: Benjamin Berg (CMU), Timothy Zhu (PennState), Siddhartha Sen (Microsoft Research),
SCALABLE DISTRIBUTED DEEP LEARNING
SEOUL Oct.7, 2016 SCALABLE DISTRIBUTED DEEP LEARNING Han Hee Song, PhD Soft On Net 10/7/2016 BATCH PROCESSING FRAMEWORKS FOR DL Data parallelism provides efficient big data processing: data collecting,
DVFS Space Exploration in Power-Constrained Processing-in-Memory Systems
DVFS Space Exploration in Power-Constrained Processing-in-Memory Systems Marko Scrbak and Krishna M. Kavi Computer Systems Research Laboratory Department of Computer Science & Engineering University of
Key aspects of cloud computing. Towards fuller utilization. Two main sources of resource demand. Cluster Scheduling
Key aspects of cloud computing Cluster Scheduling 1. Illusion of infinite computing resources available on demand, eliminating need for up-front provisioning. The elimination of an up-front commitment
Reducing DRAM Latency at Low Cost by Exploiting Heterogeneity. Donghyuk Lee Carnegie Mellon University
Reducing DRAM Latency at Low Cost by Exploiting Heterogeneity Donghyuk Lee Carnegie Mellon University Problem: High DRAM Latency processor stalls: waiting for data main memory high latency Major bottleneck
Camdoop Exploiting In-network Aggregation for Big Data Applications Paolo Costa
Camdoop Exploiting In-network Aggregation for Big Data Applications costa@imperial.ac.uk joint work with Austin Donnelly, Antony Rowstron, and Greg O Shea (MSR Cambridge) MapReduce Overview Input file
Clipper A Low-Latency Online Prediction Serving System
Clipper A Low-Latency Online Prediction Serving System Daniel Crankshaw Xin Wang, Giulio Zhou, Michael Franklin, Joseph Gonzalez, Ion Stoica NSDI 2017 March 29, 2017 Learning Big Data Training Complex
A Hierarchical Synchronous Parallel Model for Wide-Area Graph Analytics
A Hierarchical Synchronous Parallel Model for Wide-Area Graph Analytics Shuhao Liu*, Li Chen, Baochun Li, Aiden Carnegie University of Toronto April 17, 2018 Graph Analytics What is Graph Analytics? 2
Big Data Hadoop Stack
Big Data Hadoop Stack Lecture #1 Hadoop Beginnings What is Hadoop? Apache Hadoop is an open source software framework for storage and large scale processing of data-sets on clusters of commodity hardware
Accelerating Spark Workloads using GPUs
Accelerating Spark Workloads using GPUs Rajesh Bordawekar, Minsik Cho, Wei Tan, Benjamin Herta, Vladimir Zolotov, Alexei Lvov, Liana Fong, and David Kung IBM T. J. Watson Research Center 1 Outline Spark
Communication Models for Resource Constrained Hierarchical Ethernet Networks
Communication Models for Resource Constrained Hierarchical Ethernet Networks Speaker: Konstantinos Katrinis # Jun Zhu +, Alexey Lastovetsky *, Shoukat Ali #, Rolf Riesen # + Technical University of Eindhoven,
Key aspects of cloud computing. Towards fuller utilization. Two main sources of resource demand. Cluster Scheduling
Key aspects of cloud computing Cluster Scheduling 1. Illusion of infinite computing resources available on demand, eliminating need for up-front provisioning. The elimination of an up-front commitment
AutoTVM & Device Fleet
AutoTVM & Device Fleet ` Learning to Optimize Tensor Programs Frameworks High-level data flow graph and optimizations Hardware Learning to Optimize Tensor Programs Frameworks High-level data flow graph
Enabling Technology for the Cloud and AI One Size Fits All?
Enabling Technology for the Cloud and AI One Size Fits All? Tim Horel Collaborate. Differentiate. Win. DIRECTOR, FIELD APPLICATIONS The Growing Cloud Global IP Traffic Growth 40B+ devices with intelligence
Altair OptiStruct 13.0 Performance Benchmark and Profiling. May 2015
Altair OptiStruct 13.0 Performance Benchmark and Profiling May 2015 Note The following research was performed under the HPC Advisory Council activities Participating vendors: Intel, Dell, Mellanox Compute
How Might Recently Formed System Interconnect Consortia Affect PM? Doug Voigt, SNIA TC
How Might Recently Formed System Interconnect Consortia Affect PM? Doug Voigt, SNIA TC Three Consortia Formed in Oct 2016 Gen-Z Open CAPI CCIX complex to rack scale memory fabric Cache coherent accelerator
Exploiting Inter-Flow Relationship for Coflow Placement in Data Centers. Xin Sunny Huang, T. S. Eugene Ng Rice University
Exploiting Inter-Flow Relationship for Coflow Placement in Data Centers Xin Sunny Huang, T S Eugene g Rice University This Work Optimizing Coflow performance has many benefits such as avoiding application
MOHA: Many-Task Computing Framework on Hadoop
Apache: Big Data North America 2017 @ Miami MOHA: Many-Task Computing Framework on Hadoop Soonwook Hwang Korea Institute of Science and Technology Information May 18, 2017 Table of Contents Introduction
Jinho Hwang (IBM Research) Wei Zhang, Timothy Wood, H. Howie Huang (George Washington Univ.) K.K. Ramakrishnan (Rutgers University)
Jinho Hwang (IBM Research) Wei Zhang, Timothy Wood, H. Howie Huang (George Washington Univ.) K.K. Ramakrishnan (Rutgers University) Background: Memory Caching Two orders of magnitude more reads than writes
Weld: A Common Runtime for Data Analytics
Weld: A Common Runtime for Data Analytics Shoumik Palkar, James Thomas, Anil Shanbhag*, Deepak Narayanan, Malte Schwarzkopf*, Holger Pirk*, Saman Amarasinghe*, Matei Zaharia Stanford InfoLab, *MIT CSAIL
Packing Tasks with Dependencies. Robert Grandl, Srikanth Kandula, Sriram Rao, Aditya Akella, Janardhan Kulkarni
Packing Tasks with Dependencies Robert Grandl, Srikanth Kandula, Sriram Rao, Aditya Akella, Janardhan Kulkarni The Cluster Scheduling Problem Jobs Goal: match tasks to resources Tasks 2 The Cluster Scheduling
IBM CORAL HPC System Solution
IBM CORAL HPC System Solution HPC and HPDA towards Cognitive, AI and Deep Learning Deep Learning AI / Deep Learning Strategy for Power Power AI Platform High Performance Data Analytics Big Data Strategy
Sharing High-Performance Devices Across Multiple Virtual Machines
Sharing High-Performance Devices Across Multiple Virtual Machines Preamble What does sharing devices across multiple virtual machines in our title mean? How is it different from virtual networking / NSX,
Programming Systems for Big Data
Programming Systems for Big Data CS315B Lecture 17 Including material from Kunle Olukotun Prof. Aiken CS 315B Lecture 17 1 Big Data We ve focused on parallel programming for computational science There
Hammer Slide: Work- and CPU-efficient Streaming Window Aggregation
Large-Scale Data & Systems Group Hammer Slide: Work- and CPU-efficient Streaming Window Aggregation Georgios Theodorakis, Alexandros Koliousis, Peter Pietzuch, Holger Pirk Large-Scale Data & Systems (LSDS)
Sparrow. Distributed Low-Latency Spark Scheduling. Kay Ousterhout, Patrick Wendell, Matei Zaharia, Ion Stoica
Sparrow Distributed Low-Latency Spark Scheduling Kay Ousterhout, Patrick Wendell, Matei Zaharia, Ion Stoica Outline The Spark scheduling bottleneck Sparrow s fully distributed, fault-tolerant technique
The 7 deadly sins of cloud computing [2] Cloud-scale resource management [1]
![The 7 deadly sins of cloud computing [2] Cloud-scale resource management [1]](/thumbs/76/74197841.jpg "The 7 deadly sins of cloud computing [2] Cloud-scale resource management [1]")
The 7 deadly sins of [2] Cloud-scale resource management [1] University of California, Santa Cruz May 20, 2013 1 / 14 Deadly sins of of sin (n.) - common simplification or shortcut employed by ers; may
The Future of High Performance Computing
The Future of High Performance Computing Randal E. Bryant Carnegie Mellon University http://www.cs.cmu.edu/~bryant Comparing Two Large-Scale Systems Oakridge Titan Google Data Center 2 Monolithic supercomputer
Efficient, Scalable, and Provenance-Aware Management of Linked Data
Efficient, Scalable, and Provenance-Aware Management of Linked Data Marcin Wylot 1 Motivation and objectives of the research The proliferation of heterogeneous Linked Data on the Web requires data management
IBM Power Systems HPC Cluster
IBM Power Systems HPC Cluster Highlights Complete and fully Integrated HPC cluster for demanding workloads Modular and Extensible: match components & configurations to meet demands Integrated: racked &
Scaling Distributed Machine Learning
Scaling Distributed Machine Learning with System and Algorithm Co-design Mu Li Thesis Defense CSD, CMU Feb 2nd, 2017 nx min w f i (w) Distributed systems i=1 Large scale optimization methods Large-scale
Phoenix: A Constraint-aware Scheduler for Heterogeneous Datacenters
Phoenix: A Constraint-aware Scheduler for Heterogeneous Datacenters Prashanth Thinakaran, Jashwant Raj Gunasekaran, Bikash Sharma, Mahmut Taylan Kandemir, Chita R. Das Computer Science and Engineering,
PAC485 Managing Datacenter Resources Using the VirtualCenter Distributed Resource Scheduler
PAC485 Managing Datacenter Resources Using the VirtualCenter Distributed Resource Scheduler Carl Waldspurger Principal Engineer, R&D This presentation may contain VMware confidential information. Copyright
Overview. Idea: Reduce CPU clock frequency This idea is well suited specifically for visualization
Exploring Tradeoffs Between Power and Performance for a Scientific Visualization Algorithm Stephanie Labasan & Matt Larsen (University of Oregon), Hank Childs (Lawrence Berkeley National Laboratory) 26
RobinHood: Tail Latency Aware Caching Dynamic Reallocation from Cache-Rich to Cache-Poor
RobinHood: Tail Latency Aware Caching Dynamic Reallocation from Cache-Rich to Cache-Poor Daniel S. Berger and Benjamin Berg, Carnegie Mellon University; Timothy Zhu, Pennsylvania State University; Siddhartha
Coflow. Recent Advances and What s Next? Mosharaf Chowdhury. University of Michigan
Coflow Recent Advances and What s Next? Mosharaf Chowdhury University of Michigan Rack-Scale Computing Datacenter-Scale Computing Geo-Distributed Computing Coflow Networking Open Source Apache Spark Open
A Network-aware Scheduler in Data-parallel Clusters for High Performance
A Network-aware Scheduler in Data-parallel Clusters for High Performance Zhuozhao Li, Haiying Shen and Ankur Sarker Department of Computer Science University of Virginia May, 2018 1/61 Data-parallel clusters
Understanding Reduced-Voltage Operation in Modern DRAM Devices
Understanding Reduced-Voltage Operation in Modern DRAM Devices Experimental Characterization, Analysis, and Mechanisms Kevin Chang A. Giray Yaglikci, Saugata Ghose,Aditya Agrawal *, Niladrish Chatterjee
IBM Data Science Experience White paper. SparkR. Transforming R into a tool for big data analytics
IBM Data Science Experience White paper R Transforming R into a tool for big data analytics 2 R Executive summary This white paper introduces R, a package for the R statistical programming language that
Pocket: Elastic Ephemeral Storage for Serverless Analytics
Pocket: Elastic Ephemeral Storage for Serverless Analytics Ana Klimovic*, Yawen Wang*, Patrick Stuedi +, Animesh Trivedi +, Jonas Pfefferle +, Christos Kozyrakis* *Stanford University, + IBM Research 1
The Heterogeneous Programming Jungle. Service d Expérimentation et de développement Centre Inria Bordeaux Sud-Ouest
The Heterogeneous Programming Jungle Service d Expérimentation et de développement Centre Inria Bordeaux Sud-Ouest June 19, 2012 Outline 1. Introduction 2. Heterogeneous System Zoo 3. Similarities 4. Programming
Design-Induced Latency Variation in Modern DRAM Chips:
Design-Induced Latency Variation in Modern DRAM Chips: Characterization, Analysis, and Latency Reduction Mechanisms Donghyuk Lee 1,2 Samira Khan 3 Lavanya Subramanian 2 Saugata Ghose 2 Rachata Ausavarungnirun
Sub-millisecond Stateful Stream Querying over Fast-evolving Linked Data
Sub-millisecond Stateful Stream Querying over Fast-evolving Linked Data Yunhao Zhang, Rong Chen, Haibo Chen Institute of Parallel and Distributed Systems (IPADS) Shanghai Jiao Tong University Stream Query
Cross-Layer Memory Management to Reduce DRAM Power Consumption
Cross-Layer Memory Management to Reduce DRAM Power Consumption Michael Jantz Assistant Professor University of Tennessee, Knoxville 1 Introduction Assistant Professor at UT since August 2014 Before UT
Lecture 13: Memory Consistency. + a Course-So-Far Review. Parallel Computer Architecture and Programming CMU , Spring 2013
Lecture 13: Memory Consistency + a Course-So-Far Review Parallel Computer Architecture and Programming Today: what you should know Understand the motivation for relaxed consistency models Understand the
Bolt: I Know What You Did Last Summer In the Cloud
Bolt: I Know What You Did Last Summer In the Cloud Christina Delimitrou1 and Christos Kozyrakis2 1Cornell University, 2Stanford University Platform Lab Review February 2018 Executive Summary Problem: cloud
Computer Architecture: Multi-Core Processors: Why? Onur Mutlu & Seth Copen Goldstein Carnegie Mellon University 9/11/13
Computer Architecture: Multi-Core Processors: Why? Onur Mutlu & Seth Copen Goldstein Carnegie Mellon University 9/11/13 Moore s Law Moore, Cramming more components onto integrated circuits, Electronics,
Joe Wingbermuehle, (A paper written under the guidance of Prof. Raj Jain)
1 of 11 5/4/2011 4:49 PM Joe Wingbermuehle, wingbej@wustl.edu (A paper written under the guidance of Prof. Raj Jain) Download The Auto-Pipe system allows one to evaluate various resource mappings and topologies
SMiPE: Estimating the Progress of Recurring Iterative Distributed Dataflows
SMiPE: Estimating the Progress of Recurring Iterative Distributed Dataflows Jannis Koch, Lauritz Thamsen, Florian Schmidt, and Odej Kao Technische Universität Berlin, Germany {firstname.lastname}@tu-berlin.de
Lecture 9 Dynamic Compilation
Lecture 9 Dynamic Compilation I. Motivation & Background II. Overview III. Compilation Policy IV. Partial Method Compilation V. Partial Dead Code Elimination VI. Escape Analysis VII. Results Partial Method
Data Center TCP (DCTCP)
Data Center TCP (DCTCP) Mohammad Alizadeh, Albert Greenberg, David A. Maltz, Jitendra Padhye Parveen Patel, Balaji Prabhakar, Sudipta Sengupta, Murari Sridharan Microsoft Research Stanford University 1
DON T CRY OVER SPILLED RECORDS Memory elasticity of data-parallel applications and its application to cluster scheduling
DON T CRY OVER SPILLED RECORDS Memory elasticity of data-parallel applications and its application to cluster scheduling Călin Iorgulescu (EPFL), Florin Dinu (EPFL), Aunn Raza (NUST Pakistan), Wajih Ul
Designing Hybrid Data Processing Systems for Heterogeneous Servers
Designing Hybrid Data Processing Systems for Heterogeneous Servers Peter Pietzuch Large-Scale Distributed Systems (LSDS) Group Imperial College London http://lsds.doc.ic.ac.uk University
Overview of Project's Achievements
PalDMC Parallelised Data Mining Components Final Presentation ESRIN, 12/01/2012 Overview of Project's Achievements page 1 Project Outline Project's objectives design and implement performance optimised,
Preemptive, Low Latency Datacenter Scheduling via Lightweight Virtualization
Preemptive, Low Latency Datacenter Scheduling via Lightweight Virtualization Wei Chen, Jia Rao*, and Xiaobo Zhou University of Colorado, Colorado Springs * University of Texas at Arlington Data Center
Computer Architecture: Multi-Core Processors: Why? Prof. Onur Mutlu Carnegie Mellon University
Computer Architecture: Multi-Core Processors: Why? Prof. Onur Mutlu Carnegie Mellon University Moore s Law Moore, Cramming more components onto integrated circuits, Electronics, 1965. 2 3 Multi-Core Idea:
Proteus: agile ML elasticity through tiered reliability in dynamic resource markets
Proteus: agile ML elasticity through tiered reliability in dynamic resource markets Aaron Harlap, Alexey Tumanov*, Andrew Chung, Greg Ganger, Phil Gibbons (borrowed/adapted from Aaron Harlap s Eurosys
Advances of parallel computing. Kirill Bogachev May 2016
Advances of parallel computing Kirill Bogachev May 2016 Demands in Simulations Field development relies more and more on static and dynamic modeling of the reservoirs that has come a long way from being
Expressing Heterogeneous Parallelism in C++ with Intel Threading Building Blocks A full-day tutorial proposal for SC17
Expressing Heterogeneous Parallelism in C++ with Intel Threading Building Blocks A full-day tutorial proposal for SC17 Tutorial Instructors [James Reinders, Michael J. Voss, Pablo Reble, Rafael Asenjo]
EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures
EC-Bench: Benchmarking Onload and Offload Erasure Coders on Modern Hardware Architectures Haiyang Shi, Xiaoyi Lu, and Dhabaleswar K. (DK) Panda {shi.876, lu.932, panda.2}@osu.edu The Ohio State University
Workflows and Scheduling
Workflows and Scheduling Frank Röder Arbeitsbereich Wissenschaftliches Rechnen Fachbereich Informatik Fakultät für Mathematik, Informatik und Naturwissenschaften Universität Hamburg 14-12-2015 Frank Röder
LRC: Dependency-Aware Cache Management for Data Analytics Clusters. Yinghao Yu, Wei Wang, Jun Zhang, and Khaled B. Letaief IEEE INFOCOM 2017
LRC: Dependency-Aware Cache Management for Data Analytics Clusters Yinghao Yu, Wei Wang, Jun Zhang, and Khaled B. Letaief IEEE INFOCOM 2017 Outline Cache Management for Data Analytics Clusters Inefficiency
Poseidon: An Efficient Communication Architecture for Distributed Deep Learning on GPU Clusters
Poseidon: An Efficient Communication Architecture for Distributed Deep Learning on GPU Clusters Hao Zhang Zeyu Zheng, Shizhen Xu, Wei Dai, Qirong Ho, Xiaodan Liang, Zhiting Hu, Jianliang Wei, Pengtao Xie,
Getafix: Workload-aware Distributed Interactive Analytics
Getafix: Workload-aware Distributed Interactive Analytics Presenter: Mainak Ghosh Collaborators: Le Xu, Xiaoyao Qian, Thomas Kao, Indranil Gupta, Himanshu Gupta Data Analytics 2 Picture borrowed from https://conferences.oreilly.com/strata/strata-ny-2016/public/schedule/detail/51640
Varys. Efficient Coflow Scheduling. Mosharaf Chowdhury, Yuan Zhong, Ion Stoica. UC Berkeley
Varys Efficient Coflow Scheduling Mosharaf Chowdhury, Yuan Zhong, Ion Stoica UC Berkeley Communication is Crucial Performance Facebook analytics jobs spend 33% of their runtime in communication 1 As in-memory
Motivation Goal Idea Proposition for users Study
Exploring Tradeoffs Between Power and Performance for a Scientific Visualization Algorithm Stephanie Labasan Computer and Information Science University of Oregon 23 November 2015 Overview Motivation:
Dynamic Selection of Auto-tuned Kernels to the Numerical Libraries in the DOE ACTS Collection
Numerical Libraries in the DOE ACTS Collection The DOE ACTS Collection SIAM Parallel Processing for Scientific Computing, Savannah, Georgia Feb 15, 2012 Tony Drummond Computational Research Division Lawrence
LazyBase: Trading freshness and performance in a scalable database
LazyBase: Trading freshness and performance in a scalable database (EuroSys 2012) Jim Cipar, Greg Ganger, *Kimberly Keeton, *Craig A. N. Soules, *Brad Morrey, *Alistair Veitch PARALLEL DATA LABORATORY
High Performance File System and I/O Middleware Design for Big Data on HPC Clusters
High Performance File System and I/O Middleware Design for Big Data on HPC Clusters by Nusrat Sharmin Islam Advisor: Dhabaleswar K. (DK) Panda Department of Computer Science and Engineering The Ohio State
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
HPC learning using Cloud infrastructure
HPC learning using Cloud infrastructure Florin MANAILA IT Architect florin.manaila@ro.ibm.com Cluj-Napoca 16 March, 2010 Agenda 1. Leveraging Cloud model 2. HPC on Cloud 3. Recent projects - FutureGRID
Seamless Dynamic Runtime Reconfiguration in a Software-Defined Radio
Seamless Dynamic Runtime Reconfiguration in a Software-Defined Radio Michael L Dickens, J Nicholas Laneman, and Brian P Dunn WINNF 11 Europe 2011-Jun-22/24 Overview! Background on relevant SDR! Problem
ResQ: Enabling SLOs in Network Function Virtualization
ResQ: Enabling SLOs in Network Function Virtualization Amin Tootoonchian* Aurojit Panda Chang Lan Melvin Walls Katerina Argyraki Sylvia Ratnasamy Scott Shenker *Intel Labs UC Berkeley ICSI NYU Nefeli EPFL
Hadoop 2.x Core: YARN, Tez, and Spark. Hortonworks Inc All Rights Reserved
Hadoop 2.x Core: YARN, Tez, and Spark YARN Hadoop Machine Types top-of-rack switches core switch client machines have client-side software used to access a cluster to process data master nodes run Hadoop
Thread Cluster Memory Scheduling: Exploiting Differences in Memory Access Behavior. Yoongu Kim Michael Papamichael Onur Mutlu Mor Harchol-Balter
Thread Cluster Memory Scheduling: Exploiting Differences in Memory Access Behavior Yoongu Kim Michael Papamichael Onur Mutlu Mor Harchol-Balter Motivation Memory is a shared resource Core Core Core Core
The Evolution of Big Data Platforms and Data Science
IBM Analytics The Evolution of Big Data Platforms and Data Science ECC Conference 2016 Brandon MacKenzie June 13, 2016 2016 IBM Corporation Hello, I m Brandon MacKenzie. I work at IBM. Data Science - Offering
TOOLS FOR IMPROVING CROSS-PLATFORM SOFTWARE DEVELOPMENT
TOOLS FOR IMPROVING CROSS-PLATFORM SOFTWARE DEVELOPMENT Eric Kelmelis 28 March 2018 OVERVIEW BACKGROUND Evolution of processing hardware CROSS-PLATFORM KERNEL DEVELOPMENT Write once, target multiple hardware
Big-learning Systems for Big Data
n e w s l e t t e r o n p d l a c t i v i t i e s a n d e v e n t s s p r i n g 2 0 1 6 http://www.pdl.cmu.edu/ an informal publication from academia s premiere storage systems research center devoted
Principal Software Engineer Red Hat Emerging Technology June 24, 2015
USING APACHE SPARK FOR ANALYTICS IN THE CLOUD William C. Benton Principal Software Engineer Red Hat Emerging Technology June 24, 2015 ABOUT ME Distributed systems and data science in Red Hat's Emerging
CPU-GPU Heterogeneous Computing
CPU-GPU Heterogeneous Computing Advanced Seminar "Computer Engineering Winter-Term 2015/16 Steffen Lammel 1 Content Introduction Motivation Characteristics of CPUs and GPUs Heterogeneous Computing Systems
HPC-Reuse: efficient process creation for running MPI and Hadoop MapReduce on supercomputers
Chung Dao 1 HPC-Reuse: efficient process creation for running MPI and Hadoop MapReduce on supercomputers Thanh-Chung Dao and Shigeru Chiba The University of Tokyo, Japan Chung Dao 2 Running Hadoop and
The Cray Rainier System: Integrated Scalar/Vector Computing
THE SUPERCOMPUTER COMPANY The Cray Rainier System: Integrated Scalar/Vector Computing Per Nyberg 11 th ECMWF Workshop on HPC in Meteorology Topics Current Product Overview Cray Technology Strengths Rainier
Bare Metal Library. Abstractions for modern hardware Cyprien Noel
Bare Metal Library Abstractions for modern hardware Cyprien Noel Plan 1. 2. 3. Modern Hardware? New challenges & opportunities Three use cases Current solutions Leveraging hardware Simple abstraction Myself
Trends and Challenges in Big Data
Trends and Challenges in Big Data Ion Stoica November 14, 2016 PDSW-DISCS 16 UC BERKELEY Before starting Disclaimer: I know little about HPC and storage More collaboration than ever between HPC, Distributes
GPU Acceleration of the Longwave Rapid Radiative Transfer Model in WRF using CUDA Fortran. G. Ruetsch, M. Fatica, E. Phillips, N.
GPU Acceleration of the Longwave Rapid Radiative Transfer Model in WRF using CUDA Fortran G. Ruetsch, M. Fatica, E. Phillips, N. Juffa Outline WRF and RRTM Previous Work CUDA Fortran Features RRTM in CUDA
PROBABILISTIC SCHEDULING MICHAEL ROITZSCH
Faculty of Computer Science Institute of Systems Architecture, Operating Systems Group PROBABILISTIC SCHEDULING MICHAEL ROITZSCH DESKTOP REAL-TIME 2 PROBLEM worst case execution time (WCET) largely exceeds
Write a technical report Present your results Write a workshop/conference paper (optional) Could be a real system, simulation and/or theoretical
Identify a problem Review approaches to the problem Propose a novel approach to the problem Define, design, prototype an implementation to evaluate your approach Could be a real system, simulation and/or
RDMA and Hardware Support
RDMA and Hardware Support SIGCOMM Topic Preview 2018 Yibo Zhu Microsoft Research 1 The (Traditional) Journey of Data How app developers see the network Under the hood This architecture had been working
OVERHEADS ENHANCEMENT IN MUTIPLE PROCESSING SYSTEMS BY ANURAG REDDY GANKAT KARTHIK REDDY AKKATI
CMPE 655- MULTIPLE PROCESSOR SYSTEMS OVERHEADS ENHANCEMENT IN MUTIPLE PROCESSING SYSTEMS BY ANURAG REDDY GANKAT KARTHIK REDDY AKKATI What is MULTI PROCESSING?? Multiprocessing is the coordinated processing
ITTC High-Performance Networking The University of Kansas EECS 881 Architecture and Topology
High-Performance Networking The University of Kansas EECS 881 Architecture and Topology James P.G. Sterbenz Department of Electrical Engineering & Computer Science Information Technology & Telecommunications
Apache Spark 2.0 Performance Improvements Investigated With Flame Graphs. Luca Canali CERN, Geneva (CH)
Apache Spark 2.0 Performance Improvements Investigated With Flame Graphs Luca Canali CERN, Geneva (CH) Speaker Intro Database engineer and team lead at CERN IT Hadoop and Spark service Database services
PAGE PLACEMENT STRATEGIES FOR GPUS WITHIN HETEROGENEOUS MEMORY SYSTEMS
PAGE PLACEMENT STRATEGIES FOR GPUS WITHIN HETEROGENEOUS MEMORY SYSTEMS Neha Agarwal* David Nellans Mark Stephenson Mike O Connor Stephen W. Keckler NVIDIA University of Michigan* ASPLOS 2015 EVOLVING GPU
Analyzing I/O Performance on a NEXTGenIO Class System
Analyzing I/O Performance on a NEXTGenIO Class System holger.brunst@tu-dresden.de ZIH, Technische Universität Dresden LUG17, Indiana University, June 2 nd 2017 NEXTGenIO Fact Sheet Project Research & Innovation
Big Data Systems on Future Hardware. Bingsheng He NUS Computing
Big Data Systems on Future Hardware Bingsheng He NUS Computing http://www.comp.nus.edu.sg/~hebs/ 1 Outline Challenges for Big Data Systems Why Hardware Matters? Open Challenges Summary 2 3 ANYs in Big
Degrees of Freedom in Cached Interference Networks with Limited Backhaul
Degrees of Freedom in Cached Interference Networks with Limited Backhaul Vincent LAU, Department of ECE, Hong Kong University of Science and Technology (A) Motivation Interference Channels 3 No side information