TUD launches EU Horizon project to optimize cloud performance

Together with leading European partners, TUD Dresden University of Technology (TUD) is launching the research project "CHORYS: Open and Programmable Accelerators for Data-Intensive Applications in the Cloud". In collaboration with the University of Copenhagen, Politecnico di Milano and Technische Universität Darmstadt as well as the companies Menta SAS and Codasip GmbH, the project aims to significantly increase the performance of data-intensive applications in the cloud.

The main objective of CHORYS is to develop a new hardware and software architecture that is specifically tailored to the needs of data-intensive applications in the cloud. The focus is on the integration of near-data processing technologies and asynchronous data services in order to make data processing more efficient and resource-saving.

An innovative aspect of CHORYS is the close integration of RISC-V-based processor extensions and specialized accelerators such as OpenSSD-V and OpenNIC-V. These technologies make it possible to carry out database and analysis processes directly where the data is generated - and thus reduce time-consuming data movements through conventional storage and network structures.

Particularly noteworthy is the combination of programmable SSDs, SmartNICs (network interface cards) and data-parallel processing technologies that are integrated within an open architecture. This is complemented by the development of a powerful library that enables efficient execution of data-intensive computations on different hardware platforms.

The CHORYS project results have the potential to make a significant contribution to the further development of the cloud computing market and to significantly advance data-intensive business applications, scientific data processing and open source projects.

As part of the project, the Chair of Databases at the TUD is leading the "Data-Intensive Patterns Library and APIs" work package. The aim is to develop a standardized programming interface that makes it possible to efficiently distribute parts of data-intensive algorithms to different computing units and thus enable optimized use of hardware resources.

A central component of the research work is the further development of the SIMD hardware abstraction library TSL, which already enables cross-platform execution of data-parallel codes on different architectures such as x86, AArch64 and oneAPI-FPGAs.

The project is being funded by the European Union to the tune of over 4 million euros over a period of four years.