Advanced Computing in the Age of AI | Monday, June 17, 2024

The MAESTRO Project Launches Maestro Core 0.2 Software 

Oct. 29, 2021 -- The MAESTRO project is pleased to announce the availability of the first public release of Maestro Core 0.2, a C library that does multi-threaded cross-application data transfer over high-speed interconnect using lightweight data wrappers that include user metadata.

Maestro Core is part of the Maestro middleware framework, a data- and memory-aware abstraction for workflow coupling, inter- and intra-application data exchange and redistribution. A central design goal was to enable modelling of memory and storage hierarchies to allow for reasoning about data movement and placement based on costs of moving data objects. At the same time, data objects can carry user-defined metadata. This is, e.g., the case for various meteorological production workflows. Leveraging the Maestro middleware allows workflow management software to reason about the data without inspecting it, at scale.

The software is still in an early stage but has been tested in various environments with success, both functionally and performance-wise.

We make the updated Maestro Core available at: https://gitlab.jsc.fz-juelich.de/maestro/maestro-core.

Follow the documentation to get started.

About MAESTRO

Maestro will build a data-aware and memory-aware middleware framework that addresses ubiquitous problems of data movement in complex memory hierarchies and at many levels of the HPC software stack. The Maestro consortium consists of 7 expert partners (Jülich Supercomputing Centre, CEA, Appentra, ETH Zurich (CSCS), ECMWF, Seagate, HPE), each bringing specialist knowledge and expertise to the technical challenge. More information at maestro-data.eu.


Source: MAESTRO

EnterpriseAI