Follow the evolution of the algorithms

We imagine algorithms as computational formulas with standard and stable rules. They are, but in relation to the processors that process them. If the processors, architectures and operation of machines change, so do the languages and algorithms. There are several hundred computer languages and several million algorithms.

In the beginning, we calculated in a linear way, then parallel programming appeared and became more and more sophisticated. Now quantum computers are bringing a whole new way of looking at computation. The way of programming is also evolving: with the advent of machine learning, computers are able to improve and optimize their programming, which fundamentally changes the way operations are considered. Add to that economic and environmental considerations and we come to the need to do better.

But the diversity of tools and algorithms is such that it becomes difficult to keep up and take advantage of the specific benefits of each. The Concace project (Digital and Parallel Composability for High Performance Computing) aims to develop new approaches useful for numerical simulation to better exploit the possibilities and make them known.

Concace is also interested in emerging techniques in numerical simulation, such as the hybridization of computing and machine learning or quantum computing, with the objective of integrating new approaches or anticipating future developments in HPC.

"the tremendous computational power that quantum computing offers makes it necessary to rethink HPC algorithms."

For the full article: Concace: The Best of HPC Algorithms for Numerical Simulation

Visit inria.fr