Crystal Ball Panel: The Futures of Supercomputing William Gropp www.mcs.anl.gov/~gropp www.mcs.anl.gov/~gropp.

Slides:



Advertisements
Similar presentations
Chapter 26 Testing Bjarne Stroustrup
Advertisements

Three Questions You Should Ask William Gropp Mathematics and Computer Science.
Team Meeting Communication Skills
Welcome to the Be Smoke Free Now Program! The Easy, Quick & Comfortable Way to become a NON-SMOKER today and for the rest of your life.
LADIS workshop (Oct 11, 2009) A Case for the Accountable Cloud Andreas Haeberlen MPI-SWS.
Concept Summary Batesville High School Physics. Natural Philosophy  Socrates, Plato, Aristotle  Were the “authorities” in Western thought from about.
The Triumph of Hope over Experience * ? Bill Gropp *Samuel Johnson.
Error Measurement and Iterative Methods
SENG 531: Labs TA: Brad Cossette Office Hours: Monday, Wednesday.
Foundations of Network and Computer Security J J ohn Black Lecture #3 Aug 28 th 2009 CSCI 6268/TLEN 5550, Fall 2009.
1 Mathematical, Information and Computational Sciences FAST-OS Workshop July 9-10, 2002 Fred Johnson.
1 CMSC 132: Object-Oriented Programming II Software Development III Department of Computer Science University of Maryland, College Park.
ISE554 The WWW 3.4 Evaluation Methods. Evaluating Interfaces with Users Why evaluation is crucial to interface design General approaches and tradeoffs.
James Tam Evaluating Interfaces With Users Why evaluation is crucial to interface design General approaches and tradeoffs in evaluation The role of ethics.
What are the functions of an operating system? The operating system is the core software component of your computer. It performs many functions and is,
*Please put on headphones before beginning! Let’s Begin!
Abstract Load balancing in the cloud computing environment has an important impact on the performance. Good load balancing makes cloud computing more.
Welcome to the wonderful world of……. . A Quick & Easy Guide.  What IS ?  A quick, easy and convenient way to send a letter to friends, family.
First Law of Chem How to do Lab Write-up 1. Word Processor - Computer that will run: a. Word b. Open Office (click on blue dot to download for FREE ) This.
Chapter 9 Database Management Discovering Computers Fundamental.
Principles of Scalable HPC System Design March 6, 2012 Sue Kelly Sandia National Laboratories Abstract: Sandia National.
CSCI-383 Object-Oriented Programming & Design Lecture 1.
David Evans Turing Machines, Busy Beavers, and Big Questions about Computing.
Elementary School Science: Emphasizing the Basics Presented by Frank H. Osborne, Ph. D. © 2015 EMSE 3123 Math and Science in Elem. Ed.
WELCOME! Future Industry Leaders from JLS Middle School.
Making a great Project 2 OCR 1994/2360. Analysis This is the key to getting it right. Too many candidates skip through this section. It’s worth 20% of.
Results Matter. Trust NAG. Numerical Algorithms Group Mathematics and technology for optimized performance Alternative Processors Panel IDC, Tucson, Sept.
Federation Strategy Robert Ricci GENI-FIRE Workshop September 2015.
Cryptography Dec 29. This Lecture In this last lecture for number theory, we will see probably the most important application of number theory in computer.
U.S. Department of Energy’s Office of Science High Performance Computing Challenges and Opportunities Dr. Daniel Hitchcock
The Unreasonable Effectiveness of Mathematics Bram Boroson, Clayton State University, 3/20/2013.
Week 5 - Wednesday.  What did we talk about last time?  Exam 1!  And before that?  Review!  And before that?  if and switch statements.
Panel on Training and Developing HPC People HPC User Forum Dearborn MI April 13, 2010 Paul Buerger Avetec/DICE program Jim Kasdorf.
Just as there are many human languages, there are many computer programming languages that can be used to develop software. Some are named after people,
Crystal Ball Panel ORNL Heterogeneous Distributed Computing Research Al Geist ORNL March 6, 2003 SOS 7.
SILC-NET, a project of ILRU – Independent Living Research Utilization Developing an Outcomes-Focused SPIL: Three Easy Steps Module 3 A series of online.
Sponsored by the National Science Foundation Systematic Experimentation, Automation, and Scaling Up Sarah Edwards, GENI Project Office.
Software Working Group Chairman’s Note: This document was prepared by the “software and applications” working group and was received by the entire workshop.
ELPP, 15 November 2010 e-Feedback Meeting Students’ Needs & Expectations Yuhua Hu & Paul McLaughlin The School of Biological Sciences.
What is Science?. Science is… 1.What do you think of when you think of science? 2.Have you ever done science? What did you do?
The Microtheme. Step One: Thesis A thesis has the question restated and 3 prongs (ideas) What is our question? What three prongs support your answer to.
Selling With Confidence Template Organization Name Seller’s Name.
MESQUITE: Mesh Optimization Toolkit Brian Miller, LLNL
6-Dec-15CSE 60641: Operating Systems1 -The structure of the “THE” multiprogramming system – Edsger W. Dijkstra Presented at the 1 st ACM SOSP, Gatlinburg,
Appreciating OO Design Principles & Patterns Dale Skrien Dept of Computer Science Colby College Waterville, ME 04901
CSE 190p wrapup Michael Ernst CSE 190p University of Washington.
VCE IT Theory Slideshows By Mark Kelly Vceit.com Problem Solving Methodology 1 Analysis.
1 Light Tools Building and confirming tools for inquiry How?Why?
GUIDANCE Connecting Science © Hodder Education 2009 Evaluating a practical technique When we evaluate our method and results, we reflect on what we did.
Tackling I/O Issues 1 David Race 16 March 2010.
INTRODUCTION TO LANGUAGE LEARNING Basics to get you started.
The expected value The value of a variable one would “expect” to get. It is also called the (mathematical) expectation, or the mean.
THE NUCLEAR ENERGY SURVEY : Nuray ÖZDEMİR Prepared by : Nuray ÖZDEMİR EDIRNE SULEYMAN DEMIREL SCIENCE HIGH SCHOOL APRIL 2009 EDIRNE / TURKEY The use of.
The information systems lifecycle Far more boring than you ever dreamed possible!
Scenario’s PED8J-03 Siddharth Bhagavatula. Case Study 1 (Tobacco) In this scenario you are offered a smoke. 1.You could smoke with the kids and your friend.
The Law of Averages. What does the law of average say? We know that, from the definition of probability, in the long run the frequency of some event will.
IBEAR Decision making scenarios By: Ryan Johnston.
Lecture 2 Page 1 CS 236 Online Security Policies Security policies describe how a secure system should behave Policy says what should happen, not how you.
7 th Grade Science Stubbs & Musselman. Mav Marks Mav Marks are ACT based. They will answer questions on a given scientific passage each day. Mav Marks.
Code Simplicity: Software Design In Open Source Projects Max Kanat-Alexander
CSC 108H: Introduction to Computer Programming
Problem Solving Methodology 1
Section 2-1 Inductive Reasoning.
APIs and Function Parameters
Lessons from The File Copy Assignment
Title of your science project
Why PC Based Control ?.
A COLD BEER….
The Nature of Science.
A COLD BEER….
Presentation transcript:

Crystal Ball Panel: The Futures of Supercomputing William Gropp

University of ChicagoDepartment of Energy Where Are We? Scientists are doing new science We have “commodity” supercomputing But…  Programming and debugging, both for correctness and performance, is painful  System administration is hell  Key software is being developed in public, not just debugged  I/O stinks (how many talks used “I/O” and “broken” in the same breath?)  Users discover problems, not the system, not the operators, … Triumph of hope over experience

University of ChicagoDepartment of Energy But I Saw A Demo At Supercomputing! Clarke’s Third law:  Any sufficiently advanced technology is indistinguishable from magic Demo gap  Corollary to Clarke’s 3 rd law: Any sufficiently rigged demo is indistinguishable from magic  Gropp’s conjecture All supercomputing demos are sufficiently rigged There are two futures:

University of ChicagoDepartment of Energy The Demo Future Ever more impressive demos, but …  Users still tell system admins about errors in the system software  Users must choose between programming at a low level (but (maybe) getting performance) or at a high level (but losing generality/performance/portability)  Tools are fragile  Scalability means “scales to as many as two” (far future: eight)  I/O for applications measured in MB/sec

University of ChicagoDepartment of Energy The Stop Kidding Ourselves Future Applications work!  Without any handholding by the tool developers Handholding does not scale Tools work No handholding (repeat: no handholding) I/O for applications measured in GB/sec Most scientists stop programming  Instead, they use tools and environments

University of ChicagoDepartment of Energy How Do We Get There? Emphasize robust tools that scale and interoperate  E.g., Scalable Systems Software SciDAC Recognize the realities of HPC systems and design solutions (both hardware and software) that are for this universe Invest in the science of creating and maintaining high- quality software for HPC  There are reasons why there are so few examples of good HPC software, and it isn’t that the developers aren’t working hard enough  Feynman, on seeing a 10 page proof, observed that if the proof is that long, you haven’t achieved understanding. The fact that so much software is so flakey says that we don’t understand the underlying principles and approaches

University of ChicagoDepartment of Energy Learn where to be new and where to live with a less than “perfect” solution Make no little plans  If you give up standardization, you have to get a lot back for it But don’t make grandiose plans  Recall the quote about MULTICS from Dan Reed’s talk Understand application needs  Not what it desires, what it needs  “I want to invert a matrix” — Not! Work with others  Open processes to develop common interfaces

University of ChicagoDepartment of Energy We Can Get There We must set ambitious but reasonable goals We must close the demo gap We must chose solutions that scale in terms of people, not just compute processors