Supercomputers in Science from the big bang to climate change - PowerPoint PPT Presentation

Supercomputers in Science from the big bang to climate change David Henty EPCC, the University of Edinburgh d.henty@epcc.ed.ac.uk +44 131 650 5960

what is EPCC? • Edinburgh Parallel Computing Centre – founded in 1990 • The Supercomputing Centre at the University of Edinburgh January 2014 Supercomputers in Science 2

what are computers used for? January 2014 Supercomputers in Science 3

computational science experiment theory computer simulation January 2014 Supercomputers in Science 4

why? January 2014 Supercomputers in Science 5

what is the world’s yearly income? 1 Aadel Abdali Afghanistan £873 2 Aamir Abdali Afghanistan £798 3 …......... …………….. ……………………. ............ 4 …......... …………….. ……………………. ............ 5 …......... …………….. ……………………. ............ 6 …......... …………….. ……………………. ............ 7 …......... …………….. ……………………. ............ 8 …......... …………….. ……………………. ............ 9 …......... …………….. ……………………. ............ 10 …......... …………….. ……………………. ............ January 2014 Supercomputers in Science 6

… further down 5,303,422,761 ……… ………….. …. ………… 5,303,422,762 ……… ………….. …. ………… 5,303,422,763 Mark Henson UK £28,176 5,303,422,764 Mary Henson UK £37,866 5,303,422,765 ……… ………….. …. ………… 5,303,422,766 ……… ………….. …. ………… 5,303,422,767 ……… ………….. …. ………… 5,303,422,768 David Henty UK £ 1 1,234 5,303,422,769 ……… ………….. …. ………… 5,303,422,770 ……… ………….. …. ………… January 2014 Supercomputers in Science 7

… near the bottom 5,342,564,831 ……… ………….. 1 …......... …………….. ……………………. ............ …. ………… 5,342,564,832 ……… ………….. 2 …......... …………….. ……………………. ............ …. ………… 5,342,564,833 ……… ………….. 3 …......... …………….. ……………………. ............ …. ………… 5,342,564,834 Elisa Windsor 4 …......... …………….. ……………………. ............ UK £23,677 5,342,564,835 Elizabeth Windsor* UK £38,356,973 5 …......... …………….. ……………………. ............ 5,342,564,836 ……… ………….. 6 …......... …………….. ……………………. ............ …. ………… 5,342,564,837 ……… ………….. 7 …......... …………….. ……………………. ............ …. ………… 5,342,564,838 ……… ………….. 8 …......... …………….. ……………………. ............ …. ………… 5,342,564,839 ……… ………….. 9 …......... …………….. ……………………. ............ …. ………… 5,342,564,840 ……… ………….. 10 …......... …………….. ……………………. ............ …. ………… (* The Queen) January 2014 Supercomputers in Science 8

… the end 6,999,999,991 ……… ………….. …………….. ………… 6,999,999,992 ……… ………….. …………….. ………… 6,999,999,993 ……… ………….. …………….. ………… 6,999,999,994 ……… ………….. …………….. ………… 6,999,999,995 ……… ………….. …………….. ………… 6,999,999,996 ……… ………….. …………….. ………… 6,999,999,997 ……… ………….. …………….. ………… 6,999,999,998 ……… ………….. …………….. ………… 6,999,999,999 Zojj Zinyama Zimbabwe £3,564 7,000,000,000 Zuka Zinyama Zimbabwe £1,236 January 2014 Supercomputers in Science 9

computer program set running total to zero start at top of list add income to total go to next item in list repeat if not at end of list print total January 2014 Supercomputers in Science 10

how long does it take? Year Intel Frequ- Operations Time per Time per Total time CPU ency per second operation loop 1966 1971 1993 2012 January 2014 Supercomputers in Science 11

how long does it take? Year Intel Frequ- Operations Time per Time per Total time CPU ency per second operation loop 1966 1 Hz 1 1 second 3 seconds 650 years Me 1971 1993 2012 January 2014 Supercomputers in Science 12

how long does it take? Year Intel Frequ- Operations Time per Time per Total time CPU ency per second operation loop 1966 1 Hz 1 1 second 3 seconds 650 years Me 100 KHz 100 thousand 10 micro- 30 micro- 2½ days 1971 seconds seconds i4004 (millionths of a second) 1993 2012 January 2014 Supercomputers in Science 13

how long does it take? Year Intel Frequ- Operations Time per Time per Total time CPU ency per second operation loop 1966 1 Hz 1 1 second 3 seconds 650 years Me 100 KHz 100 thousand 10 micro- 30 micro- 2½ days 1971 seconds seconds i4004 (millionths of a second) 1993 60 MHz 60 million 17 nano- 50 nano- 6 minutes seconds seconds Pentium 2012 January 2014 Supercomputers in Science 14

how long does it take? Year Intel Frequ- Operations Time per Time per Total time CPU ency per second operation loop 1966 1 Hz 1 1 second 3 seconds 650 years Me 100 KHz 100 thousand 10 micro- 30 micro- 2½ days 1971 seconds seconds i4004 (millionths of a second) 1993 60 MHz 60 million 17 nano- 50 nano- 6 minutes seconds seconds Pentium 3 GHz 3 billion 0.3 nano- 1 nano- 7 seconds seconds second 2012 Core i7 (billionths of a second) January 2014 Supercomputers in Science 15

how much faster? 300 1 1971 1975 1980 1985 1990 January 2014 Supercomputers in Science 16

how much faster? 30,000 300 1 1990 1995 2000 2005 2010 January 2014 Supercomputers in Science 17

two thousand transistors (1971) Moore’s Law Number of transistors doubles every 2 years January 2014 Supercomputers in Science 18

three million transistors (1993) January 2014 Supercomputers in Science 19

100 million transistors (2004) January 2014 Supercomputers in Science 20

200 million transistors (2006) dual-core processor same frequency = less power = less heat January 2014 Supercomputers in Science 21

serial vs parallel computing serial ► adjective [ attrib ] Computing parallel processing ► noun [ mass noun ] Computing (of a processor) running only a single task. a mode of operation in which a process is split into many parts, which are executed simultaneously on different processors attached to the same computer. January 2014 Supercomputers in Science 22

parallel computer program if I am core 1: total1 = sum top half list if I am core 2: total2 = sum bottom half list wait for both cores to finish total = total1 + total2 print total January 2014 Supercomputers in Science 23

four billion transistors (2012) 32 cores “multicore” January 2014 Supercomputers in Science 24

how much faster in parallel? 120,000 300 1990 1995 2000 2005 2010 January 2014 Supercomputers in Science 25

parallel supercomputers + = fast + = super fast January 2014 Supercomputers in Science 26

Cray XE6 run by EPCC at Edinburgh 5632 CPUs 90,112 cores 90,112 GB 1 Megawatt very fast network January 2014 Supercomputers in Science 27

Cray XC30 run by EPCC at Edinburgh 9840 CPUs 118,080 cores 338,944 GB 1 Megawatt even faster network and CPUs January 2014 Supercomputers in Science 28

parallel computer simulations January 2014 Supercomputers in Science 29

universal experiment January 2014 Supercomputers in Science 30

big bang: R.Cen & J. Ostriker, Illinois January 2014 Supercomputers in Science 31

big bang: R.Cen & J. Ostriker, Illinois January 2014 Supercomputers in Science 32

big bang: project stargate at SC’09 January 2014 Supercomputers in Science 33

big bang: project stargate at SC’09 January 2014 Supercomputers in Science 34

climate modelling: UCAR January 2014 Supercomputers in Science 35

climate modelling: UCAR January 2014 Supercomputers in Science 36

ocean modelling: ECCO2 / NASA January 2014 Supercomputers in Science 37

ocean modelling: ECCO2 / NASA January 2014 Supercomputers in Science 38

ocean modelling: close up January 2014 Supercomputers in Science 39

ocean modelling: close up January 2014 Supercomputers in Science 40

cell membranes: Science, 296:525-530, 2002 January 2014 Supercomputers in Science 41

cell membranes: Science, 296:525-530, 2002 January 2014 Supercomputers in Science 42

four legs or two? Dr Lee Margetts January 2014 Supercomputers in Science 43

four legs or two? Dr Lee Margetts January 2014 Supercomputers in Science 44

running hadrosaur January 2014 Supercomputers in Science 45

hopping hadrosaur January 2014 Supercomputers in Science 48

Supercomputers in Science from the big bang to climate change - PowerPoint PPT Presentation

Supercomputers in Science from the big bang to climate change David Henty EPCC, the University of Edinburgh d.henty@epcc.ed.ac.uk +44 131 650 5960 what is EPCC? Edinburgh Parallel Computing Centre founded in 1990 The

Supercomputers and Supercomputers and Clusters and Clusters and Grid, Grid, Oh My! Oh My!

The Integrative Role of COWs and Supercomputers in Research and Education Activities Don

LARGE SCALE VISUALIZATION ON GPU ACCELERATED SUPERCOMPUTERS Peter Messmer, 11/16/2015

Improving Hadoop MapReduce Performance on Supercomputers with JVM Reuse Thanh-Chung Dao and

Efficient Scientific Data Efficient Scientific Data Management on Supercomputers Management on

Use of NSF Supercomputers Rob Gardner, University of Chicago OSG Council, Indianapolis, October

transfer package for next-generation supercomputers Approved for public release Benjamin R.

Adaptive Metric-Aware Job Scheduling for Production Supercomputers Wei Tang, Dongxu Ren,

SUPERCOMPUTERS TO SUPERCARS Bill Veenhuis Sr. Solutions Architect, Automotive

Capacity Planning of Supercomputers Simulating MPI Applications at Scale Tom Cornebize Under the

Simulating Supernovae with Supercomputers Don Willcox Center for Computational Sciences and

BENCHMARKING SUPERCOMPUTERS IN THE POST-MOORE ERA Dan Stanzione Executive Director, Texas

IS TOPOLOGY IMPORTANT AGAIN? Effects of contention on message latencies in large supercomputers

Black-hole binary simulations on supercomputers U. Sperhake CSIC-IEEC Barcelona 2 nd Iberian

Scheduling Many-Task Workloads on Supercomputers Dealing with Trailing Tasks Timothy G.

CHAIRS AND SUPERCOMPUTERS: CREATIVE COLLABORATIONS IN AI Jeff Thompson Assistant

Periodic I/O Scheduling for Supercomputers Guillaume Aupy 1 , Ana Gainaru 2 , Valentin Le F` evre 3

2010 Computing on Grids and Supercomputers Improving Many-Task Computing in Scientific Workflows

Realizing Extremely LargeScale Stencil Applications on GPU Supercomputers Toshio Endo, Yuki

Installing SPRINT Supercomputers SPRINT is already installed on HPC Wales and

Experiences with Charm++ and NAMD on Knights Landing Supercomputers 15 th Annual Workshop on

Topology-Aware Data Aggregation for Intensive I/O on Large-Scale Supercomputers Franois Tessier

Fault Tolerance Support for Supercomputers with Multicore Nodes Esteban Meneses Xiang Ni

Power!Use!of!Disk!Subsystems!in! Supercomputers! Ma8hew!L.!Curry! Sandia!Na?onal!Laboratories!