Remote Access to MPI+SMP+GPU-clusters in a Grid Environment

Tutorial Chair: Alexandre Santos, Univ. Minho

Goals: to engage young researchers in Computational Eng & Science to listen experts' talks in Grid and Cluster computing during the morning, and have hands-on experimentation on the remote access to heterogeneous resources during the afternoon. This tutorial will mainly address grid technologies, namely at our national INGRID (EGEE-based) and at Ibergrid level, and different cluster technologies.

Agenda
Morning (talks in an auditorium)
09h00-09h45	Access to resources in the Grid environment INGRID + EGEE + IBERGRID (focus on grid issues, lead by Jorge Gomes, LIP, Portugal)
09h45-10h30	Access to a thin-node MPI cluster in a Grid environment (focus on the central cluster node at INGRID, lead by Jorge Gomes, LIP, Portugal)
10h30-11h00	Coffee break
11h00-11h45	Access to a fat-node SMP cluster: Finis Terrae (focus on the Ibergrid access to Superdome in FinisTerrae, SMP Itanium nodes with 1TB RAM; lead by Carlos Fernandez, CESGA, Spain)
11h45-12h30	Access to remote visualization and data analisys HPC resources (GPU-cluster) (focus on Longhorn, the largest GPU-cluster in the world for visualization and data analysis; lead by Paul Navratil, TACC, USA)
Afternoon (Hands-on, including examples brought by participants; parallel sessions)
14h30-16h00	Parallel Sessions P1. Hands-on MPI-cluster: main cluster in INGRID P2. Hands-on SMP-cluster: Finis Terrae at CESGA P3. Hands-on GPU-cluster: Longhorn at TACC
16h00-16h30	Coffee break
16h30-18h00	Parallel Sessions P1. Hands-on MPI-cluster: main cluster in INGRID P2. Hands-on SMP-cluster: Finis Terrae at CESGA P3. Hands-on GPU-cluster: Longhorn at TACC