SNIC Documentation - User contributions [en]

Alya performance benchmark

2013-04-24T11:48:37Z

Jing Gong (PDC):

{{project info
|description=Alya performance benchmark
|fields=Computational fluid dynamics
|financing=PRACE
|active=yes
|start date=2012-06-01
|end date=2013-08-01
}}

The Alya system is a high performance computational mechanics code developed at the Barcelona Supercomputing Center (BSC). Alya is designed from scratch to take advantage of parallel architectures.

However, it is necessary to discover performance bottlenecks originating from the increase of complexity of each level of concurrency and to correct them in the source code Alya. Together with the main developer (Dr. Guillaume Houzeaux) we will use the performance analysis tools (Vampir and TAU) to investigate the performance and scalability of Alya and do some benchmarks with Alya on different computer platforms.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

User:Jing Gong (PDC)

2013-04-17T09:43:36Z

Jing Gong (PDC):

{{application expert info
|first name=Jing
|last name=Gong
|centre=PDC
|fields=Computational fluid dynamics
|fte=100
|snic ae financing=0
|image=Jing_Gong.png
|financing=SeRC,CRESTA
|office= PDC; Teknikringen 14 ;100 44 Stockholm
|general activities=Application expert
}}



== Quick facts ==
* Works at PDC since 2012
* Ph.D. 2007 in Scientific Computing, Uppsala University
* Working on CFD

== Expertise ==
* [[expertise::C++]]
* [[expertise::MPI]]
* [[expertise::ANSYS]]
* [[expertise::Fluent]]

== Projects ==
* [[project::Alya performance benchmark]]
* [[project::Nek5000 OpenACC]]

User:Jing Gong (PDC)

2012-09-05T09:10:42Z

Jing Gong (PDC): /* Projects */

{{application expert info
|first name=Jing
|last name=Gong
|centre=PDC
|fields=Computational fluid dynamics
|fte=100
|snic ae financing=0
|image=Jing_Gong.png
|financing=SeRC
|office= PDC; Teknikringen 14 ;100 44 Stockholm
|general activities=Application expert within SeRC.
}}



== Quick facts ==
* Works at PDC since 2012
* Ph.D. 2007 in Scientific Computing, Uppsala University
* Working on CFD

== Expertise ==
* [[expertise::C++]]
* [[expertise::MPI]]
* [[expertise::ANSYS]]
* [[expertise::Edge]]
* [[expertise::Fluent]]

== Projects ==
* [[project::Alya performance benchmark]]
* [[project::Nek5000 BC]]
* [[project::Nek5000 OpenACC]]

Nek5000 OpenACC

2012-09-04T19:31:23Z

Jing Gong (PDC):

{{project info
|description=Nek5000 with OpenACC
|fields=Computational fluid dynamics
|financing=CRESTA
|active=yes
|start date=2012-05-01
|end date=
}}

The co-design team OpenACC in Nek5000 aims to enable Nek5000 efficiently to utilize the massively parallel GPU accelerated system based on OpenACC directive. Nek5000 is an open-source code for the simulation of incompressible flow in complex geometries. Nek5000 is written in mixed Fortran77/C and designed to employ fully large-scale parallelism.

In the present state, Nek5000 does not employ any hybrid approach to parallelization. All communication is handled by MPI, which has proven to be very efficient, mainly due to the element structure of the mesh [3]. In [4] a Lattice Boltzmann code has been successfully adapted to use a large number of GPUs in parallel. In order to adapt the Nek5000 code for the exascale computation, a hybrid approach, that is, C with MPI for communication and OpenACC for GPU acceleration, should be considered.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

Nek5000 OpenACC

2012-09-04T19:31:06Z

Jing Gong (PDC): Created page with "{{project info |description=Nek5000 with OpenACC |fields=Computational fluid dynamics |financing=SeRC |active=yes |start date=2012-05-01 |end date= }} The co-design team OpenACC..."

{{project info
|description=Nek5000 with OpenACC
|fields=Computational fluid dynamics
|financing=SeRC
|active=yes
|start date=2012-05-01
|end date=
}}

The co-design team OpenACC in Nek5000 aims to enable Nek5000 efficiently to utilize the massively parallel GPU accelerated system based on OpenACC directive. Nek5000 is an open-source code for the simulation of incompressible flow in complex geometries. Nek5000 is written in mixed Fortran77/C and designed to employ fully large-scale parallelism.

In the present state, Nek5000 does not employ any hybrid approach to parallelization. All communication is handled by MPI, which has proven to be very efficient, mainly due to the element structure of the mesh [3]. In [4] a Lattice Boltzmann code has been successfully adapted to use a large number of GPUs in parallel. In order to adapt the Nek5000 code for the exascale computation, a hybrid approach, that is, C with MPI for communication and OpenACC for GPU acceleration, should be considered.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

Nek5000 BC

2012-09-04T19:27:51Z

Jing Gong (PDC): Created page with "{{project info |description=Boundary conditions for Nek5000 |fields=Computational fluid dynamics |financing=SeRC |active=yes |start date=2012-02-01 |end date= }} Nek5000 is writ..."

{{project info
|description=Boundary conditions for Nek5000
|fields=Computational fluid dynamics
|financing=SeRC
|active=yes
|start date=2012-02-01
|end date=
}}

Nek5000 is written in mixed Fortran77/C and designed to employ fully larger-scale parallelism. The code has a long history of HPC development. Recently the larger-scale simulations were successful performed on the Cray XE6 system at PDC, KTH with 32,768 cores (“Coherent structures and dominant frequencies in a turbulent three-dimensional diffuser” by J. Malm, P. Schlatter and D. S. Henningson, accepted by JFM) and on the IBM BG/P Eugene with 262144 cores (“Exterme Scaling Worskop 2010 Report”)

In the project, we investigate the boundary conditions in Nek5000 for exascale computing. The definition of boundary conditions requires special attention, especially for the turbulent boundary layers in the exascale computations. Because only a part of the turbulent spectrum is resolved in the spectral element methods, the challenge is to regenerate and preserve the turbulence at the boundaries. At the inflow boundary, the challenge is to prescribe transient turbulent velocity profiles from DNS data. Simply adding random fluctuations to the DNS profiles lacks the temporal and spatial correlation of real turbulence. The fluctuations must be pre-computed and stored in a database or computed on the fly from an auxiliary computation.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

Alya performance benchmark

2012-09-04T19:18:40Z

Jing Gong (PDC):

{{project info
|description=Alya performance benchmark
|fields=Computational fluid dynamics
|financing=PRACE
|active=yes
|start date=2012-06-01
|end date=2013-05-01
}}

The Alya system is a high performance computational mechanics code and developed at the Barcelona Supercomputing Center (BSC). Alya is designed from scratch to take profit of parallel architectures.

However, it is necessary to discover performance bottlenecks originating from the increase of complexity of each level of concurrency and to correct them in the source code Alya. Together with the main developer (Dr. Guillaume Houzeaux) we will use the performance analysis tools (Vampir and TAU) to investigate the performance analysis and scalability of Alya and do some benchmarks for Alya on different computer platforms.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

Alya performance benchmark

2012-08-31T11:40:24Z

Jing Gong (PDC):

{{project info
|description=Alya performance benchmark
|fields=Computational fluid dynamics
|financing=PRACE
|active=yes
|start date=2012-06-01
|end date=2013-05-01
}}

The Alya System is a Computational Mechanics (CM) code with two main features. Firstly, it is specially designed for running with the highest efficiency standards in large scale supercomputing facilities. Secondly, it is capable of solving different physics, each one with its own modelization characteristics, in a coupled way.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

Alya performance benchmark

2012-08-31T11:37:40Z

Jing Gong (PDC): /* Members */

{{project info
|description=Alya performance benchmark
|fields=Computational fluid dynamics
|financing=PRACE
|active=yes
|start date=2012-06-01
|end date=2013-05-01
}}

The Alya System is a Computational Mechanics (CM) code with two main features. Firstly, it is specially designed for running with the highest efficiency standards in large scale supercomputing facilities. Secondly, it is capable of solving different physics, each one with its own modelization characteristics, in a coupled way.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|PDC
|Application expert
|Computational fluid dynamics
}}

Alya performance benchmark

2012-08-31T11:35:23Z

Jing Gong (PDC): Created page with "{{project info |description=Alya performance benchmark |fields=Computational fluid dynamics |financing=PRACE |active=yes |start date=2012-06-01 |end date=2013-05-01 }} The Alya ..."

{{project info
|description=Alya performance benchmark
|fields=Computational fluid dynamics
|financing=PRACE
|active=yes
|start date=2012-06-01
|end date=2013-05-01
}}

The Alya System is a Computational Mechanics (CM) code with two main features. Firstly, it is specially designed for running with the highest efficiency standards in large scale supercomputing facilities. Secondly, it is capable of solving different physics, each one with its own modelization characteristics, in a coupled way.

== Members ==
{{#ask: [[Category:Person]] [[project::{{PAGENAME}}]]
|?centre
|?role
|?field
}}

File:Jing Gong.png

2012-01-25T19:53:48Z

Jing Gong (PDC): uploaded a new version of "File:Jing Gong.png"

User:Jing Gong (PDC)

2012-01-24T12:16:45Z

Jing Gong (PDC):

{{application expert info
|first name=Jing
|last name=Gong
|centre=PDC
|fields=Computational fluid dynamics
|fte=100
|snic ae financing=100
|image=Jing_Gong.png
|financing=SNIC
|office= PDC; Teknikringen 14 ;100 44 Stockholm
}}



== Quick facts ==
* Work at PDC since 2012
* Ph.D. 2007 in Scientific Computing, Uppsala University
* Working on CFD

== Expertise ==
* [[expertise::C++]]
* [[expertise::MPI]]

File:Jing Gong.png

2012-01-24T12:16:17Z

Jing Gong (PDC):

User:Jing Gong (PDC)

2012-01-24T12:09:46Z

Jing Gong (PDC):

{{application expert info
|first name=Jing
|last name=Gong
|centre=PDC
|fields=Computational fluid dynamics
|fte=100
|snic ae financing=100
|image=Jing_Gong.JPG
|financing=SNIC
|office= PDC; Teknikringen 14 ;100 44 Stockholm
}}



== Quick facts ==
* Work at PDC since 2012
* Ph.D. 2007 in Scientific Computing, Uppsala University
* Working on CFD

== Expertise ==
* [[expertise::C++]]
* [[expertise::MPI]]

User:Jing Gong (PDC)

2012-01-19T07:09:14Z

Jing Gong (PDC):

{{application expert info
|first name=Jing
|last name=Gong
|centre=PDC
|fields=Computational fluid dynamics
|fte=100
|snic ae financing=100
|financing=SNIC
|office= PDC; Teknikringen 14 ;100 44 Stockholm
}}



== Quick facts ==
* Work at PDC since 2012
* Ph.D. 2007 in Scientific Computing, Uppsala University
* Working on CFD

== Expertise ==
* [[expertise::C++]]
* [[expertise::MPI]]