SC²S Colloquium - July 4, 2018
|Date:||July 4, 2018|
|Time:||15:00 - 15:45|
Andreas Reiser: Extending a CFD Lab Course by a preCICE Conjugate Heat Transfer Tutorial
This is a Master's Thesis introduction advised by Benjamin Uekermann
Conjugate heat transfer refers to the coupled analysis of the thermal interactions between fluids and solids. The topic of my work is to create an educational concept to teach students about multi-physics coupling with preCICE using this phenomenon as demonstrative example. For this purpose the students are asked to implement a preCICE adapter in order to couple their fluid-solver to an (black-box) openFOAM solid-solver. In this talk I will present the considerations which have to be made and the example scenarios which the students will use to test their implementation.
Keywords: preCICE, CFD, FSI
Michael Lettrich: Design of a Throughput Oriented Network Transport Layer Based on MPI for the Data Acquisition System of the CMS Detector at CERN
This is a Master's Thesis submission talk advised by Steffen Seckler
The Large Hadron Collider (LHC) at CERN generates unstable sub-atomic particles by colliding two opposed particle beams with up to 14TeV in predefined interaction points. As one of four LHC experiments, the Compact Muon Solenoid (CMS) experiment registers them in its set of different sub-detectors. With collisions producing 1MB of data at a rate of 40MHz, the data rate must be reduced by the data acquisition system (DAQ) before sent to mass storage. After reduction by hardware filtering to 100kHz, a second, high level software filter reduces the rate even more. Between these steps, an event-building cluster merges data from the same collision at 100kHz using an RDMA Interconnect programmed at driver level. This thesis investigates how the Message Passing Interface (MPI), mostly used in high performance computing, can be leveraged to implement an efficient, scalable and portable network transport that can handle the high bandwidths required for the CMS event-builder.
Keywords: MPI, LHC, CERN