Course schedule

6 April
9:00 – 9:15Welcome and introduction
9:15 – 10:00Introduction to HPC concepts (theoretical part)

  • Parallel architectures
  • Parallel programming languages and paradigms
  • Program analysis, program transformations, data locality optimizations
  • Parallelization techniques, run-time systems, software environments
  • Master-slave model
  • Map – Reduce techniques
  • Load distribution and balancing
  • Scheduling by priorities
10:00 – 10:30Connecting and copying files from/to the execution cluster
10:30 – 11:00Introduction to SLURM and job submission
11:00 – 11:30Coffee break
11:30 – 12:30Introduction to OpenMP and Practical session on introductory shared memory exercises:

  • HelloWorld in OpenMP
  • Loop-work sharing
  • Sections-work sharing
  • Reduction
  • Varying the number of threads
12:30 – 13:30Introduction to MPI and Practical session on introductory distributed memory exercises:

  • HelloWorld in MPI
  • Master-Slave approach
  • Network bandwidth exercise
  • Map-reduce static data distribution
13:30 – 14:30Lunch
14:30 – 15:00Introduction to sequence database search and dynamic data distribution
15:00 – 16:00Practical session on sequence database search with MPI and with static and dynamic data distribution (part 1)
16:00 – 16:30Coffee break
16:30 – 18:00Practical session on sequence database search with MPI and with static and dynamic data distribution (part 2)

 

7 April
08:00 – 8:15Summary of the first day
8:15 – 8:45Second day presentation.

Background on pairwise and multiple genome comparison.

Introduction to GECKO.

08:45 – 9:45 Practical session:

  • retrieving the code, compiling it
  • using GECKO to compare 3 or 4 genomes.
  • taking a look at the structure looking for possible parallelization strategies.
09:45 – 10:30Description and pseudo-code of the implemented parallel strategy. Coarse-grained at the pairwise comparison level and fine-grained inside each module composing GECKO. Measuring application speedup and efficiency
10:30 – 11:00Coffee break
11:00 – 13:00Groups organization and start working with the parallel strategy.

Practical session: Self-organized group work to benchmark the described parallel strategy

13:00 – 14:00Lunch
14:00 – 14:30Motivational discussion about parallelisation and its costs and benefits in GECKO:

  • Each group will present their benchmarking results (i.e. the obtained speedup). Each group will have 5 minutes approximately
  • Analysis of the difficulties encountered by each of the groups. Analysis of the differences in the obtained speedups.
14:30 – 14:45Quick recapitulation of the course
GitHub

All material will be available to download under Attribution-NonCommercial-NoDerivatives 4.0 International at the beginning of the course from the official course github web/repository

ELIXIR-SI eLearning Platform

All material is also available in the ELIXIR-SI eLearning platform web

Course Streaming

The course will be streamed (live) in the following web (Google chrome required)

Overview

This course to be held 6-7 April 2017 in Malaga, Spain is part of the Train-the-Researcher (TtR) subtask of the ELIXIR-EXCELERATE project. The course is organised as a two-day mini-symposium, and aims at introducing the participants in the complexities of parallel programming with emphasis on genome-scale comparison algorithms.

  • The theoretical aspects of this course cover (a) the parallel programing background, with a quick overview in the architectures and programming models; and (b) the basis for genome-scale sequence comparison algorithms.
  • The practical aspects are organised to master the concepts of data distribution and balancing using a Map-Reduce strategy and internal coding with MPI.

Organization overview

This course has been prepared for two days including theoretical and practical parts. The first day will start with a brief introduction to HPC condensed during the morning. A practical introductory part to HPC will complete the first day. The second day will firstly provide the necessary background on the multiple genome comparison application GECKO. Secondly, on the same day, we will motivate the potential HPC techniques to be applied to this application and will let the students organize in groups to start thinking about a parallelization approach and later on implementing it. At the end of the day each group will briefly describe how they have implemented it and will present their results (i.e. speedup).

Methods

The course is comprised of practical exercises preceded by short lectures. Exercises will be conducted primarily in the C programming language using the MPI library.

Target Audience

This course is open to bioinformatics users and developers that aim at using HPC resources to extend their activity beyond the limitations of sequential applications on commodity hardware. This course will not suit participants that are looking for entry-level concepts in HPC.

20 participants will be selected by CV and motivation. Gender equality will be an additional aspect to select the participants. As an applicant to this course you need to fill out your personal information, attach your CV and provide a short paragraph to explain why you want to apply for this course. Optionally if you are part of an ELIXIR node you can also attach a letter of endorsement from the node manager. The last application day will be March 16th, the result of the candidate selection will be notified March 17th. The course registration is free, but by sending this participation you commit to attend the course in case you are selected as it is stated in the conditions/policy document of the registration form.

Pre-requisites

This course has a broad technical scope and hence programming skills are required. Experience in the command-line, writing and compiling C code would make the course easier to follow.

City center to course venue

From the city center (Alameda principal) to the course venue (room 2.0.10, indications will be provided in-place) . Bus line 11 operates every 10 minutes

Course venue to city center

From the the course venue to the city center. Bus line 11 operates every 10 minutes

City center to course venue

From Hotel Tribuna (C/ Carretería) to the bus stop (Alameda principal. Bus line 11)

Points of interest in Google maps (available here) https://drive.google.com/open?id=19Xv5N5qgNdgcSreAhmC08-oAE8w&usp=sharing

Uso de cookies

Este sitio web utiliza cookies para que usted tenga la mejor experiencia de usuario. Si continúa navegando está dando su consentimiento para la aceptación de las mencionadas cookies y la aceptación de nuestra política de cookies, pinche el enlace para mayor información.plugin cookies

ACEPTAR
Aviso de cookies