We are currently at a point where the gap between the scientific programming and the computing platforms, on which the scientific codes must be computed, is growing bigger and bigger. The importance of scientific programming for High Performance Computing (HPC) is increasing and has become as one of the foremost fields of research. This has raised many issues, such as computing architectures, memory, networking, and programming models. In turn, this forces us to adapt our codes, or implement new ones, to take advantage of the latest computational features.

This special issue focuses on these challenges, which arise when programming scientific codes over massively parallel HPC platforms composed of a high number of cores dealing with communication, programming, specialized architectures, load balancing, benchmarking, etc. We hope to provide a platform to discuss how modern scientific programming, which has important challenges and high computational requirements, can be efficiently mapped on current and upcoming high-performance platforms.

The aim of this special issue is to bridge the gap between the theory of scientific problems (machine learning, scientific simulations, image processing, computational fluid dynamics, bioinformatics, linear algebra, big data computing, etc.) and their implementation on HPC platforms by exploring, proposing, and evaluating new trends/directions in scientific programming. Authors are invited to submit original research and review articles. Works focused on emerging programming models and computer architectures are especially welcome.

Potential topics include but are not limited to the following:

• Scientific programming on specialized hardware (GPUs, Tensor Core, FPGAs, Heterogeneous Memory, etc.)
• Numerical modelling of scientific applications to increase performance (novel methodologies to achieve more scalability, less synchronization-communication, less memory occupancy, mixed-precision, etc.)
• Programming techniques for communication, synchronization, and load balancing
• Evaluation of scientific applications including scalability, benchmarking, performance, and numerical accuracy analysis
• Use of programming models for scientific applications (OpenMP, MPI, CUDA, etc.)
• Autotuning in scientific programming