Project Ermes is an internal passion project at CodeZero and the time commitment provided by our team is on a voluntary basis as its a research project focussed on CFD (Computational Fluid Dynamics).
CFD is a form of science that, with the help of HPC, produces quantitative predictions of fluid-flow phenomena based on the conservation laws (conservation of mass, momentum, and energy) governing fluid motion. CFD has increased in importance and in accuracy; however, its predictions are never completely exact. This is due to many potential sources of error that may be involved, one has to be very careful when interpreting the results produced by CFD techniques.
The key to various numerical methods is to convert the partial different equations that govern a physical phenomenon into a system of algebraic equations. Different techniques are available for this conversion. CFD is merely a tool for analysing fluid-flow problems. If it is used correctly, it can provide useful information.
At CodeZero we believe HPC is HPC whether it is 1x node or 64x nodes, the difference being how much data do you actually need and how much computing power is required to run the complex calculations to derive insights. If a typical CFD dataset is 20TB, how much of the total dataset is meaningful in terms of the problem you are trying to solve. If we apply this project to something like Immersion Cooling, we could start to run an analysis on how the immersion oil behaves under certain conditions. A fully populated Immersion tank and rack will most likely provide a situation where the oil behaves differently, if the tank is only part populated. How the oil behaves may also have a direct impact on how oil performs, which in turn affects server performance. Another test could be to track how the oil performs under different types of load which could be simulated by running different types of AI, ML & HPC workloads on the server CPUs, GPUs, Memory & Storage.