HPC Yearbook 19/20 As a result, the simulation result data can

be visualised in situ, while the simulation is still running, by accessing shared memory. ‘Tis saves both time and memory, and greatly supports the real-time nature of the Vestec project,’ according to Johannes Guenther, a senior graphics soſtware engineer at Intel. Tis is an important point because

visualisation with ray tracing is a computationally-intensive process, providing what is essentially a physics simulation of light propagation. ‘We take care that our rendering libraries do effectively utilise the resources of HPC systems,’ Pawlowski added. But the ray-tracing algorithm is

‘embarrassingly parallel’ where the pixels of an image can largely be computed independently from each other, according to Guenther, who added: ‘Tus, we apply parallelisation at multiple levels, by using MPI (message passing interface) to scale across many compute nodes, as well as threading and SIMD (single instruction, multiple data) vectorisation local to a node.’

Wildfire watch

Te VESTEC project is now being used to investigate three different but highly impactful disaster scenarios: wildfires, mosquito-borne diseases and space weather. Wildfires are unpredictable, as Miguel

Mendes, R&D department project manager at specialist soſtware house Tecnosylva, explained: ‘Reducing the fire simulation uncertainties will greatly improve the accuracy and processing speed of results, which shall lead to faster and more accurate operational decisions by the practitioners when dealing with forest fires.’ So, the next iteration of the Wildfire

Analyst fire simulation code, which is currently under development at Tecnosylva, will better predict the behaviour of fire using stochastic simulations. Mendes said: ‘[Our] novel technique for

near-real-time data assimilation of satellite data to calibrate the simulation results will

High-Performance Computing 2019-20

With the massively increased power of modern supercomputers, it should be possible to run these simulations while a disaster is on-going, to help inform response teams and aid decision-making

take the forest fire modelling to the next level. As soon as new satellite data of the ongoing fire becomes available, the simulator will integrate this to calibrate simulation results.’ Te Wildfire Analyst analytics soſtware is

currently used operationally by firefighters to help them make important decisions when dealing with forest fires. Mendes added: ‘Tis project shall take the Wildfire Analyst simulation capabilities to the next level through the operational use of stochastic forest fire forecast and data assimilation.’

Space weather

Space weather forecasting is still a relatively new phenomenon. But, as the space industry gathers pace and brings in more data while our computational powers increase, space weather modelling is an emerging field to help us predict extraterrestrial threats, such as solar storms. For the Vestec project, the aim is to

develop efficient, in situ visualisation and data analysis of space weather in a highly parallel production simulator. Te KTH Royal Institute of Technology

Fire presence probability for a given evacuation line drawn by the user (“Analysis line”)


has developed a soſtware solution for this part of the project, called MHD-Epic. Vyacheslav Olshevsky, a researcher at KTH, said: ‘It is unique because it simulates the whole heliosphere, while resolving most important kinetic processes happening in the

Earth’s magnetosphere.’ Te two codes are coupled using a

hierarchical approach, with information passing both from large global scales to the smallest kinetic scales, and vice versa. Olshevsky added: ‘Te information exchange is very efficient and takes less than two per cent of the execution time in all simulations.’

Mosquito maps

Another team is developing code to create risk maps for mosquito-borne diseases that evolve as a consequence of variations in climatic variables. Giorgio Guzzetta, a researcher from

the Fondazione Bruno Kessler, said: ‘Being able to predict the areas at highest risk will enhance our preparedness in response to epidemics: such as by allocating resources where they are most needed, by increasing the awareness of local health providers, and prioritising the logistics of control interventions (such as insecticide spraying).’ Currently, there are global maps

estimating the existence of suitable habitats for the world’s mosquito species. But they don’t estimate their abundance, a key factor when calculating disease transmission risk. Te new code provides an ‘educated

estimate of the mosquito abundance in any location of the world, and includes seasonal variations,’ according to Piero Poletti, a fellow

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32