Recently the HPCf participated in a NIH proposal submitted to the BRIN program. This project was fully funded in September 2001.

More information can be found at http://brin.hpcf.upr.edu/

With this project, we committed ourselves to expand the High Performance Computing facility with the intent of creating the Bioinformatics Resource Center (BiRC).

This facility will contain high-end hardware, software and services essential to all researchers and educators in biomedical sciences at Puerto Rican institutions, with specific tools to support the following three research clusters: Neurobiology, Mental Health and Medical Biotechnology.

The areas we consider as essential to be developed and supported at the Bioinformatics Resource Center are:

1. genomic and proteomic databases
2. sequence analysis software
3. phylogeny software
4. protein structure prediction and visualization
5. bioinformatics programming
6. microarray data visualization and analysis
7. biostatistics
8. research support services and training

In order to alleviate the accessibility problems of CPU cycles for the bioinformatics community, we have proposed to deploy a second supercomputing server totally dedicated to the bioinformatics community.

We recently deployed a Silicon Graphics Origin 300 server with the following characteristics:

• 32 MIPS R14000 600MHz processors
• 24 Gigabytes of system memory
• 306 Gigabytes of UltraSCSI disk storage
• Gigabit Ethernet network interfaces

The name of this server is cafeina.hpcf.upr.edu. To access this machine, one must first fill out an allocation request form, found in the "User Info" section of our website.

In choosing this platform, we insure that we continue to capitalize with the expertise we have been developing in the past years in administrating SGI supercomputers. We also insure that the SGI-specific services continue for the full benefit of the HPCf user community.

To complement this ?traditional? supercomputing server, we plan to also deploy a Linux cluster consisting initially of 78 dual-processor AMD 1800+ (1.533 GHz) nodes, each with 1024 Megabytes of memory and a 20 Gigabytes IDE hard drive. A master node consisting of a dual-processor AMD 1800+ computer with 1024 Megabytes of memory and a total of 600Gigabytes of storage, will be controlling the Linux cluster.This cluster will then offer a total of 2.16 Terabytes of on-line storage and 79 Gigabytes of distributed memory.

The storage and backup facility will be upgraded by the implementation of a hierarchical management system, which will handle the dynamic distribution of users? files depending upon the frequency of access.

Three types of storage will be present

• 1.2 Terabyte of FiberChannel storage on the SGI supercomputers
• Linux-based mirrored RAID system with 4.4 Terabytes capacity, and
• the DLT storage through the StorageTek tape robot.