Imagine Folding on THIS!

[BigIan88]

Boasting a 1,008 Intel core cluster, which we're told is the equivalent of 1,000 desktop systems, the supercomputer will simulate vast regions of the Universe. Supplied by Dell and operated by the University's Institute of Cosmology and Gravitation, the 1,008 core machine is a distributed memory cluster that uses an open source software stack powered by 2.66GHz Intel Xeon processors, each with 2GB of memory per core.

The machine also has 85TB of fast parallel storage and 10TB of NFS storage. Try and build that in your living room. Capable of doing a measly billion calculations each second, Sciama will investigate the properties of hundreds of millions of galaxies and solve complex cosmological problems. Maybe it will even have a crack at answering the question, what is the meaning of life?

Without a hint of modesty, Portsmouth's Sciama manager Gary Burton said, "Using it will allow us to explore the whole of cosmic history and analyse data that contains fundamental clues about the origins of the Universe."

To achieve this feat of hyperbole and the exploration of the whole of cosmic history, Sciama will crunch observational data coming from satellites, telescopes and other detectors.

The supercomputer was named after Dennis Sciama, a leading figure internationally in the development of astrophysics and cosmology. The name is also an acronym for SEPnet Computing Infrastructure for Astrophysical Modelling and Analysis.

But we think Portsmouth senior research fellow, Dr David Bacon, said it best. "We're absolutely thrilled to have Sciama...to make real breakthroughs...and by calculating the consequences of mind-boggling new theories." Mind boggling, indeed.

Source - http://www.theinquirer.net/inquirer...822/mind-boggling-tackles-universal-questions

 

[adicool96]

Something like a billion points per minute!!!!

 

[wotevajjjj]

it's not overclocked so i don't like it

 

[lasher]

Tut we already know the meaning of life ... hitchikers anyone??

Bloody cosmic reasearch who give a .....

Used for medical research it could potentially do a lot more, what a sodding waste of money, energy and poeples time.

 

[adicool96]

Tut we already know the meaning of life ... hitchikers anyone??

Bloody cosmic reasearch who give a .....

Used for medical research it could potentially do a lot more, what a sodding waste of money, energy and poeples time.

Maybe they could discover aliens who have the cure for cancer and aid's and other uncurable diseases

. Li

 

[thestepster]

oh thats cool i love space stuff always fascinated me

 

[BigIan88]

anyone fancy emailing them and asking them to get it folding for 48 hours? im pretty sure it would be in the top 10 within that timeframe!

 

[thestepster]

u try setting up a few hundred clients

 

[BigIan88]

might take a little while to sort out but the payoff is significant in comparison

 

[Xochielt]

Just a toy!

1,008 cores @ 2,66GHz and 2GB memory per core... boring

I maintain the VSC.

Nearly 4,000 nehalem-cores @ 2,66GHz and 3GB memory per core

Turbo-mode up to 2,93GHz.

Just to imagine what this 'rig' can do:

During the time a jet aircraft needs to travel a millimeter, our

VSC can sum up 90.000.000 numbers.

About 30/35 TFlops of pure calculation-power...

And even this beast is listed worse 350...

Just a few impressions:

The boxes with the SunBlades

Right and left the racks with the servers

4/16 racks with the servers

Would be really interesting, how much PPD this THING can

push...

 

[FoodStampz]

I wonder what the A/C is set to in that room?

 

[Xochielt]

I wonder what the A/C is set to in that room?

Set to 18°C - cooling down to sth. like 22°C/23°C. Have to mention,

that the small racks after two serverracks are liquid cooling. There

is one built in transferring the heat from the 2nd floor where the

VSC is to the ( 13th? 14th? ) where a cooling is.

Inside the Servers the air reaches about 40°C.

 

[hmmblah]

Set to 18°C - cooling down to sth. like 22°C/23°C. Have to mention,

that the small racks after two serverracks are liquid cooling. There

is one built in transferring the heat from the 2nd floor where the

VSC is to the ( 13th? 14th? ) where a cooling is.

Inside the Servers the air reaches about 40°C.

Keep talking dirty!

What kind of work does that cluster do?

 

[Xochielt]

The VSC was built to satisfy the High Performance Computing

demands of Viennas three biggest Universities, which are:

University of Vienna, Vienna University of Technology and

University of Natural Resources and Applied Life Science Vienna.

The installed software is:

Compilers:

• Intel Fortran & C
  
• GNU Fortran G77
  
• gfortran
  
• gcc
  
• Portland Group Comiler
  
• Glasgow Haskell Compiler
  

Applications:

• GAUSSIAN 09
  
• R
  
• MATLAB
  
• Mathematica
  
• HDF5
  
• Global Arrays
  
• LAMMPS
  

Parallel Computing:

• QLogic MPI 0.1.0
  
• OpenMPI 1.3.2
  

I do not really have a clue what they do exactly with

this thing...

...but when you think this was dirty - listen:

Since today ( 02.02.2011 ) it is official:

Vienna goes performance computing!

New cluster in Vienna got announced today. The so called

'VSC-2' ( the 'old' VSC is now called 'VSC-1' ) is planned

to be finished within Q2/2011. The VSC-1 stays where it is,

the VSC-2 comes additional to it.

Specs ( this is dirty! ):

1.314 nodes ( 476@VSC-1 )

2 8-core AMD Opteron 6132 HE @2,2 GHz per node

21,024 cores total ( 3,968@VSC-1 )

internal connection: 32GBit/s

~150TFlops ( 35@VSC-1 )

costs more than 4.000.000€

This one is going to rock!

But we currently own two more clusters.

IBM Cluster:

• 54 nodes
  
• 2 IBM Power5+ @1.9GHz per node
  
• 16GB memory per node
  
• 73GB SCSI HDD per node
  
• 1 Access-node with 8 IBM Power5+ @1.5GHz, 32GB memory, 2x146GB SCSI + 2x300GB SCSI HDD
  
• 1 Fileserver with 2 IBM Power5+ @1.9GHz, 16GB memory, 2x146GB SCSI + 2x300GB SCSI HDD
  

Phoenix Linux Cluster:

• 67 Sun V20z nodes with 2 AMD Opteron 250 @2.4GHz, 4GB memory, 73GB SCSI HDD
  
• 11 Sun X4100 nodes with 2 AMD Opteron 2200 @2.8GHz DualCore, 8GB memory, 73GB SAS HDD
  
• 4 Sun X4600 nodes with 4 AMD Opteron 8220 @2.8GHz DualCore, 64GB memory, 114GB SAS HDD
  
• 1 Sunfire V40z frontend node with 2 AMD Opteron 248, 4GB memory, 73GB mirrored HDD
  
• 1 Storage node Sun X4500 with 48x 500 GB SATA HDD
  

Running F@H at Phoenix, IBM, VSC-1 and VSC-2 for a week would be

overkill

Oh yeah - and here a pic of the IBM-cluster:

That is geekporn - isn't it?

Oh - and a short question:

Who has to maintain this new VSC-2-beast without getting more money

monthly or being employeed more hours per week? I think you got the answer

 

[SnW]

Good luck setting it up

 

[stevehat1]

Good luck setting it up

I haven't used it yet, but it sure looks like a job for the GPU Tracker to me