jueves, 23 de febrero de 2012

4th Week

For this week I didn't do much I looked for some algorithms to test the cluster efficiency, I found this tutorial which includes some steps to test an algorithm that its also included in a file, what the algorithm basically do is calculates an approximate value for pi using a Monte Carlo technique. So I tried following the steps it mention but when it ask for the password it doesn't work I don't know why. The tutorial tells to open this for when u can run the algorithm u can see the time it takes and all the test results.

For next week I hope me and some other members of the algorithm team can make this algorithm work or find another one or maybe do one to have it ready and test the cluster in some weeks.

For lab: I found a video of a cluster made by some guy while he is testing it,he uses a algorithm of prime numbers to test it and some 3d images like the ones we saw at class.

jueves, 16 de febrero de 2012

3rd Week

For my contribution in this 3rd week me and some of the members of the application team have been talking about the type of applications to work in, we haven't decide yet we will be working on that in this week, the idea is to divide all the members of the parallel applications group in little groups of 3 and each group work in a different area and then vote for the best application and all the groups will work on it. As we haven't decide yet I looked for some examples and I found one that I think its really interesting its called Yabi.

Wiki link: Yabi

I didnt know how to post the video of Yabi in the wiki so i post it here.

jueves, 9 de febrero de 2012

Semana 2

For this week contribution I looked for some examples of parallel algorithms. I also did a small introduction for the parallel algorithm section in the main page.
Here is the link for the wiki:

miércoles, 8 de febrero de 2012

Columbia Supercomputer from NASA

As a contribution for this week in lab ill talk about one example of an important supercomputer. The name of this supercomputer is Columbia from the NASA named this way in honor to the crew who died in Columbia disaster; it may not be the most powerful computer nowadays but it’s really important for the history of NASA because with this one they achieved a lot of breakthroughs in science. It was original installed in 2004 and in that year it was ranked in second place of the top most powerful supercomputers. Nowadays it’s being replaced with the new Pleiades Supercomputer which its ranked in 7th place.

The Columbia System Architecture includes:
System Architecture

• Manufacturer: SGI
• 40 compute node cabinets
• 30 Tflop/s theoretical peak (original 10,240 system: 63 Tflop/s)
• Total Memory: 9 TB
• Nodes
o 1 SGI® Altix® 4700 (512 cores)
o Intel® Itanium® processors (Montecito)
o 1.6 GHz
o 9MB
o 1 SGI Altix 4700 (2,048 cores)
o Intel Itanium processors (Montecito)
o 1.6 GHz
o 9 MB
o 2 SGI Altix 4700 (1,024 cores)
o Intel Itanium processors (Montvale)
o 1.6 GHz
o 9 MB
o 4 total compute nodes (4,608 total cores)
o Memory Type: Double data rate synchronous dynamic random access memory (DDR SDRAM)
o 2 GB per core

• 1 front-end node

• SGI® NUMAlink® interconnected single-system image compute nodes (enable large SSI)
• Internode
o InfiniBand®: 4x (Single Data Rate, Double Data Rate)
o 10 Gb Ethernet LAN/WAN interconnect
o 1 Gb Ethernet LAN/WAN interconnect

• Online: DataDirect Networks® and LSI® RAID, 800 TB (raw)
o 1 SGI CXFS domain
o Local SGI XFS
• Archival: Attached to high-end computing SGI CXFS storage area network filesystem
Operating Environment

• Operating system: SUSE® Linux® Enterprise
• Job scheduler: PBS®
• Compilers - C, Intel Fortran, SGI MPT

For more information or if you want to know about the new NASA supercomputer you can look at: http://www.nas.nasa.gov/hecc/resources/columbia.html