Posts by Collection


How to Contact


Technische Universität Hamburg
Institute of Massively Parallel Systems (E-EXK5)
Am Schwarzenberg-Campus 3 (E), 4.046
21073 Hamburg, Germany





GPU Architectures

MS, Summer Semester, 2022

Module Gerneral Information

The module consists of lectures and a semester-long problem-based project.

  • Level: MS
  • Credit points: 6
  • Semester hours per week (SWS): 4
  • Instructor: Sohan Lal
  • Time and Localtion: Thursday Lecture 13:15 - 15:00, Project Meeting: 15:00 - 17:00, TBA

Advanced Topics in Computer Architecture

BS/MS, Winter semester 2022, 2023

Seminar Gerneral Information

  • Level: BS/MS
  • Credit points: 3
  • Semester hours per week (SWS): 2
  • Instructor: Sohan Lal
  • Time and Localtion: TBA

Massively Parallel Systems: Architecture and Programming

MS, Winter semester 2022, 2023

Module Gerneral Information

The module consists of lectures and several programming assignments.

  • Level: MS
  • Credit points: 6
  • Semester hours per week (SWS): 4
  • Instructor: Sohan Lal
  • Lecture: Thursday, 11:00-13:00, room O-0.018
  • Lab: Thursday, 14:00-16:00, Linux pool, room CIP/E-2.012P3b


Short Bio

Manuel Renz is a Ph.D. student under the supervision of Prof. Sohan Lal since July, 2022. After receiving his M.Sc. in Computer Science from TU Berlin, he worked at CARIAD SE (former Carmeq GmbH) in the domain of image-processing. During his masters, his research focused on heterogenous multi-/manycore systems and he worked in cooperation with DFKI (Deutsches Forschungszentrum für Künstliche Intelligenz) and Technical University of Berlin. Due to his keen interest in computer architecture, he decided to continue research in this area. In his Ph.D., he is working to increase the performance and energy-efficiency of multi-/many-core systems such as GPUs.

Short Bio

Sohan Lal is a junior professor and head of the Institue of Massively Parallel Systems at Hamburg University of Technology (TUHH). Before joining TUHH, he was a postdoctoral researcher at the Technical University of Berlin (TU Berlin), working on a DFG funded research project on advanced modeling and runtime support for large-scale HPC clusters (Celerity). He graduated with a Ph.D. in Computer Science from TU Berlin in August 2019. His dissertation was titled “Power Modeling and Architectural Techniques for Energy-Efficient GPUs” and was supervised by Prof. Ben Juurlink. At TU Berlin, he worked on two EU funded research projects on low power parallel computing on GPUs (LPGPU), where he led several tasks, collaborated with consortium members to deliver joint deliverables and contributed significantly to their success. His Ph.D. dissertation work was also conducted in the context of LPGPU projects. For his dissertation, he investigated bottlenecks that cause low performance and low energy efficiency in GPUs and proposed architectural techniques to improve performance and energy efficiency. The results of the dissertation were published in several reputed conferences such as IPDPS, DATE, ISPASS. He won several grants such as HiPEAC travel grants, a HiPEAC collaboration grant to visit Prof. Henk Coporaal (TU/e) that led to a joint publication at DATE. He was a semifinalist at ACM SRC held at MICRO'18. He is interested in computer architecture in general and graphics processing unit (GPU) architecture in particular and his broad research interests include power and performance modeling, parallel systems, memory systems, heterogeneous computing, approximate computing, applied machine learning, and GPU security.

Short Bio

Tim Lühnen is a Ph.D. student at the Institute of Massively Parallel Systems under the supervision of Prof. Sohan Lal since November 2022. He studied Electrical Engineering at Technische Universität Braunschweig between 2016 and 2021. The title of his master thesis was "An Evaluation of Softcore Processors for Fault-tolerant Computing Platforms".



Below is a list of open, ongoing, and completed theses. If you are intested in an open thesis, or you alreay have a topic and looking for a supervisor, please contact me.