SlideShare una empresa de Scribd logo
1 de 13
Descargar para leer sin conexión
SVG Upgrade 2012
AGENDA


•Upgraded SVG
   • Motivation
   • Hardware configuration
   • Infiniband Network
   • Environment
   • Queues configuration
   • Benchmarks
•CESGA Supercomputers in 2013
   • Distribution of jobs
Hardware

Thin-nodes                                         Total
                                                    Total
  8 HP SL230s Gen8 each with                          24 24 Intel Sandy CPUs
                                                         Intel Sandy Bridge Bridge   processors
     2x Intel Xeon E5-2670, 8 cores each, 2.6GHz        192 cores
                                                      192 cores
                                                         1,5 TB memory
     64GB main memory DDR3-1600MHz                    1,5 TB memory
                                                         28 TB disk
     2TB SAS hard disk                                28 Peak Performance 3788 GFLops
                                                         TB disk
     2x1GbE
                                                      3788 Gflops peak performance
     2xInfiniband FDR 56Gb
     Peak Performance 332GFlops
Fat-nodes
  2 HP DL560 Gen8 each with
     4x Intel Xeon E5-4620, 8 cores each, 2.2GHz
     512GB main memory DDR3-1600MHz
     6 hard disks each 1TB
     4x1GbE
     Infiniband FDR 56Gb
     10GbE
     Peak Performance 563 GFlops
Motivation
Target
• Competitive solution for:
  • Parallel MPI & OpenMP applications
  • Memory Intensive
• Alternative for Finis Terrae
• Lower cost of operation and maintenance
• Finis Terrae II prototype
  • To define new requirements
Infiniband network
Mellanox SX6036 switch
36 port FDR 56Gb/s
4Tb/s aggregated non-blocking BW
1 microseconds MPI latency
Dual connection:
  High availability – same BW
Environment
Integrated in the SVG cluster:
  Scientific Linux 6.3 (Red Hat)
  Common /opt/cesga
  Common /home, stores…
  Same gateway: svg.cesga.es
  Interactive use: compute –arch sandy
  Jobs: qsub –arch sandy
  Binary compatible - no need to recompile
Usage
              compute –arch amd
              qsub –arch amd




      svg.cesga.es

                        compute
SSH                       qsub


              compute –arch sandy
              qsub –arch sandy
Configuration
Full production phase (November 2012)
  •   Only runs jobs with –sandy option
  •   General availability of applications specifically compiled
  •   Maximum wall-clock time 12 hours
  •   Maximum 2 jobs per node fat nodes


Under consideration near future:
  Jobs without –sandy option
  Higher wall-clock time
Queues Configuration
Exclusive nodes
  To take advantage of Infiniband
  To take advantage of Turboboost
  Jobs not interferring each other
  Maximum performance
Maximum 2 jobs on Fat nodes
  32 cores nodes
  Exclusive if required by the jobs (cores, memory)
Queues: Limits
“module help sge”
Up to 112 cores (MPI)
Up to 32 cores shared memory (OpenMP)
Memory:
  up to 64GB per core
  up to 512GB for non MPI jobs
  up to 1024GB per job
Scratch up to 1,7TB
Execution time: 12 hours
   If needed ask for more resources in https://www.altausuarios.cesga.es/
Intel® Xeon® Processor E5-2600 Turbo Boost




1Max   Turbo Boost frequency based on number of 100 MHz increments above marked frequency (+1 = 0.100 GHz, +2 = 0.200 GHz, +3 = 0.300 GHz, etc. )
CESGA Supercomputers 2013



                              Shared storage: /home
                                      /store
                                   Linux O. S.
                           Grid Engine Batch Scheduler




           Finis Terrae (FT)                  Superordenador Virtual Gallego (SVG)
         Capability computing                    Throughput and Capacity computing
    Parallel jobs (>4 ... 1024 cores)                 Sequential & parallel jobs
                                                 up to 32 cores per node and 112 cores
     Huge memory (>4... 1024GB)                                   MPI
 Huge parallel scratch (>50... 10,000GB) Low-medium-large memory (up to 512GB!)
                                              Medium single node scratch (<1000GB)
                                                Customized clusters – Cloud services
Other Improvements
VPN for home connection
Storage:
  Do not use SFS from SVG
  Use “store”
High availability front-ends:
  svg.cesga.es

Más contenido relacionado

La actualidad más candente

Seastar @ SF/BA C++UG
Seastar @ SF/BA C++UGSeastar @ SF/BA C++UG
Seastar @ SF/BA C++UGAvi Kivity
 
Kubernetes and OpenStack at Scale
Kubernetes and OpenStack at ScaleKubernetes and OpenStack at Scale
Kubernetes and OpenStack at ScaleStephen Gordon
 
Containers on Baremetal and Preemptible VMs at CERN and SKA
Containers on Baremetal and Preemptible VMs at CERN and SKAContainers on Baremetal and Preemptible VMs at CERN and SKA
Containers on Baremetal and Preemptible VMs at CERN and SKABelmiro Moreira
 
K8s cluster autoscaler
K8s cluster autoscaler K8s cluster autoscaler
K8s cluster autoscaler k8s study
 
Using ansible to core os &amp; kubernetes clusters
Using ansible to core os &amp; kubernetes clustersUsing ansible to core os &amp; kubernetes clusters
Using ansible to core os &amp; kubernetes clustersmagicmarkup
 
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...Stephen Gordon
 
Moving from CellsV1 to CellsV2 at CERN
Moving from CellsV1 to CellsV2 at CERNMoving from CellsV1 to CellsV2 at CERN
Moving from CellsV1 to CellsV2 at CERNBelmiro Moreira
 
Heketi Functionality into Glusterd2
Heketi Functionality into Glusterd2Heketi Functionality into Glusterd2
Heketi Functionality into Glusterd2Gluster.org
 
Web後端技術的演變
Web後端技術的演變Web後端技術的演變
Web後端技術的演變inwin stack
 
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStack
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStackPeanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStack
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStackSean Cohen
 
HybridAzureCloud
HybridAzureCloudHybridAzureCloud
HybridAzureCloudChris Condo
 
10 Years of OpenStack at CERN - From 0 to 300k cores
10 Years of OpenStack at CERN - From 0 to 300k cores10 Years of OpenStack at CERN - From 0 to 300k cores
10 Years of OpenStack at CERN - From 0 to 300k coresBelmiro Moreira
 
Seastar / ScyllaDB, or how we implemented a 10-times faster Cassandra
Seastar / ScyllaDB,  or how we implemented a 10-times faster CassandraSeastar / ScyllaDB,  or how we implemented a 10-times faster Cassandra
Seastar / ScyllaDB, or how we implemented a 10-times faster CassandraTzach Livyatan
 
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and Ceph
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and CephProtecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and Ceph
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and CephSean Cohen
 
Ceph Performance and Optimization - Ceph Day Frankfurt
Ceph Performance and Optimization - Ceph Day Frankfurt Ceph Performance and Optimization - Ceph Day Frankfurt
Ceph Performance and Optimization - Ceph Day Frankfurt Ceph Community
 
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...NETWAYS
 
ELK: Moose-ively scaling your log system
ELK: Moose-ively scaling your log systemELK: Moose-ively scaling your log system
ELK: Moose-ively scaling your log systemAvleen Vig
 
Meet the Experts: InfluxDB Product Update
Meet the Experts: InfluxDB Product UpdateMeet the Experts: InfluxDB Product Update
Meet the Experts: InfluxDB Product UpdateInfluxData
 
Load balancing in the SRE way
Load balancing in the SRE wayLoad balancing in the SRE way
Load balancing in the SRE wayShawn Zhu
 
Designing HPC & Deep Learning Middleware for Exascale Systems
Designing HPC & Deep Learning Middleware for Exascale SystemsDesigning HPC & Deep Learning Middleware for Exascale Systems
Designing HPC & Deep Learning Middleware for Exascale Systemsinside-BigData.com
 

La actualidad más candente (20)

Seastar @ SF/BA C++UG
Seastar @ SF/BA C++UGSeastar @ SF/BA C++UG
Seastar @ SF/BA C++UG
 
Kubernetes and OpenStack at Scale
Kubernetes and OpenStack at ScaleKubernetes and OpenStack at Scale
Kubernetes and OpenStack at Scale
 
Containers on Baremetal and Preemptible VMs at CERN and SKA
Containers on Baremetal and Preemptible VMs at CERN and SKAContainers on Baremetal and Preemptible VMs at CERN and SKA
Containers on Baremetal and Preemptible VMs at CERN and SKA
 
K8s cluster autoscaler
K8s cluster autoscaler K8s cluster autoscaler
K8s cluster autoscaler
 
Using ansible to core os &amp; kubernetes clusters
Using ansible to core os &amp; kubernetes clustersUsing ansible to core os &amp; kubernetes clusters
Using ansible to core os &amp; kubernetes clusters
 
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...
Containers for the Enterprise: Delivering OpenShift on OpenStack for Performa...
 
Moving from CellsV1 to CellsV2 at CERN
Moving from CellsV1 to CellsV2 at CERNMoving from CellsV1 to CellsV2 at CERN
Moving from CellsV1 to CellsV2 at CERN
 
Heketi Functionality into Glusterd2
Heketi Functionality into Glusterd2Heketi Functionality into Glusterd2
Heketi Functionality into Glusterd2
 
Web後端技術的演變
Web後端技術的演變Web後端技術的演變
Web後端技術的演變
 
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStack
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStackPeanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStack
Peanut Butter and jelly: Mapping the deep Integration between Ceph and OpenStack
 
HybridAzureCloud
HybridAzureCloudHybridAzureCloud
HybridAzureCloud
 
10 Years of OpenStack at CERN - From 0 to 300k cores
10 Years of OpenStack at CERN - From 0 to 300k cores10 Years of OpenStack at CERN - From 0 to 300k cores
10 Years of OpenStack at CERN - From 0 to 300k cores
 
Seastar / ScyllaDB, or how we implemented a 10-times faster Cassandra
Seastar / ScyllaDB,  or how we implemented a 10-times faster CassandraSeastar / ScyllaDB,  or how we implemented a 10-times faster Cassandra
Seastar / ScyllaDB, or how we implemented a 10-times faster Cassandra
 
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and Ceph
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and CephProtecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and Ceph
Protecting the Galaxy - Multi-Region Disaster Recovery with OpenStack and Ceph
 
Ceph Performance and Optimization - Ceph Day Frankfurt
Ceph Performance and Optimization - Ceph Day Frankfurt Ceph Performance and Optimization - Ceph Day Frankfurt
Ceph Performance and Optimization - Ceph Day Frankfurt
 
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...
OpenNebula Conf 2014 | Building Hybrid Cloud Federated Environments with Open...
 
ELK: Moose-ively scaling your log system
ELK: Moose-ively scaling your log systemELK: Moose-ively scaling your log system
ELK: Moose-ively scaling your log system
 
Meet the Experts: InfluxDB Product Update
Meet the Experts: InfluxDB Product UpdateMeet the Experts: InfluxDB Product Update
Meet the Experts: InfluxDB Product Update
 
Load balancing in the SRE way
Load balancing in the SRE wayLoad balancing in the SRE way
Load balancing in the SRE way
 
Designing HPC & Deep Learning Middleware for Exascale Systems
Designing HPC & Deep Learning Middleware for Exascale SystemsDesigning HPC & Deep Learning Middleware for Exascale Systems
Designing HPC & Deep Learning Middleware for Exascale Systems
 

Destacado (7)

Can You Get Performance from Xeon Phi Easily? Lessons Learned from Two Real C...
Can You Get Performance from Xeon Phi Easily? Lessons Learned from Two Real C...Can You Get Performance from Xeon Phi Easily? Lessons Learned from Two Real C...
Can You Get Performance from Xeon Phi Easily? Lessons Learned from Two Real C...
 
CESGA: Curso Básico SVG
CESGA: Curso Básico SVGCESGA: Curso Básico SVG
CESGA: Curso Básico SVG
 
Fortissimo Enabling manufacturing SMEs to benefit from HPC
FortissimoEnabling manufacturing SMEs to benefit from HPCFortissimoEnabling manufacturing SMEs to benefit from HPC
Fortissimo Enabling manufacturing SMEs to benefit from HPC
 
A Web-platform for radiotherapy, a new workflow concept and an information sh...
A Web-platform for radiotherapy, a new workflow concept and an information sh...A Web-platform for radiotherapy, a new workflow concept and an information sh...
A Web-platform for radiotherapy, a new workflow concept and an information sh...
 
Energy Efficiency Policy at CESGA
Energy Efficiency Policy at CESGAEnergy Efficiency Policy at CESGA
Energy Efficiency Policy at CESGA
 
La Virtualización y el Cloud en el CESGA: Proyecto de Escritorios Virtuales e...
La Virtualización y el Cloud en el CESGA: Proyecto de Escritorios Virtuales e...La Virtualización y el Cloud en el CESGA: Proyecto de Escritorios Virtuales e...
La Virtualización y el Cloud en el CESGA: Proyecto de Escritorios Virtuales e...
 
Workshop on Fostering Innovation for Cyber-Physical Systems, Advanced Comput...
 Workshop on Fostering Innovation for Cyber-Physical Systems, Advanced Comput... Workshop on Fostering Innovation for Cyber-Physical Systems, Advanced Comput...
Workshop on Fostering Innovation for Cyber-Physical Systems, Advanced Comput...
 

Similar a Workshop actualización SVG CESGA 2012

Exaflop In 2018 Hardware
Exaflop In 2018   HardwareExaflop In 2018   Hardware
Exaflop In 2018 HardwareJacob Wu
 
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」PC Cluster Consortium
 
cachegrand: A Take on High Performance Caching
cachegrand: A Take on High Performance Cachingcachegrand: A Take on High Performance Caching
cachegrand: A Take on High Performance CachingScyllaDB
 
Accelerating hbase with nvme and bucket cache
Accelerating hbase with nvme and bucket cacheAccelerating hbase with nvme and bucket cache
Accelerating hbase with nvme and bucket cacheDavid Grier
 
Argonne's Theta Supercomputer Architecture
Argonne's Theta Supercomputer ArchitectureArgonne's Theta Supercomputer Architecture
Argonne's Theta Supercomputer Architectureinside-BigData.com
 
Theta and the Future of Accelerator Programming
Theta and the Future of Accelerator ProgrammingTheta and the Future of Accelerator Programming
Theta and the Future of Accelerator Programminginside-BigData.com
 
Accelerating HBase with NVMe and Bucket Cache
Accelerating HBase with NVMe and Bucket CacheAccelerating HBase with NVMe and Bucket Cache
Accelerating HBase with NVMe and Bucket CacheNicolas Poggi
 
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-GeneOpenStack Korea Community
 
The Power of HPC with Next Generation Supermicro Systems
The Power of HPC with Next Generation Supermicro Systems The Power of HPC with Next Generation Supermicro Systems
The Power of HPC with Next Generation Supermicro Systems Rebekah Rodriguez
 
lecture11_GPUArchCUDA01.pptx
lecture11_GPUArchCUDA01.pptxlecture11_GPUArchCUDA01.pptx
lecture11_GPUArchCUDA01.pptxssuser413a98
 
Modern network servers
Modern network serversModern network servers
Modern network serversAPNIC
 
Red Hat Gluster Storage Performance
Red Hat Gluster Storage PerformanceRed Hat Gluster Storage Performance
Red Hat Gluster Storage PerformanceRed_Hat_Storage
 
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...Tokyo Institute of Technology
 
OSS Presentation VMWorld 2011 by Andy Bennett & Craig Morgan
OSS Presentation VMWorld 2011 by Andy Bennett & Craig MorganOSS Presentation VMWorld 2011 by Andy Bennett & Craig Morgan
OSS Presentation VMWorld 2011 by Andy Bennett & Craig MorganOpenStorageSummit
 
Hadoop on a personal supercomputer
Hadoop on a personal supercomputerHadoop on a personal supercomputer
Hadoop on a personal supercomputerPaul Dingman
 
Introduction to DPDK
Introduction to DPDKIntroduction to DPDK
Introduction to DPDKKernel TLV
 
AI Accelerators for Cloud Datacenters
AI Accelerators for Cloud DatacentersAI Accelerators for Cloud Datacenters
AI Accelerators for Cloud DatacentersCastLabKAIST
 

Similar a Workshop actualización SVG CESGA 2012 (20)

Exaflop In 2018 Hardware
Exaflop In 2018   HardwareExaflop In 2018   Hardware
Exaflop In 2018 Hardware
 
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」
PCCC23:筑波大学計算科学研究センター テーマ1「スーパーコンピュータCygnus / Pegasus」
 
cachegrand: A Take on High Performance Caching
cachegrand: A Take on High Performance Cachingcachegrand: A Take on High Performance Caching
cachegrand: A Take on High Performance Caching
 
Accelerating hbase with nvme and bucket cache
Accelerating hbase with nvme and bucket cacheAccelerating hbase with nvme and bucket cache
Accelerating hbase with nvme and bucket cache
 
Argonne's Theta Supercomputer Architecture
Argonne's Theta Supercomputer ArchitectureArgonne's Theta Supercomputer Architecture
Argonne's Theta Supercomputer Architecture
 
Theta and the Future of Accelerator Programming
Theta and the Future of Accelerator ProgrammingTheta and the Future of Accelerator Programming
Theta and the Future of Accelerator Programming
 
Accelerating HBase with NVMe and Bucket Cache
Accelerating HBase with NVMe and Bucket CacheAccelerating HBase with NVMe and Bucket Cache
Accelerating HBase with NVMe and Bucket Cache
 
GIST AI-X Computing Cluster
GIST AI-X Computing ClusterGIST AI-X Computing Cluster
GIST AI-X Computing Cluster
 
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene
[OpenStack Days Korea 2016] Track3 - OpenStack on 64-bit ARM with X-Gene
 
The Power of HPC with Next Generation Supermicro Systems
The Power of HPC with Next Generation Supermicro Systems The Power of HPC with Next Generation Supermicro Systems
The Power of HPC with Next Generation Supermicro Systems
 
QNAP TS-832PX-4G.pdf
QNAP TS-832PX-4G.pdfQNAP TS-832PX-4G.pdf
QNAP TS-832PX-4G.pdf
 
lecture11_GPUArchCUDA01.pptx
lecture11_GPUArchCUDA01.pptxlecture11_GPUArchCUDA01.pptx
lecture11_GPUArchCUDA01.pptx
 
Modern network servers
Modern network serversModern network servers
Modern network servers
 
Red Hat Gluster Storage Performance
Red Hat Gluster Storage PerformanceRed Hat Gluster Storage Performance
Red Hat Gluster Storage Performance
 
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...
Applying Recursive Temporal Blocking for Stencil Computations to Deeper Memor...
 
OSS Presentation VMWorld 2011 by Andy Bennett & Craig Morgan
OSS Presentation VMWorld 2011 by Andy Bennett & Craig MorganOSS Presentation VMWorld 2011 by Andy Bennett & Craig Morgan
OSS Presentation VMWorld 2011 by Andy Bennett & Craig Morgan
 
Hadoop on a personal supercomputer
Hadoop on a personal supercomputerHadoop on a personal supercomputer
Hadoop on a personal supercomputer
 
Advances in GPU Computing
Advances in GPU ComputingAdvances in GPU Computing
Advances in GPU Computing
 
Introduction to DPDK
Introduction to DPDKIntroduction to DPDK
Introduction to DPDK
 
AI Accelerators for Cloud Datacenters
AI Accelerators for Cloud DatacentersAI Accelerators for Cloud Datacenters
AI Accelerators for Cloud Datacenters
 

Más de CESGA Centro de Supercomputación de Galicia

FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...
FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...
FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...CESGA Centro de Supercomputación de Galicia
 

Más de CESGA Centro de Supercomputación de Galicia (13)

Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2
 
Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2
 
Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2
 
Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2Jornada convocatoria experimentos H2020 FORTISSIMO2
Jornada convocatoria experimentos H2020 FORTISSIMO2
 
Novedades de gestión del H2020
Novedades de gestión del H2020Novedades de gestión del H2020
Novedades de gestión del H2020
 
Spatial data infraestructure ID-Patri
Spatial data infraestructure ID-PatriSpatial data infraestructure ID-Patri
Spatial data infraestructure ID-Patri
 
CLOUDPYME: Servicios en Cloud para la PYMEs innovadoras
CLOUDPYME: Servicios en Cloud para la PYMEs innovadorasCLOUDPYME: Servicios en Cloud para la PYMEs innovadoras
CLOUDPYME: Servicios en Cloud para la PYMEs innovadoras
 
21 anos en apoyo de la investigación en Galicia
21 anos en apoyo de la investigación en Galicia21 anos en apoyo de la investigación en Galicia
21 anos en apoyo de la investigación en Galicia
 
CloudPYME: Cloud para empresas que Innovan
CloudPYME: Cloud para empresas que InnovanCloudPYME: Cloud para empresas que Innovan
CloudPYME: Cloud para empresas que Innovan
 
HP E INTEL CONMEMORAN LOS 20 AÑOS DEL CENTRO DE SUPERCOMPUTACIÓN DE GALICIA
 HP E INTEL CONMEMORAN LOS 20 AÑOS DEL CENTRO DE SUPERCOMPUTACIÓN DE GALICIA HP E INTEL CONMEMORAN LOS 20 AÑOS DEL CENTRO DE SUPERCOMPUTACIÓN DE GALICIA
HP E INTEL CONMEMORAN LOS 20 AÑOS DEL CENTRO DE SUPERCOMPUTACIÓN DE GALICIA
 
FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...
FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...
FUJITSU celebra en Santiago los 20 años de la instalación de su primer Superc...
 
Workshop actualización SVG CESGA 2012. Aplicaciones
Workshop actualización SVG CESGA 2012. AplicacionesWorkshop actualización SVG CESGA 2012. Aplicaciones
Workshop actualización SVG CESGA 2012. Aplicaciones
 
DATA CENTERS 2012 Eficacia, Disponibilidad y Seguridad
DATA CENTERS 2012 Eficacia, Disponibilidad y SeguridadDATA CENTERS 2012 Eficacia, Disponibilidad y Seguridad
DATA CENTERS 2012 Eficacia, Disponibilidad y Seguridad
 

Workshop actualización SVG CESGA 2012

  • 2. AGENDA •Upgraded SVG • Motivation • Hardware configuration • Infiniband Network • Environment • Queues configuration • Benchmarks •CESGA Supercomputers in 2013 • Distribution of jobs
  • 3. Hardware Thin-nodes Total Total 8 HP SL230s Gen8 each with 24 24 Intel Sandy CPUs Intel Sandy Bridge Bridge processors 2x Intel Xeon E5-2670, 8 cores each, 2.6GHz 192 cores 192 cores 1,5 TB memory 64GB main memory DDR3-1600MHz 1,5 TB memory 28 TB disk 2TB SAS hard disk 28 Peak Performance 3788 GFLops TB disk 2x1GbE 3788 Gflops peak performance 2xInfiniband FDR 56Gb Peak Performance 332GFlops Fat-nodes 2 HP DL560 Gen8 each with 4x Intel Xeon E5-4620, 8 cores each, 2.2GHz 512GB main memory DDR3-1600MHz 6 hard disks each 1TB 4x1GbE Infiniband FDR 56Gb 10GbE Peak Performance 563 GFlops
  • 4. Motivation Target • Competitive solution for: • Parallel MPI & OpenMP applications • Memory Intensive • Alternative for Finis Terrae • Lower cost of operation and maintenance • Finis Terrae II prototype • To define new requirements
  • 5. Infiniband network Mellanox SX6036 switch 36 port FDR 56Gb/s 4Tb/s aggregated non-blocking BW 1 microseconds MPI latency Dual connection: High availability – same BW
  • 6. Environment Integrated in the SVG cluster: Scientific Linux 6.3 (Red Hat) Common /opt/cesga Common /home, stores… Same gateway: svg.cesga.es Interactive use: compute –arch sandy Jobs: qsub –arch sandy Binary compatible - no need to recompile
  • 7. Usage compute –arch amd qsub –arch amd svg.cesga.es compute SSH qsub compute –arch sandy qsub –arch sandy
  • 8. Configuration Full production phase (November 2012) • Only runs jobs with –sandy option • General availability of applications specifically compiled • Maximum wall-clock time 12 hours • Maximum 2 jobs per node fat nodes Under consideration near future: Jobs without –sandy option Higher wall-clock time
  • 9. Queues Configuration Exclusive nodes To take advantage of Infiniband To take advantage of Turboboost Jobs not interferring each other Maximum performance Maximum 2 jobs on Fat nodes 32 cores nodes Exclusive if required by the jobs (cores, memory)
  • 10. Queues: Limits “module help sge” Up to 112 cores (MPI) Up to 32 cores shared memory (OpenMP) Memory: up to 64GB per core up to 512GB for non MPI jobs up to 1024GB per job Scratch up to 1,7TB Execution time: 12 hours If needed ask for more resources in https://www.altausuarios.cesga.es/
  • 11. Intel® Xeon® Processor E5-2600 Turbo Boost 1Max Turbo Boost frequency based on number of 100 MHz increments above marked frequency (+1 = 0.100 GHz, +2 = 0.200 GHz, +3 = 0.300 GHz, etc. )
  • 12. CESGA Supercomputers 2013 Shared storage: /home /store Linux O. S. Grid Engine Batch Scheduler Finis Terrae (FT) Superordenador Virtual Gallego (SVG) Capability computing Throughput and Capacity computing Parallel jobs (>4 ... 1024 cores) Sequential & parallel jobs up to 32 cores per node and 112 cores Huge memory (>4... 1024GB) MPI Huge parallel scratch (>50... 10,000GB) Low-medium-large memory (up to 512GB!) Medium single node scratch (<1000GB) Customized clusters – Cloud services
  • 13. Other Improvements VPN for home connection Storage: Do not use SFS from SVG Use “store” High availability front-ends: svg.cesga.es