The Course is over, thanks a lot to everyone who participated

Scientific Computing in Practice: Summer Kickstart 2016

Part of Scientific Computing in Practice lecture series at Aalto University.

Audience: Aalto employees and students looking for extensive intro into Triton cluster scientific computing facilities at Aalto.

About the course: kickstart for Aalto researchers to get started with the available HPC computational resources at Aalto. The course is focused on Trtion usage practicalities and HPC in general. Course is given in the hands-on style tutorial, there will be lot of practical things that you end up trying on your own on Triton. We give you hints, ready solutions and copy/paste ready examples on how to compile, run and monitor your app on Triton, in addition to how to optimize your workflow in terms of filesystem traffic, memory usage etc.

Time, date: Mon 6.6 — Wed 8.6.2016, from 10:00 – 15:00 all days

Place: Otaniemi, Otakaari 1, U7 (first floor, next to student HUB)

Days’ time schedule is flexible, here is the tentative plan. 30 minutes break around 12:00, plus short 10 minutes breaks in between. You will be given time to try and ask, it’s more like an informal help session to get you started on Triton.

Day #1:

  • Triton basics: access, SLURM, monitoring, environment modules
  • HPC crash course: what is behind the front-end

Will cover HPC fundamentals: terminology, architectures, interconnects, infrastructure behind. Traditionally will stay focus on Triton: how to get started, where one finds help and list of available apps, how to run and fetch results from your job. SLURM advances like array jobs, using GPUs, Matlab runs etc. Whole introduction is build around concrete use cases that end user will have a chance to try on its own

Triton basics (PDF; updated on 2016-06-07)

Day #2:

  • Troubleshooting: solving running jobs issues.
  • Code building
  • Lustre filesystem: how to optimize your IO workflow

We will continue with Triton practicalities, what are the possible issues and how to resolve them, how to compile a serial code and how to build an application. We will cover IO issues, telling what are pros and cons of parallel filesystem used on Triton.

Troubleshooting on Triton, Building the code, IO on Triton: Lustre and others (PDF)

Day #3:

  • Parallel basics: MPI and OpenMP
  • Advanced jobs running: running in parallel
  • Debugging serial code

Paralellization, profiling and debugging are keywords of this day sessions.

Debugging on TritonMPI and OpenMP on Triton

Lectures: Janne Blomqvist (D. Sc.), Mikko Hakala (D. Sc.), Ivan Degtyarenko (D. Sc.), Simo Tuomisto

Cost: Free of charge for Aalto employees and students.

Credits (opintopisteet): Full course hours correspond roughly to 1 ECTS. Requires all three days attendance plus homework. Homework is a set tasks to be done either during the course or shortly after and to be returned to course organizers. See details at SCIP2016 summer homework (PDF).

Other comments: All parts will include live examples, demos and hands-on tutorials. Participants will be provided with access to Triton for running examples. Participants are expected to bring their laptops with the pre-installed SSH client, configured for using WiFi (preferably Eduroam, otherwise Aalto or Aalto Open).

Additional course info at: Ivan Degtyarenko, ivan.degtyarenko at aalto.fi