Top500 November 2013

  • 7,523 views
Uploaded on

Slides from the TOP500 BOF session during SC13 in Denver, Co.

Slides from the TOP500 BOF session during SC13 in Denver, Co.

More in: Technology
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
  • It is possible download the PDF?
    Are you sure you want to
    Your message goes here
No Downloads

Views

Total Views
7,523
On Slideshare
0
From Embeds
0
Number of Embeds
17

Actions

Shares
Downloads
0
Comments
1
Likes
1

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide
  • Tianhee-1A #1 Nov 10
  • Nov09 N500 growth starts lagging – dropped from 200%to 150% per year
  • Nov09 N500 growth starts lagging – dropped from 200%to 150% per year
  • Gini: 0 = perfect equality (All system are the same size); 1 = total inequality (one system has all Rmax)
  • Gini: 0 = perfect equality (All system are the same size); 1 = total inequality (one system has all Rmax)
  • Gini: 0 = perfect equality (All system are the same size); 1 = total inequality (one system has all Rmax)
  • Both curves are identical and show ”knee” in 2008;The low end of both populations started to lag
  • Fujitsu, Dell, and Dawning have 2 each
  • Light blue diamonds: GigE systemsDark Blue Circles: 10G systemsRed Squares: Infinband and custom interconnectsGolden Triangle: All Accelerator systems
  • 38% and 26% per year

Transcript

  • 1. Highlights of the nd 42 TOP500 List SC13, Denver, CO
  • 2. # Site 1 2 3 41st Computer Country National University of Defense Technology NUDT Tianhe-2 NUDT TH-IVB-FEP, China Oak Ridge National Laboratory Cray Lawrence Livermore National Laboratory RIKEN Advanced Institute 4 for Computational Science 5 6 7 8 9 10 List: The TOP10 Manufacturer Argonne National Laboratory Swiss National Supercomputing Centre (CSCS) Texas Advanced Computing Center/UT Rmax Power [Pflops] [MW] 3,120,000 33.9 17.8 USA 560,640 17.6 8.21 USA 1,572,864 17.2 7.89 795,024 10.5 12.7 Xeon 12C 2.2GHz, IntelXeon Phi Titan Cray XK7, Opteron 16C 2.2GHz, Gemini, NVIDIA K20x IBM Sequoia BlueGene/Q, Power BQC 16C 1.6GHz, Custom Fujitsu IBM K Computer SPARC64 VIIIfx 2.0GHz, Tofu Interconnect Mira BlueGene/Q, Japan USA 786,432 8.59 3.95 Switzerland 115,984 6.27 2.33 USA 462,462 5.17 4.51 Germany 458,752 5.01 2.30 USA 393,216 4.29 1.97 Germany 147,456 2.90 3.52 Power BQC 16C 1.6GHz, Custom Cray Piz Daint Cray XC30, Xeon E5 8C 2.6GHz, Aries, NVIDIA K20x Dell Stampede PowerEdge C8220, Xeon E5 8C 2.7GHz, Intel Xeon Phi Forschungszentrum Juelich (FZJ) IBM Lawrence Livermore National Laboratory IBM Leibniz Rechenzentrum Cores JuQUEEN BlueGene/Q, Power BQC 16C 1.6GHz, Custom Vulcan BlueGene/Q, Power BQC 16C 1.6GHz, Custom IBM SuperMUC iDataPlex DX360M4, Xeon E5 8C 2.7GHz, Infiniband FDR
  • 3. 100 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 Replacement Rate 350 300 250 200 150 137 50 0
  • 4. Performance Development 1 Eflop/s 1E+09 250 PFlop/s 100 Pflop/s 10000000 33.9 PFlop/s 10 Pflop/s 10000000 1 Pflop/s 1000000 118 TFlop/s SUM 100 Tflop/s 100000 10 Tflop/s 10000 N=1 1 1000 Tflop/s 1.17 TFlop/s 100 Gflop/s 100 10 Gflop/s 10 59.7 GFlop/s N=500 1 Gflop/s 1 2012 2010 2008 2006 2004 2002 2000 1998 400 MFlop/s 1996 0.1 1994 100 Mflop/s
  • 5. Projected Performance Development 1E+11 1E+10 1 Eflop/s 1E+09 100 Pflop/s 000000 10 Pflop/s 000000 1 Pflop/s 000000 100 Tflop/s 100000 10 Tflop/s 10000 1 Tflop/s 1000 100 Gflop/s 100 10 Gflop/s 10 1 Gflop/s 1 100 Mflop/s 0.1 SUM N=1 2020 2014 2008 2002 1996 N=500
  • 6. Accelerators 70 60 50 40 Clearspeed IBM Cell 30 ATI Radeon 20 Nvidia Kepler Nvidia Fermi 10 2013 2012 2011 2010 2009 2008 2007 0 2006 Systems Intel Xeon Phi
  • 7. 2013 2012 2011 2010 2009 2008 2007 2006 Fraction of Total TOP500 Performance Performance Share of Accelerators 40% 35% 30% 25% 20% 15% 10% 5% 0%
  • 8. Projected Performance Development 1E+11 1E+10 11E+09 Eflop/s 100 Pflop/s 10000000 10 Pflop/s 10000000 1000000 1 Pflop/s 100000 100 Tflop/s 10000 10 Tflop/s 1000 1 Tflop/s 100 100 Gflop/s 10 10 Gflop/s 1 1 Gflop/s 0.1 100 Mflop/s SUM N=1 2020 2014 2008 2002 1996 N=500
  • 9. Rank at which Half of total Performance is accumulated 100 90 80 70 60 50 40 30 20 10 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 0
  • 10. Gini Coefficient • A measure of statistical dispersion intended to represent inequality – Area A above the Lorenz curve (cummulative distribution) – Gini = A/(A+B) – 0: All members have the same – 1: One member has everything
  • 11. Gini Coefficient of the TOP500 Gini 70 65 60 55 50 45 40 35 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 30
  • 12. Gini Coefficient of the TOP50 Research/Academic/Classified Systems Gini 70 65 60 55 50 45 40 35 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 30
  • 13. Gini Coefficient of the TOP50 Research and Industry Systems 70 Research 60 Industry 50 40 30 20 10 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 0
  • 14. Performance Development of TOP50 Research and Industry Systems 1 Eflop/s 1E+09 100 Pflop/s 10000000 10 Pflop/s 10000000 1 Pflop/s 1000000 100 Tflop/s 100000 10 Tflop/s 10000 1 Tflop/s 1000 100 Gflop/s 10100 Gflop/s 10 1 Gflop/s 100 Mflop/s 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1
  • 15. Performance Development of “Bottom-50” Research and Industry Systems 1 Eflop/s 10000000 100 Pflop/s 10000000 10 Pflop/s 1000000 1 Pflop/s 100 Tflop/s 100000 10 Tflop/s 10000 1 Tflop/s 1000 100 Gflop/s 10100 Gflop/s 1 Gflop/s 10 100 Mflop/s 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1
  • 16. Countries / System Share United States China Canada 2% India 2% Japan Others 12% United Kingdom France United States 53% Germany 4% France 4% Germany India Canada Others United Kingdom 5% Japan 6% China 12%
  • 17. Performance of Countries US EU 10,000 Japan 1,000 China 100 10 1 2012 2010 2008 2006 2004 2002 0 2000 Total Performance [Tflop/s] 100,000
  • 18. Vendors / System Share Hitachi NEC Others 4 33 4 1% 6% NUDT 1% 4 Dell 1% 8 Fujitsu 2% 8 2% Cray Inc. SGI 48 Bull 17 9% 14 3% 3% IBM HP IBM 164 33% Cray Inc. SGI Bull Fujitsu Dell NUDT Hitachi HP 196 39% NEC Others
  • 19. Vendors (TOP50) / System Share IBM Cray Inc Bull 3 6% Others 9 18% SGI IBM 19 38% NUDT Bull Others NUDT 3 6% Fujitsu 3 6% SGI 3 6% Fujitsu Cray Inc 10 20%
  • 20. Linpack Efficiency 120% Linpack Efficiency 100% 80% 60% 40% 20% 0% 0 100 200 300 400 500
  • 21. Power Consumption 8 TOP10 7 3.0 x in 5 y Power [MW] 6 5 4 TOP50 3 2.6 x in 5 y 2 TOP500 1 0 2008 2009 2010 2011 2012 2013 2.7 x in 5 y
  • 22. Power Efficiency Linpack/Power [Gflops/kW] 2,000 TOP10 1,800 1,600 1,400 1,200 TOP50 1,000 800 600 TOP500 400 200 0 2008 2009 2010 2011 2012 2013
  • 23. Linpack/Power [Gflops/kW] Power Efficiency 2000 1800 1600 1400 1200 1000 800 600 400 200 0 TOP10 TOP50 TOP500
  • 24. Linpack/Power [Gflops/kW] Power Efficiency 5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0 TOP10 TOP50 TOP500
  • 25. Tsubame KFC Linpack/Power [Gflops/kW] Power Efficiency NVIDIA K20x AMD FirePro 4,000 3,500 Max-Efficiency 3,000 Mic BlueGene/Q 2,500 2,000 1,500 1,000 500 0 Cell TOP10 TOP50 TOP500
  • 26. Most Power Efficient Architectures Rmax/ Power Computer Tsubame KFC, NEC, Xeon 6C 2.1GHz, Infiniband FDR, NVIDIA K20x 3,418 HA-PACS TCA, Cray Cluster, Xeon 10C 2.8GHz, QDX, NVIDIA K20x 2,980 SANAM, Adtech, ASUS, Xeon 8C 2.0GHz, Infiniband FDR, AMD FirePro 2,973 iDataPlex DX360M4, Xeon 8C 2.6GHz, Infiniband FDR14, NVIDIA K20x 2,702 Piz Daint, Cray XC30, Xeon 8C 2.6GHz, Aries, NVIDIA K20x 2,697 BlueGene/Q, Power BQC 16C 1.60 GHz, Custom 2,300 HPCC, Cluster Platform SL250s, Xeon 8C 2.4GHz, FDR, NVIDIA K20m 2,243 Titan, Cray XK7, Opteron 16C 2.2GHz, Gemini, NVIDIA K20x 2,143 [Mflops/Watt]