Our research activities involve the design and implementation of the following software platforms.

  • Neomem: a software prototype designed to mitigate the issue of catastrophic forgetting that arises when training neural networks with continuously generated data. By employing the rehearsal approach, Neomem effectively retains previously observed data, preserving associated knowledge. Neomem’s advanced data management capabilities enable data-parallel training across dozens of GPUs, ensuring both scalability and excellent predictive performance of continual learning workloads.
  • E2Clab: a framework that implements a rigorous methodology for designing experiments with real-world workloads on the Edge-to-Cloud Computing Continuum. This methodology provides guidelines to move from real-world use cases to the design of relevant testbed setups for experiments enabling researchers to understand performance and to support the reproducibility of the experiments.
  • StorAlloc: StorAlloc is a simulator of a storage-aware job scheduler. Through our simulator, we want to demonstrate how storage resources can be allocated the same way as computing resources on large-scale systems equipped with heterogeneous storage spaces.
  • KerA: a software prototype which addresses the limitations of state-of-the-art storage solutions for stream processing. It illustrates for the first time the idea of unified storage for data ingestion and storage in support of low-latency stream processing.
  • Damaris: Damaris is a middleware for multicore SMP nodes allowing them to efficiently handle data transfers for storage and visualization by dedicating one or a few cores to the application I/O. The current version allows efficient asynchronous I/O, hiding all I/O related overheads such as data compression and post-processing. It was evaluated with the CM1 tornado simulation, one of the Blue Waters target applications. Future work is targeting fast direct access to the data from running simulations and efficient I/O scheduling.
  • BlobSeer: a distributed blob management service enabling multiversioning and efficient fine-grain access to huge blobs under heavy concurrency. Several of our current research activities are originated from BlobSeer.

Comments are closed.