Presentation is loading. Please wait.

Presentation is loading. Please wait.

Comparing dual- and quad-core performance

Published byArthur Sparks Modified over 6 years ago

Similar presentations

Presentation on theme: "Comparing dual- and quad-core performance"— Presentation transcript:

1 Comparing dual- and quad-core performance
Harald Vogt (DESY) Michael Volkmann (HUB) Peter Wegner (DESY) Stephan Wiesand (DESY) Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

2 Comparing dual- and quad-core performance
Outline Benchmark hardware and software Architectural aspects Results Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

3 Benchmark systems System Model OS CPU RAM
blade01 PE1955 SL4/64 Clovertown 5345, 2.33 Ghz x 4 8 x 2 GB DDR FBD blade SL3/64 32-app blade02 SL3//32 cube6.ht Prec SL3/32 Xeon, 2.4 Ghz x 1, +HT 2 GB RD-800 cube6.noht SL3//32 erinye1 PE1950 SL4/64 Woodcrest 5160, 3.0 Ghz x 2 4 x 2 GB DDR FBD galaxy30 X4100 SL4/64 Opteron 280, 2.4 Ghz x 2 4 x 2 GB DDR-400 linfini V20z SL4/64 Opteron 250, 2.4 Ghz x 1 4 x 1 GB DDR-400 victoria PE1950 SL4/64 Dempsey, 3.0 Ghz x 2, +HT 4 x 512 MB DDR FBD Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

4 Software ROOT Stress Benchmark Athena Framework: ROOT: 5.08.00
n parallel stress benchmarks run in batch mode, 3000 events (stress -b 3000), 2 consecutive runs, results are from first run (2nd run is to keep load at same level during all measurements). Test directories were created in /dev/shm (no I/O to disk). Power management (Demand Based Switching, PowerNow) was enabled on blade01, galaxy30 resp. Athena Framework: Bit Bit (simulation only) Simulation, Reconstruction: 10 Higgs Events, H->ZZ->llqq (DC PythiaH185zzqqll.py) Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

5 Architectural aspects - Chipset
Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

6 Architectural aspects – Chipset, cont.
Bottlenecks ?? Core1 2 Core3 4 Role of snoop filter in 5000X Chipset ? Core Core 2 x 10.6 GB/s 4 x 5.3 GB/s Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

7 ROOT stress test ( ROOT 5.08.00)
64-Bit ROOT on quad-core Clovertown 32-Bit ROOT on quad-core Clovertown 64-Bit ROOT on dual-core Woodcrest 64-Bit ROOT on dual-core Opteron Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

8 ROOT stress test ( ROOT 5.08.00)
Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

9 ATLAS Athena Simulation – processing time
Processing time (sec) Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

10 ATLAS Athena Simulation – scaling behavior
Tbest/Tn Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

11 ATLAS Athena Simulation – memory usage per core, 8 cores (top)
Virt Res Shr 32-Bit 608m 498m 79m 64-Bit 1221m (2.0) 719m (1.44) 85m Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

12 ATLAS Athena Reconstruction – processing time
Processing time (sec) Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

13 ATLAS Athena Reconstruction – scaling behavior
Tbest/Tn Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

14 ATLAS Athena event generation – processing time
Processing time (sec) Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

15 ATLAS Athena event generation – scaling behavior
Tbest/Tn Number of Jobs Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

16 Performance per Watt Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

17 Comparing dual- and quad-core performance
Conclusions With more than 4 Jobs per system quad-cores are scaling extremely good for simulation and good for reconstruction compared to dual-cores – even when quad-core CPUs with lower clock frequency are used 64-Bit Appl. result in approx. 20% performance gain against 32-Bit Open issues: Particle Astrophysics-, LQCD-Benchmarks References Atlas Athena framework: Christian Vilsbeck ( ) Peter Wegner Multi-core performance in High Energy Physics HEPiX Workshop 2007

Download ppt "Comparing dual- and quad-core performance"

Similar presentations

Presented by: Yash Gurung, ICFAI UNIVERSITY.Sikkim BUILDING of 3 R'sCLUSTER PARALLEL COMPUTER.

Presented by: Yash Gurung, ICFAI UNIVERSITY.Sikkim BUILDING of 3 R'sCLUSTER PARALLEL COMPUTER.
The Protein Folding Problem David van der Spoel Dept. of Cell & Mol. Biology Uppsala, Sweden

The Protein Folding Problem David van der Spoel Dept. of Cell & Mol. Biology Uppsala, Sweden
Exploring The Green Blade Ken Lutz University of California, Berkeley LoCal Retreat, June 8, 2009.

Exploring The Green Blade Ken Lutz University of California, Berkeley LoCal Retreat, June 8, 2009.
Peter Wegner, DESY CHEP03, 25 March 2003 1 LQCD benchmarks on cluster architectures M. Hasenbusch, D. Pop, P. Wegner (DESY Zeuthen), A. Gellrich, H.Wittig.

Peter Wegner, DESY CHEP03, 25 March LQCD benchmarks on cluster architectures M. Hasenbusch, D. Pop, P. Wegner (DESY Zeuthen), A. Gellrich, H.Wittig.
HS06 on the last generation of CPU for HEP server farm Michele Michelotto 1.

HS06 on the last generation of CPU for HEP server farm Michele Michelotto 1.
Gordon: Using Flash Memory to Build Fast, Power-efficient Clusters for Data-intensive Applications A. Caulfield, L. Grupp, S. Swanson, UCSD, ASPLOS’09.

Gordon: Using Flash Memory to Build Fast, Power-efficient Clusters for Data-intensive Applications A. Caulfield, L. Grupp, S. Swanson, UCSD, ASPLOS’09.
Test results Test definition (1) Istituto Nazionale di Fisica Nucleare, Sezione di Roma; (2) Istituto Nazionale di Fisica Nucleare, Sezione di Bologna.

Test results Test definition (1) Istituto Nazionale di Fisica Nucleare, Sezione di Roma; (2) Istituto Nazionale di Fisica Nucleare, Sezione di Bologna.
COLLABORATIVE EXECUTION ENVIRONMENT FOR HETEROGENEOUS PARALLEL SYSTEMS Aleksandar Ili´c, Leonel Sousa 2010 IEEE International Symposium on Parallel & Distributed.

COLLABORATIVE EXECUTION ENVIRONMENT FOR HETEROGENEOUS PARALLEL SYSTEMS Aleksandar Ili´c, Leonel Sousa 2010 IEEE International Symposium on Parallel & Distributed.
Energy Profiling And Analysis Of The HPC Challenge Benchmarks Scalable Performance Laboratory Department of Computer Science Virginia Tech Shuaiwen Song,

Energy Profiling And Analysis Of The HPC Challenge Benchmarks Scalable Performance Laboratory Department of Computer Science Virginia Tech Shuaiwen Song,
Cluster currently consists of: 1 Dell PowerEdge 2950 3.6Ghz Dual, quad core Xeons (8 cores) and 16G of RAM Original GRIDVM - SL4 VM-Ware host 1 Dell PowerEdge.

Cluster currently consists of: 1 Dell PowerEdge Ghz Dual, quad core Xeons (8 cores) and 16G of RAM Original GRIDVM - SL4 VM-Ware host 1 Dell PowerEdge.
Service Review and Plans Carmine Cioffi Database Administrator and Developer 3D Workshop, Barcelona (ES), 20-21 April 2009.

Service Review and Plans Carmine Cioffi Database Administrator and Developer 3D Workshop, Barcelona (ES), April 2009.
Parallelizing Security Checks on Commodity Hardware E.B. Nightingale, D. Peek, P.M. Chen and J. Flinn U Michigan.

Parallelizing Security Checks on Commodity Hardware E.B. Nightingale, D. Peek, P.M. Chen and J. Flinn U Michigan.
AUTHORS: STIJN POLFLIET ET. AL. BY: ALI NIKRAVESH Studying Hardware and Software Trade-Offs for a Real-Life Web 2.0 Workload.

AUTHORS: STIJN POLFLIET ET. AL. BY: ALI NIKRAVESH Studying Hardware and Software Trade-Offs for a Real-Life Web 2.0 Workload.
CERN - IT Department CH-1211 Genève 23 Switzerland www.cern.ch/i t Tier0 database extensions and multi-core/64 bit studies Maria Girone, CERN IT-PSS LCG.

CERN - IT Department CH-1211 Genève 23 Switzerland t Tier0 database extensions and multi-core/64 bit studies Maria Girone, CERN IT-PSS LCG.
Hardware Trends. Contents Memory Hard Disks Processors Network Accessories Future.

Hardware Trends. Contents Memory Hard Disks Processors Network Accessories Future.
High Performance Computing Processors Felix Noble Mirayma V. Rodriguez Agnes Velez Electric and Computer Engineer Department August 25, 2004.

High Performance Computing Processors Felix Noble Mirayma V. Rodriguez Agnes Velez Electric and Computer Engineer Department August 25, 2004.
Energy Savings with DVFS Reduction in CPU power Extra system power.

Energy Savings with DVFS Reduction in CPU power Extra system power.
HS06 on new CPU, KVM virtual machines and commercial cloud Michele Michelotto 1.

HS06 on new CPU, KVM virtual machines and commercial cloud Michele Michelotto 1.
Hyper Threading Technology. Introduction Hyper-threading is a technology developed by Intel Corporation for it’s Xeon processors with a 533 MHz system.

Hyper Threading Technology. Introduction Hyper-threading is a technology developed by Intel Corporation for it’s Xeon processors with a 533 MHz system.
Performance and Scalability of xrootd Andrew Hanushevsky (SLAC), Wilko Kroeger (SLAC), Bill Weeks (SLAC), Fabrizio Furano (INFN/Padova), Gerardo Ganis.

Performance and Scalability of xrootd Andrew Hanushevsky (SLAC), Wilko Kroeger (SLAC), Bill Weeks (SLAC), Fabrizio Furano (INFN/Padova), Gerardo Ganis.

Similar presentations

Ads by Google