Mixed precision, approximate computing, and data compression for computational fluid dynamics

Figure 1: (a) NVIDIA A100 (source of imge of NVIDIA A100). (b) Turbulent flow simulations using GPUs and LBM-based CFD

Computer hardware is changing noticeably at the fundamental, computer arithmetic level in the last decade. Graphical processing units (GPUs) and some CPUs increasingly started introducing 16-and 8-bit hardware for computing mathematical operations, which work in parallel with classical 32-and 64-bit units. These low precision units are very efficient and to utilize them we need the software to mix precisions. However, they are not always compliant with the IEEE 754 standard which has been around for almost 40 years, and that brings certain challenges at the fundamental level. Apart from that, researchers are trying to gain further performance gains through the techniques of approximate computing such as loop perforation, under designed arithmetic operators, or reduced voltage memory. You will be exploring all of this in the context of computational fluid dynamics (CFD). This project is a perfect opportunity to learn about fundamental (and currently evolving) aspects of numerical computer hardware and then combine that with your knowledge of CFD algorithms to explore pushing the performance further. Researchers with experience on both the hardware level and the application level are rare and this is a unique opportunity to gain that.