Input/Output and visualisation in the HPC era. In the three CompBioMed exemplars, input, output and the way data is analysed are key issues. Molecular dynamics, 3D flow features, muscular fibre contraction, particle transport or bone structure, are clear examples of the difficulties and the power of visualisation once the problematic issues are solved. Elegant yet efficient solutions are required, always keeping in mind the large-scale sizes of the resulting datasets. In this section we summarise and discuss the different strategies we employ to tackle the visualisation problems of CompBioMed exemplars. As discussed in [1], scientific visualisation focuses on the creation of images to provide important information about underlying data and processes. In recent decades, the unprecedented growth in computing and sensor performance has led to the ability to capture the physical world in unprecedented levels of detail and to model and simulate complex physical phenomena. Visualisation plays a decisive role in the extraction of knowledge from these data—as the mathematician Xxxxxxx Xxxxxxx famously said, “The purpose of computing is insight, not numbers...” [22]. Visualisation supports improved understanding of large and complex data in two, three, or more dimensions from different applications. In the kind of problems we are dealing with, visualisation is of great importance, as simulation results are often best represented in the time-dependent three-dimensional form we are familiar with. Traditionally, I/O and visualisation are closely related as, in most workflows, data used for visualisation are written to disk and then read by a separate visualisation tool. This is also called “post-mortem” visualisation, since the visualisation may be done after the simulations have finished. Other modes of interaction with visualisation are becoming more common, such as in situ visualisation (in which the simulation code directly produces visualisation images, using the same nodes and partitioning), or “in- transit” visualisation (in which the simulation code is coupled to a visualisation program, possibly running on other nodes and with a different partitioning scheme). Input/Output. Complex multi-physics, organ level simulations (cardiovascular, respiratory, etc.) writing files for post-mortem visualisation usually involve the highest volume of output. There are however other operations, especially explicit check- pointing at restart, that require the writing and reading of large datasets. Logging or output of dat...
