Course PM

This Canvas page describes the lectures and computer labs of the course. For information on learning outcomes, literature, teachers, and examination (in the Swedish variant of English known as "Course PM") please confer a separate Canvas page.

Website

This course builds heavily on Chalmers-independent, external material provided on a separate HPC website created and curated by Martin Raum, who is also the lecturer and examiner of this course.

Program

The schedule of the course is available on TimeEdit.

The events formerly known at the lectures

Lectures scheduled in TimeEdit will complement the videos on the HPC website. Currently, the subdivision into weeks on this site is aligned with TMA882/MMA620.

Lectures are a possibility to review any material from those videos that required further clarification. Please do not hesitate to let the lecturer know by mail in advance of these sessions when you wish specific topics to be highlighted. If possible bring your laptop to actively follow the discussion.

Recommended exercises

After watching the videos  on the HPC website, replicate what is demonstrated there on the training system gantenbein. The HPC website also contains specific tasks and the assignments, the first of which currently are identical with the assignments required to pass the course TMA882/MMA620.

Computer labs

Students are hoped to join two lab sessions per week, while there is usually five of such sessions scheduled per week. See TimeEdit for exact time and dates.

During the lab sessions it is suggested to work through the videos on the HPC website or advance on the assignments; And to request help from the examiner and lab assistants when needed. Please do not hesitate to request such help. We not only provide guidance on these videos and assignments, but any programming and HPC related question is welcome. In case you have questions that go beyond the course material, we might prioritise students who are not yet there. This is not to discourage you—please persist—but rather a concession to an ILO-based teaching setup at Chalmers.

Reference literature

The main literature for the website

Our main resource are the videos provided on this website. The code that is available for download there, is also made accessible on the training system gantenbein in the folder /home/hpc2024/code.

A major secondary resource, however far beyond the scope of TMA882/MMA620, are the four volumes of

• The Art of HPC by Victor Eijkhout.

A good, and free, but slightly technical summary of the language C can be found in

• Essential C by Nick Parlante.

One essential part of your training is to learn how to solve specific problems in groups using independent resources, because this is how problems are commonly solved in real programmer’s life. Keep in mind that it is a common job interview question where you find solutions to unprecedented problems.

Policy on AI Usage

AI (that is, LLM) support is permitted, excluding the completion of assignments in full. The use of LLMs to obtain additional explanations beyond the tutors' advice of any material of the course is recommended; Not the least to compensate for the scarce resources available at university for individual teaching. Learners are equally encouraged to obtain guiding feedback on their code and on their terminal interaction from LLMs.

Policy Rational

Purpose: The key aspect of the course is to put relevant learning uncompromisingly in first place. AI support is highly likely to be part of future work, employment, and life. To empower and enable students in the use of these tools thus becomes imperative.

Fairness: Violations of stricter AI policies are not checkable without severe impact on the learning climate, resource allocation beyond the available, or relying on random checks that degrade enforcement to a coin flip.

Equality: Advice presented by people is no more sanctified than advice presented by machines (LLMs). The ban of AI would thus warrant for instance the ban of conversations with academic relatives --- an ahistoric and unrealistic proposition. Thus stricter AI policies would merely appear as a classist tool to maintain the edge of those ahead on the social ladder.

Limitations: However, creation of full solutions to real-world applications is currently not feasible with LLMs. Therefore, to generate full solutions to the assignments sabotages relevant learning opportunities. Even in the hypothetical case that next generation AI can solve realistically complex problems, the operators' tasks will shift correspondingly.