Tópicos em paralelismo: High Performance Computing

Summary

The course will be conducted through online tools. We will adopt Google Classroom as our main tool. To get access to Google Classroom, students must use their DAC accounts. Invites to the course page with access codes will be sent to your emails.
Instructor Hervé Yviquel
Credits 2
Support The support will happen through virtual channels scheduled
with the professor
Course Goals By the end of the course, students must have a good knowledge
of the current challenges of high-performance computing and
understand the state of art technologies to solve them.
Classes Every Monday, 19:00 – 21:00, Google Meet

Importante Notice:

  • [05/08/2021] The first class introducing the course will happen synchronously Monday 9/08 at 19H on Google Meet and will be recorded.

Syllabus

Topics to be presented in the course include:

Introduction to High-Performance Computing ● Supercomputer Architecture ● Supercomputer Software Stack ● HPC Containers ● Distributed programming models, languages and frameworks ● Domain-Specific languages and libraries for HPC ● Distributed Task scheduling ● Collective operations ● Distributed programming in Python

The course material will be available on the course page and the Classroom.

Activities

The course will have the following activities:

  • Classes will happen on Mondays from 19pm to 21pm via Google Meet, with some classes being held synchronously and other classes taking place asynchronously. For all synchronous classes, the class videos will be made available later. * Q&A sessions will also happen during class hours to interact with students online and answer any questions. * Online quizzes, paper review activities and practical activities will be provided along the course to assist in fixing the concepts covered by the classes. Those activities will be made available on Google Classroom.

Tasks carried out during the course must be submitted through Google and Github Classroom in the area corresponding to the course.

Evaluation

The evaluation of the discipline will be conducted based on a set of varied tasks that will have grades distributed proportionally:

  • MF = (T1 + T2 + … + Tn) / n, where n is the number of activities carried out throughout the semester
  • The grade range will be:
    • A: >=8.5
    • B: >=7.0 and <8.5
    • C: >=5 e <7.0
    • D: <5

Bibliography

The class will not follow any particular book but the following book is a good reference on some of the aspects studied during the class:

  • Sterling, T., Brodowicz, M., Anderson, M. (2017). High Performance Computing: Modern Systems and Practices. Germany: Elsevier Science.

Observations

  • This discipline has no final exam.
  • Any fraud attempt on the projects will result in a final score of MF = 0 (zero) for all involved.