You are here

Research Seminar Series

We run a DIR Seminar Series where group members, visitors and guests present their work followed by lively discussions. The seminars are run every Friday at 10am in the Informatics Forum. The organiser of the series is Dr Rosa Filgueira Vicente.

We also list talks relevant to our work in the seminar series from the Centre for Intelligent Systems and their Applications (CISA), the Software Systems and Processes research group (SPP).

Regaining Control: Multi-Level Scheduling, Pilot-Jobs and a Fresh Perspective on Application Optimisation.

Pilot-jobs are an effective and widely-used concept to realise multi-level scheduling and user-space resource management in heterogeneous distributed environments. In the first part of our talk, we will give an overview of multi-level scheduling and pilot-jobs and discuss examples of how and under which circumstances different types of distributed applications can benefit from these. In the second part of our talk, we will introduce our ongoing research ideas by presenting a fresh perspective on transparent and automated application optimization on workload level.

13 Apr 2012

Database Wiki

Wikis have become popular as lightweight Web-based collaboration tools, but they do not provide great support for structured data. Relational databases provide very strong support for storing, managing and querying structured data but are very difficult to use. Both kinds of systems are currently used in bioinformatics projects - wikis often used in early stages, followed by a painful transition to database-backed web sites once the data needs exceed what a wiki can provide.

25 Nov 2011

Efficient Distributed Cross Match Queries in the Virtual Observatory

The problem of cross matching astronomical catalogs for information of interest is a non-trivial problem requiring complex mathematical computations for comparison of each pair of objects in the sky. A discussion would be presented on the different strategies used for performing efficient cross matching of astronomical objects, thereby stating their advantages and disadvantages as well as presenting respective experimental results.

Date: Friday 18th November 2011, 10:30, Turing Room.

18 Nov 2011

OMERO.rapid: Adding High Performance Computing to the Open Microscopy Environment.

The Open Microscopy Environment (OME) is a multi-site collaborative effort among academic laboratories and a number of commercial entities that produces open tools to support data management for biological light microscopy. OMERO is client-server software for visualisation, management and analysis of biological microscope images. In this presentation we describe OMERO and how we have integrated RAPID into the OMERO server providing high performance computing(HPC) to the OMERO Server.

Date: Friday 11th November 2011, 10:30, Turing Room

11 Nov 2011

Dynamic and Adaptive optimization techniques to enhance the performance of MPI applications by using HECToR and Eddie clusters.

NeSC Research Seminar Series
Rosa Filgueira

Dynamic and Adaptive optimization techniques to enhance the performance of MPI applications by using HECToR and Eddie clusters.

Date and time: 
Friday, 4 November, 2011 - 10:30
60 minutes

DISPEL optimisation

7th October: Chee Sun Liew

The Performance and Cost Variability of Amazon EC2

This presentation will focus on some of Amazon EC2's unique performance and cost characteristics. We test the impact of regional cost differences by submitting a job to EC2 from both Thailand and the UK. Though startup cost differences are small, cost differences scale with use and hence impact overall business competitiveness. Predicting this is complicated by variations in the underlying cloud load that mean application performance can differ upon each run. Costs can also vary as data transfer usage may or may not be recorded, and hence charged.

23 Sep 2011

Efficient Analytical Data Processing with In-Memory SQL/MR

16th September: Makoto

I will talk about analytical data processing on shared-nothing machines,
particularly on a memory-rich cluster. When I was a visiting post-doc of
CWI, I designed and implemented MonetDB/MR, a parallel shared-nothing
database using MonetDB, with Prof. Martin Kersten and Peter Boncz. A
key idea behind MonetDB/MR is exploiting memory-resident MapReduce

Use of parallelism on MSA tools

9th September 2011: Miquel Orobitg, PhD Student, HPC_Europe visitor

Abstract: Multiple Sequence Alignment (MSA) constitutes an extremely powerful tool for important
biological applications such as phylogenetic analysis, identification of conserved motifs and
domains and structure prediction. In spite of the improvement in speed and accuracy introduced by