Projektkurs i högprestandaberäkningar

This is a project applied course in the research area of High-Performance Computing. 

As part of the project, you will identify and formulate a two months project with the guidance of the course instructors. 

Potential project topics (but not limited to):

• Development of a specific algorithm for using GPUs with CUDA, OpenCL or HIP and performance evaluation.
• Developing an algorithm using MPI and/or OpenMP for using supercomputers and performance evaluation with HPC tools.
• Benchmarking of emerging architecture systems (Nvidia tensor cores, TPU, M1 matrix engines, ...) and performance analysis.
• A performance evaluation of large-scale parallel codes using HPC tools. Relevant codes are GROMACS, Neko, iPIC3D, Vlasiator, PIConGPU, AutoDOCK.
• Evaluation of precision impact of novel floating-point formats (Posit, bfloat, ...) on scientific calculations
• An evaluation of quantum arithmetics and Fourier transform using quantum computing framework.
• An evaluation of molecular / DNA computing for computation and storing data.

Potential algorithms you would work on are matrix multiply, convolution, FFTs, n-body simulators, ray-casting, Montecarlo solvers, ...

The project will require:

1. Submission of the project specifications and plan
2. Submission of providing the literature regarding the research background and previous related work.
3. Submission of a final report and presentation

 Course Objectives: 

• Provide an overview of relevant research publication, the structure of a research paper and how to effectively search for relevant literature.
• Read, present and discuss the latest research reports in high performance computing
• Implement a presented method, or parts thereof, from a research paper

Intended Learning Outcomes

After passing the course, the student must be able to:

• Implement and present all or parts of a method from a current research report in high performance calculations
• Report on current research challenges in high-performance computing
• Effectively search for research literature relevant to high performance computing
• Orient yourself in the scientific publication landscape, including conferences, journals and bibliometrics
• Effectively read research articles and evaluate them with respect to structure and aspects such as originality and technical values, evaluation and usefulness with the aim of being able to benefit from academic research in their future career and thus contribute to knowledge transfer to industry.