Publications – HiCrest

Bibliography

2026

John Gliksberg, Luca Andreetti, Daniele Di Bari, Giuseppe Pagano, Andrea Monterubbiano, Matteo Turisini, Lilia Zaourar, Mohamed Benazouz, Antony Pottier, Khaled Takhoubit, Julien Jaeger, Corentin Beaulieu, Anis Peysieux, Florian Reynier, Hugo Taboada, Angelos Bilas, Fabien Chaix, Nikolaos Chrysos, Evangelos Mageiropoulos, Gilles Moreau, Roberto Ammendola, Andrea Biagioni, Carlotta Chiarini, Ottorino Frezza, Francesca {Lo Cicero. NET4EXA: Toward Ethernet-native interconnects for exascale HPC and AI systems. Microprocessors and Microsystems. (2026)

@article{GLIKSBERG2026105296,
title = {NET4EXA: Toward Ethernet-native interconnects for exascale HPC and AI systems},
journal = {Microprocessors and Microsystems},
pages = {105296},
year = {2026},
issn = {0141-9331},
doi = {https://doi.org/10.1016/j.micpro.2026.105296},
url = {https://www.sciencedirect.com/science/article/pii/S0141933126000530},
author = {John Gliksberg and Luca Andreetti and Daniele Di Bari and Giuseppe Pagano and Andrea Monterubbiano and Matteo Turisini and Lilia Zaourar and Mohamed Benazouz and Antony Pottier and Khaled Takhoubit and Julien Jaeger and Corentin Beaulieu and Anis Peysieux and Florian Reynier and Hugo Taboada and Angelos Bilas and Fabien Chaix and Nikolaos Chrysos and Evangelos Mageiropoulos and Gilles Moreau and Roberto Ammendola and Andrea Biagioni and Carlotta Chiarini and Ottorino Frezza and Francesca {Lo Cicero} and Alessandro Lonardo and Michele Martinelli and Pier Stanislao Paolucci and Elena Pastorelli and Pierpaolo Perticaroli and Luca Pontisso and Cristian Rossi and Francesco Simula and Piero Vicini and Samuel Rodrigo Mocholi and Marzio Vallero and Kaijie Fan and Sandro Luigi Fiore and Fabrizio Granelli and Thomas Pasquali and Marco Patrignani and Simone Pezzuto and Lorenzo Pichetti and Raffaello Potestio and Jacopo Raffi and Domenico Siracusa and Luca Tubiana and Philippe Velha and Flavio Vella and Daniele {De Sensi} and Francesco Fazzari and Vladimiro Paschali and Saverio Pasqualoni and Lorenzo Piarulli and Salvatore Pontarelli and Taha Abdelazziz Rahmani},
keywords = {High-performance interconnects, Exascale computing, SmartNIC, RDMA, HPC and AI systems},
abstract = {High-Performance Computing (HPC) and Artificial Intelligence (AI) infrastructures are evolving toward highly heterogeneous exascale systems, where communication scalability and data movement increasingly limit overall performance and efficiency. This paper presents NET4EXA, a European initiative targeting the development of BXIv3, the third generation of the Bull eXascale Interconnect (BXI), based on an Ethernet-native architecture combining FPGA-based SmartNICs, custom switching ASICs, and hardware-accelerated communication semantics. The paper reports a selection of mid-project research results spanning multiple layers of the communication stack, including NIC- and switch-level collective offloading, TCP/RDMA acceleration, on-NIC processing, communication–computation overlap through triggered operations, and optimized data movement using IOVEC mechanisms. Additional contributions include OpenSHMEM integration over a Portals-based runtime, congestion-aware routing and mapping strategies for Dragonfly＋ topologies, and benchmarking frameworks for HPC and AI communication workloads. Finally, the paper discusses emerging research directions toward future interconnect architectures based on silicon photonics and co-packaged optics targeting multi-terabit communication systems.}
}

Unat, Didem, Turimbetov, Ilyas, Issa, Mohammed, Sagbili, Dogan, Vella, Flavio, Sensi, Daniele, Ismayilov, Ismayil. The Landscape of GPU-Centric Communication. ACM Comput. Surv.. (2026)

@article{10.1145/3813799,
author = {Unat, Didem and Turimbetov, Ilyas and Issa, Mohammed and Sagbili, Dogan and Vella, Flavio and Sensi, Daniele and Ismayilov, Ismayil},
title = {The Landscape of GPU-Centric Communication},
year = {2026},
issue_date = {September 2026},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {58},
number = {12},
issn = {0360-0300},
url = {https://doi.org/10.1145/3813799},
doi = {10.1145/3813799},
abstract = {In recent years, GPUs have become the preferred accelerators for HPC and ML applications due to their parallelism and high memory bandwidth. While GPUs boost computation, inter-GPU communication can create scalability bottlenecks, especially as the number of GPUs per node and cluster grows. Traditionally, the CPU managed multi-GPU communication, but advancements in GPU-centric communication now challenge this CPU dominance by reducing its involvement, granting GPUs more autonomy in communication tasks, and addressing mismatches in multi-GPU communication and computation.This article provides a landscape of GPU-centric communication, focusing on vendor mechanisms and user-level library supports. It aims to clarify the complexities and diverse options in this field, define the terminology, and categorize existing approaches within and across nodes. The article discusses vendor-provided mechanisms for communication and memory management in multi-GPU execution and reviews major communication libraries, their benefits, challenges, and performance insights. Then, it explores key research paradigms, future outlooks, and open research questions. By extensively describing GPU-centric communication techniques across the software and hardware stacks, we provide researchers, programmers, engineers, and library designers insights on how to exploit multi-GPU systems at their best.},
journal = {ACM Comput. Surv.},
month = jun,
articleno = {322},
numpages = {36},
keywords = {GPUs, communication, MPI, NVSHMEM, NCCL, RCCL, Peer-to-peer communication, collective communication, GPUDirect technologies}
}

Casagranda, Gioele, Auden, Elizabeth, Cazzaniga, Carlo, Kastriotou, Maria, Frost, Christopher, Vallero, Marzio, Vella, Flavio, Rech, Paolo. SQUID G.A.M.E.: Gamma, Atmospheric, and Mono-Energetic Neutron Effects on Quantum Devices. IEEE Transactions on Nuclear Science. (2026)

@ARTICLE{11361247,
  author={Casagranda, Gioele and Auden, Elizabeth and Cazzaniga, Carlo and Kastriotou, Maria and Frost, Christopher and Vallero, Marzio and Vella, Flavio and Rech, Paolo},
  journal={IEEE Transactions on Nuclear Science}, 
  title={SQUID G.A.M.E.: Gamma, Atmospheric, and Mono-Energetic Neutron Effects on Quantum Devices}, 
  year={2026},
  volume={73},
  number={4},
  pages={1089-1099},
  keywords={SQUIDs;Neutrons;Josephson junctions;Superconducting integrated circuits;Circuits;Reliability;Gamma-rays;Yttrium barium copper oxide;Substrates;Extraterrestrial measurements;Gamma rays;neutrons;quantum devices;radiation;reliability;simulation;superconducting quantum interference device (SQUID)},
  doi={10.1109/TNS.2026.3657307}}

Anas Jnini, Flavio Vella, Marius Zeinhofer. Gauss-Newton Natural Gradient Descent for Physics-informed Computational Fluid Dynamics. Computers & Fluids. (2026)

@article{JNINI2026106955,
title = {Gauss-Newton Natural Gradient Descent for Physics-informed Computational Fluid Dynamics},
journal = {Computers & Fluids},
volume = {307},
pages = {106955},
year = {2026},
issn = {0045-7930},
doi = {https://doi.org/10.1016/j.compfluid.2025.106955},
url = {https://www.sciencedirect.com/science/article/pii/S0045793025004153},
author = {Anas Jnini and Flavio Vella and Marius Zeinhofer},
keywords = {Gauss-Newton method, physics-informed neural networks, Navier-Stokes equations, natural gradient, computational fluid dynamics},
abstract = {We propose Gauss-Newton’s method in function space for the solution of the Navier-Stokes equations in the physics-informed neural network (PINN) framework. Upon discretization, this yields a natural gradient method that provably mimics the function space dynamics. Our computational results demonstrate close to single-precision accuracy measured in relative L2 norm on a number of benchmark problems. To the best of our knowledge, this constitutes the first contribution in the PINN literature that solves the Navier-Stokes equations to this degree of accuracy. Finally, we show that given a suitable integral discretization, the proposed optimization algorithm agrees with Gauss-Newton’s method in parameter space. This allows a matrix-free formulation enabling efficient scalability to large network sizes.}
}

Marzio Vallero, Paolo Rech, Flavio Vella. State of practice: Evaluating GPU performance of state vector and tensor network methods. Future Generation Computer Systems. (2026)

@article{Vallero2026Practice,
title = {State of practice: Evaluating GPU performance of state vector and tensor network methods},
journal = {Future Generation Computer Systems},
volume = {174},
pages = {107927},
year = {2026},
issn = {0167-739X},
doi = {https://doi.org/10.1016/j.future.2025.107927},
url = {https://www.sciencedirect.com/science/article/pii/S0167739X25002225},
author = {Marzio Vallero and Paolo Rech and Flavio Vella},
keywords = {Quantum, Tensor network, HPC},
abstract = {The frontier of quantum computing (QC) simulation on classical hardware is quickly reaching the hard scalability limits for computational feasibility. Nonetheless, there is still a need to simulate large quantum systems classically, as the Noisy Intermediate Scale Quantum (NISQ) devices are yet to be considered fault tolerant and performant enough in terms of operations per second. Each of the two main exact simulation techniques, state vector and tensor network simulators, boasts specific limitations. This article investigates the limits of current state-of-the-art simulation techniques on a test bench made of eight widely used quantum subroutines, each in different configurations, with a special emphasis on performance. We perform both single process and distributed scaleability experiments on a supercomputer. We correlate the performance measures from such experiments with the metrics that characterise the benchmark circuits, identifying the main reasons behind the observed performance trends. Specifically, we perform distributed sliced tensor contractions, and we analyse the impact of pathfinding quality on contraction time, correlating both results with topological circuit characteristics. From our observations, given the structure of a quantum circuit and the number of qubits, we highlight how to select the best simulation strategy, demonstrating how preventive circuit analysis can guide and improve simulation performance by more than an order of magnitude.}
}

2025

Martinelli, Michele, Ammendola, Roberto, Biagioni, Andrea, Chiarini, Carlotta, Frezza, Ottorino, Cicero, Francesca Lo, Lonardo, Alessandro, Paolucci, Pier Stanislao, Pastorelli, Elena, Perticaroli, Pierpaolo, Pontisso, Luca, Rossi, Cristian, Simula, Francesco, Vicini, Piero, Colin, David, Pichon, Grégoire, Louvet, Alexandre, Gliksberg, John, Chen, Claire, Turisini, Matteo, Monterubbiano, Andrea, Nominé, Jean-Philippe, Dutoit, Denis, Taboada, Hugo, Zaourar, Lilia, Benazouz, Mohamed, Bilas, Angelos, Chaix, Fabien, Katevenis, Manolis, Chrysos, Nikolaos, Mageiropoulos, Evangelos, Kozanitis, Christos, Moen, Thomas, Persvold, Steffen, Rustad, Einar, Fiore, Sandro, Granelli, Fabrizio, Pezzuto, Simone, Potestio, Raffaello, Tubiana, Luca, Velha, Philippe, Vella, Flavio, Sensi, Daniele De, Pontarelli, Salvatore. NET4EXA: Pioneering the Future of Interconnects for Supercomputing and AI. 2025 28th Euromicro Conference on Digital System Design (DSD). (2025)

@INPROCEEDINGS{11272719,
  author={Martinelli, Michele and Ammendola, Roberto and Biagioni, Andrea and Chiarini, Carlotta and Frezza, Ottorino and Cicero, Francesca Lo and Lonardo, Alessandro and Paolucci, Pier Stanislao and Pastorelli, Elena and Perticaroli, Pierpaolo and Pontisso, Luca and Rossi, Cristian and Simula, Francesco and Vicini, Piero and Colin, David and Pichon, Grégoire and Louvet, Alexandre and Gliksberg, John and Chen, Claire and Turisini, Matteo and Monterubbiano, Andrea and Nominé, Jean-Philippe and Dutoit, Denis and Taboada, Hugo and Zaourar, Lilia and Benazouz, Mohamed and Bilas, Angelos and Chaix, Fabien and Katevenis, Manolis and Chrysos, Nikolaos and Mageiropoulos, Evangelos and Kozanitis, Christos and Moen, Thomas and Persvold, Steffen and Rustad, Einar and Fiore, Sandro and Granelli, Fabrizio and Pezzuto, Simone and Potestio, Raffaello and Tubiana, Luca and Velha, Philippe and Vella, Flavio and Sensi, Daniele De and Pontarelli, Salvatore},
  booktitle={2025 28th Euromicro Conference on Digital System Design (DSD)}, 
  title={NET4EXA: Pioneering the Future of Interconnects for Supercomputing and AI}, 
  year={2025},
  volume={},
  number={},
  pages={153-159},
  keywords={Training;Large language models;Europe;Switches;Supercomputers;Software;Network interfaces;IP networks;Portfolios;Next generation networking},
  doi={10.1109/DSD67783.2025.00032}}

Giusto, Edoardo, Núñez-Corrales, Santiago, Smith, Kaitlin N. N., Cao, Phuong, Younis, Ed, Rech, Paolo, Vella, Flavio, Baheri, Betis, Cilardo, Alessandro, Montrucchio, Bartolomeo, Jiang, Weiwen, Xu, Shuai, Dasgupta, Samudra, Iyer, Ravishankar K., Humble, Travis S. S.. Dependable classical-quantum computing systems engineering. Frontiers in Computer Science. (2025)

@ARTICLE{10.3389/fcomp.2025.1520903,
    
AUTHOR={Giusto, Edoardo  and Núñez-Corrales, Santiago  and Smith, Kaitlin N. N. and Cao, Phuong  and Younis, Ed  and Rech, Paolo  and Vella, Flavio  and Baheri, Betis  and Cilardo, Alessandro  and Montrucchio, Bartolomeo  and Jiang, Weiwen  and Xu, Shuai  and Dasgupta, Samudra  and Iyer, Ravishankar K.  and Humble, Travis S. S.},
           
TITLE={Dependable classical-quantum computing systems engineering},
          
JOURNAL={Frontiers in Computer Science},
          
VOLUME={Volume 7 - 2025},
  
YEAR={2025},
  
URL={https://www.frontiersin.org/journals/computer-science/articles/10.3389/fcomp.2025.1520903},
  
DOI={10.3389/fcomp.2025.1520903},
  
ISSN={2624-9898},
  
ABSTRACT={Increasing evidence suggests quantum computing (QC) complements traditional High-Performance Computing (HPC) by leveraging its unique capabilities, leading to the emergence of a new, hybrid paradigm, QHPC. However, this integration introduces new challenges, with dependability–defined by reproducibility, resiliency, and security and privacy–emerging as a central concern for building trustworthy systems that provide an advantage to the users. This paper proposes a framework for dependable QHPC system design, organized around these three pillars. We identify integration challenges, anticipate roadblocks, and highlight productive synergies across QC, HPC, cloud platforms, and network security. Drawing from both classical computing principles and quantum-specific insights, we present a roadmap for co-design that supports robust hybrid architectures. Our approach offers concrete metrics for assessing dependability, provides design guidance for engineers working at the QC-HPC interface, and surfaces new engineering questions around complexity, scale, and fault tolerance. Ultimately, designing for dependability is key to realizing practical, scalable QHPC systems and accelerating the broader quantum ecosystem capable of translating quantum promises into actual application delivery.}}

Agosta, Giovanni, Cherubin, Stefano, Christ, Derek, Conti, Francesco, Djupdal, Asbjørn, Jung, Matthias, Keramidas, Georgios, Passerone, Roberto, Rech, Paolo, Ricci, Elisa, Velha, Philippe, Vella, Flavio, Yildirim, Kasim Sinan, Wilbert, Nils. Architecture, Simulation and Software Stack to Support Post-CMOS Accelerators: The ARCHYTAS Project. 2025 IEEE Computer Society Annual Symposium on VLSI (ISVLSI). (2025)

@INPROCEEDINGS{11130220,
  author={Agosta, Giovanni and Cherubin, Stefano and Christ, Derek and Conti, Francesco and Djupdal, Asbjørn and Jung, Matthias and Keramidas, Georgios and Passerone, Roberto and Rech, Paolo and Ricci, Elisa and Velha, Philippe and Vella, Flavio and Yildirim, Kasim Sinan and Wilbert, Nils},
  booktitle={2025 IEEE Computer Society Annual Symposium on VLSI (ISVLSI)}, 
  title={Architecture, Simulation and Software Stack to Support Post-CMOS Accelerators: The ARCHYTAS Project}, 
  year={2025},
  volume={1},
  number={},
  pages={1-6},
  keywords={Space vehicles;Nonvolatile memory;Neuromorphics;Surveillance;Scalability;Memory management;Full stack;AI accelerators;Systems architecture;Software;AI Accelerators;Processing In Memory;Heterogeneous Architectures;Opto-electronics},
  doi={10.1109/ISVLSI65124.2025.11130220}}

Jenihhin, M., Raik, J., Jutman, A., Cherezova, N., Ubar, R., Miclea, L., Enyedi, S., Ştefan, I., Stan, O., Corcheş, C., Peng, Z., Eles, P., Drechsler, R., Eggersglüß, S., Fey, G., Glowatz, A., Tille, D., Gielen, G., Coyette, A., Dobbelaere, W., Vanhooren, R., Chuang, P.-Y., Marinissen, E. J., Di Natale, G., Barragan, M., Maistri, P., Mir, S., Vatajelu, E.-I., Bernardi, P., Di Carlo, S., Prinetto, P., Reorda, M. Sonza, Violante, M., Stratigopoulos, H.-G., Michael, M. K., Neophytou, S., Hadjitheophanous, S., Christou, K., Skitsas, M., Bosio, A., Deveautour, B., Girard, P., Traiola, M., Virazel, A., Dos Santos, F. Fernandes, Kritikakou, A., Casagranda, G., Vallero, M., Vella, F., Rech, P., Bolzani Poehls, L. M., Krstic, M., Andjelkovic, M., Vargas, F., Tshagharyan, G., Harutyunyan, G., Vardanian, V., Shoukourian, S., Zorian, Y., Dworak, J., Nepal, K., Manikas, T., Taouil, M., Fieback, M., Gebregiorgis, A., Bishnoi, R., Hamdioui, S., Chatterjee, A., Saha, A., Komarraju, S., Ma, K., Amarnath, C., Tahoori, M., Mayahinia, M., Rajabalipanah, M., Basharkhah, K., Nosrati, N., Jahanpeima, Z., Navabi, Z., Wunderlich, H.-J., Hellebrand, S.. European Test Symposium Teams: an Anniversary Snapshot. 2025 IEEE European Test Symposium (ETS). (2025)

@INPROCEEDINGS{11049652,
  author={Jenihhin, M. and Raik, J. and Jutman, A. and Cherezova, N. and Ubar, R. and Miclea, L. and Enyedi, S. and Ştefan, I. and Stan, O. and Corcheş, C. and Peng, Z. and Eles, P. and Drechsler, R. and Eggersglüß, S. and Fey, G. and Glowatz, A. and Tille, D. and Gielen, G. and Coyette, A. and Dobbelaere, W. and Vanhooren, R. and Chuang, P.-Y. and Marinissen, E. J. and Di Natale, G. and Barragan, M. and Maistri, P. and Mir, S. and Vatajelu, E.-I. and Bernardi, P. and Di Carlo, S. and Prinetto, P. and Reorda, M. Sonza and Violante, M. and Stratigopoulos, H.-G. and Michael, M. K. and Neophytou, S. and Hadjitheophanous, S. and Christou, K. and Skitsas, M. and Bosio, A. and Deveautour, B. and Girard, P. and Traiola, M. and Virazel, A. and Dos Santos, F. Fernandes and Kritikakou, A. and Casagranda, G. and Vallero, M. and Vella, F. and Rech, P. and Bolzani Poehls, L. M. and Krstic, M. and Andjelkovic, M. and Vargas, F. and Tshagharyan, G. and Harutyunyan, G. and Vardanian, V. and Shoukourian, S. and Zorian, Y. and Dworak, J. and Nepal, K. and Manikas, T. and Taouil, M. and Fieback, M. and Gebregiorgis, A. and Bishnoi, R. and Hamdioui, S. and Chatterjee, A. and Saha, A. and Komarraju, S. and Ma, K. and Amarnath, C. and Tahoori, M. and Mayahinia, M. and Rajabalipanah, M. and Basharkhah, K. and Nosrati, N. and Jahanpeima, Z. and Navabi, Z. and Wunderlich, H.-J. and Hellebrand, S.},
  booktitle={2025 IEEE European Test Symposium (ETS)}, 
  title={European Test Symposium Teams: an Anniversary Snapshot}, 
  year={2025},
  volume={},
  number={},
  pages={1-48},
  keywords={},
  doi={10.1109/ETS63895.2025.11049652}}

Jnini, Anas, Breschi, Lorenzo, Vella, Flavio. Riemann Tensor Neural Networks: Learning Conservative Systems with Physics-Constrained Networks. Proceedings of the 42nd International Conference on Machine Learning. (2025)

@InProceedings{pmlr-v267-jnini25a,
  title = 	 {Riemann Tensor Neural Networks: Learning Conservative Systems with Physics-Constrained Networks},
  author =       {Jnini, Anas and Breschi, Lorenzo and Vella, Flavio},
  booktitle = 	 {Proceedings of the 42nd International Conference on Machine Learning},
  pages = 	 {28304--28326},
  year = 	 {2025},
  editor = 	 {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry},
  volume = 	 {267},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {13--19 Jul},
  publisher =    {PMLR},
  pdf = 	 {https://raw.githubusercontent.com/mlresearch/v267/main/assets/jnini25a/jnini25a.pdf},
  url = 	 {https://proceedings.mlr.press/v267/jnini25a.html},
  abstract = 	 {Divergence-free symmetric tensors (DFSTs) are fundamental in continuum mechanics, encoding conservation laws such as mass and momentum conservation. We introduce Riemann Tensor Neural Networks (RTNNs), a novel neural architecture that inherently satisfies the DFST condition to machine precision, providing a strong inductive bias for enforcing these conservation laws. We prove that RTNNs can approximate any sufficiently smooth DFST with arbitrary precision and demonstrate their effectiveness as surrogates for conservative PDEs, achieving improved accuracy across benchmarks. This work is the first to use DFSTs as an inductive bias in neural PDE surrogates and to explicitly enforce the conservation of both mass and momentum within a physics-constrained neural architecture.}
}

Andaloro, Salvatore Domenico, Pasquali, Thomas, Vella, Flavio. Cache-optimized BFS on multi-core CPUs. Proceedings of the 1st FastCode Programming Challenge. (2025)

@inproceedings{10.1145/3711708.3723452,
author = {Andaloro, Salvatore Domenico and Pasquali, Thomas and Vella, Flavio},
title = {Cache-optimized BFS on multi-core CPUs},
year = {2025},
isbn = {9798400714467},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3711708.3723452},
doi = {10.1145/3711708.3723452},
abstract = {Breadth-First Search (BFS) performance on shared-memory systems is often limited by irregular memory access and cache inefficiencies. This work presents two optimizations for BFS graph traversal: a bitmap-based algorithm designed for small-diameter graphs and MergedCSR, a graph storage format that improves cache locality for large-scale graphs. Experimental results on real-world datasets show an average 1.3\texttimes{} speedup over a state-of-the-art implementation, with MergedCSR reducing RAM accesses by approximately 15\%.},
booktitle = {Proceedings of the 1st FastCode Programming Challenge},
pages = {23–27},
numpages = {5},
keywords = {graph, algorithm, breadth-first search, parallel computing, multi-core CPU},
location = {The Westin Las Vegas Hotel \& Spa, Las Vegas, NV, USA},
series = {FCPC '25}
}

Bellavita, Julian, Pasquali, Thomas, Del Rio Martin, Laura, Vella, Flavio, Guidi, Giulia. Popcorn: Accelerating Kernel K-means on GPUs through Sparse Linear Algebra. Proceedings of the 30th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming. (2025)

@inproceedings{10.1145/3710848.3710887,
author = {Bellavita, Julian and Pasquali, Thomas and Del Rio Martin, Laura and Vella, Flavio and Guidi, Giulia},
title = {Popcorn: Accelerating Kernel K-means on GPUs through Sparse Linear Algebra},
year = {2025},
isbn = {9798400714436},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3710848.3710887},
doi = {10.1145/3710848.3710887},
abstract = {K-means is a popular clustering algorithm with significant applications in numerous scientific and engineering areas. One drawback of K-means is its inability to identify non-linearly separable clusters, which may lead to inaccurate solutions in certain cases. Kernel K-means is a variant of classical K-means that can find non-linearly separable clusters. However, it scales quadratically with respect to the size of the dataset, taking several minutes to cluster even medium-sized datasets on traditional CPU-based machines.In this paper, we present a formulation of Kernel K-means using sparse-dense matrix multiplication (SpMM) and sparse matrix-vector multiplication (SpMV), and we show that our formulation enables the rapid implementation of a fast GPU-based version of Kernel K-means with little programming effort. Our implementation, named Popcorn, is the first open-source GPU-based implementation of Kernel K-means.Popcorn achieves a speedup of up to 123.8\texttimes{} over a CPU implementation of Kernel K-means and a speedup of up to 2.6\texttimes{} over a GPU implementation of Kernel K-means that does not use sparse matrix computations. Our results support the effectiveness of sparse matrices as tools for efficient parallel programming.},
booktitle = {Proceedings of the 30th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming},
pages = {426–440},
numpages = {15},
keywords = {GPU, Kernel K-Means, SpMM, SpMV, Sparse Matrix},
location = {Las Vegas, NV, USA},
series = {PPoPP '25}
}

Anas Jnini, Flavio Vella. Dual Natural Gradient Descent for Scalable Training of Physics-Informed Neural Networks. Transactions on Machine Learning Research. (2025)

@article{
jnini2025dual,
title={Dual Natural Gradient Descent for Scalable Training of Physics-Informed Neural Networks},
author={Anas Jnini and Flavio Vella},
journal={Transactions on Machine Learning Research},
issn={2835-8856},
year={2025},
url={https://openreview.net/forum?id=GDHVRy6SDd},
note={}
}

Marzio Vallero, Gioele Casagranda, Flavio Vella, Paolo Rech. Radiation-Induced Fault Detection in Superconducting Quantum Devices. (2025)

@misc{Vallero2025Detection,
      title={Radiation-Induced Fault Detection in Superconducting Quantum Devices}, 
      author={Marzio Vallero and Gioele Casagranda and Flavio Vella and Paolo Rech},
      year={2025},
      eprint={2506.16834},
      archivePrefix={arXiv},
      primaryClass={quant-ph},
      url={https://arxiv.org/abs/2506.16834}, 
}

Casagranda, Gioele, Vallero, Marzio, Vella, Flavio, Rech, Paolo. Understanding the Contributions of Terrestrial Radiation Sources to Error Rates in Quantum Devices. IEEE Transactions on Nuclear Science. (2025)

@ARTICLE{Casagranda2025Understanding,
  author={Casagranda, Gioele and Vallero, Marzio and Vella, Flavio and Rech, Paolo},
  journal={IEEE Transactions on Nuclear Science}, 
  title={Understanding the Contributions of Terrestrial Radiation Sources to Error Rates in Quantum Devices}, 
  year={2025},
  volume={72},
  number={4},
  pages={1324-1334},
  keywords={Qubit;SQUIDs;Josephson junctions;Substrates;Neutrons;Mesons;Particle measurements;Atmospheric measurements;Superconducting epitaxial layers;Sea level;Alpha particles;gamma rays;muons;neutrons;quantum bit (qubit);quantum computing (QC) reliability;simulations;SQUID},
  doi={10.1109/TNS.2025.3537036}}

2024

Pichetti, Lorenzo, De Sensi, Daniele, Sivalingam, Karthee, Nassyr, Stepan, Cesarini, Daniele, Turisini, Matteo, Pleiter, Dirk, Artigiani, Aldo, Vella, Flavio. Benchmarking Ethernet Interconnect for HPC/AI workloads. SC24-W: Workshops of the International Conference for High Performance Computing, Networking, Storage and Analysis. (2024)

@INPROCEEDINGS{10820803,
  author={Pichetti, Lorenzo and De Sensi, Daniele and Sivalingam, Karthee and Nassyr, Stepan and Cesarini, Daniele and Turisini, Matteo and Pleiter, Dirk and Artigiani, Aldo and Vella, Flavio},
  booktitle={SC24-W: Workshops of the International Conference for High Performance Computing, Networking, Storage and Analysis}, 
  title={Benchmarking Ethernet Interconnect for HPC/AI workloads}, 
  year={2024},
  volume={},
  number={},
  pages={869-875},
  keywords={Ranking (statistics);Industries;High performance computing;Ethernet;Bandwidth;Benchmark testing;Artificial intelligence;Low latency communication;Standards;Optimization;interconnect;ethernet;infiniband;hpc/ai workloads;gigabit ethernet},
  doi={10.1109/SCW63240.2024.00124}}

De Sensi, Daniele, Pichetti, Lorenzo, Vella, Flavio, De Matteis, Tiziano, Ren, Zebin, Fusco, Luigi, Turisini, Matteo, Cesarini, Daniele, Lust, Kurt, Trivedi, Animesh, Roweth, Duncan, Spiga, Filippo, Di Girolamo, Salvatore, Hoefler, Torsten. Exploring GPU-to-GPU Communication: Insights into Supercomputer Interconnects. SC24: International Conference for High Performance Computing, Networking, Storage and Analysis. (2024)

@INPROCEEDINGS{10793179,
  author={De Sensi, Daniele and Pichetti, Lorenzo and Vella, Flavio and De Matteis, Tiziano and Ren, Zebin and Fusco, Luigi and Turisini, Matteo and Cesarini, Daniele and Lust, Kurt and Trivedi, Animesh and Roweth, Duncan and Spiga, Filippo and Di Girolamo, Salvatore and Hoefler, Torsten},
  booktitle={SC24: International Conference for High Performance Computing, Networking, Storage and Analysis}, 
  title={Exploring GPU-to-GPU Communication: Insights into Supercomputer Interconnects}, 
  year={2024},
  volume={},
  number={},
  pages={1-15},
  keywords={Performance evaluation;Noise;Bandwidth;Routing;Supercomputers;Software;Distance measurement;Libraries;Optimization;Tuning},
  doi={10.1109/SC41406.2024.00039}}

Okanovic, Patrik, Kwasniewski, Grzegorz, Labini, Paolo Sylos, Besta, Maciej, Vella, Flavio, Hoefler, Torsten. High Performance Unstructured SpMM Computation Using Tensor Cores. SC24: International Conference for High Performance Computing, Networking, Storage and Analysis. (2024)

@INPROCEEDINGS{10793184,
  author={Okanovic, Patrik and Kwasniewski, Grzegorz and Labini, Paolo Sylos and Besta, Maciej and Vella, Flavio and Hoefler, Torsten},
  booktitle={SC24: International Conference for High Performance Computing, Networking, Storage and Analysis}, 
  title={High Performance Unstructured SpMM Computation Using Tensor Cores}, 
  year={2024},
  volume={},
  number={},
  pages={1-14},
  keywords={Training;Schedules;Tensors;Scientific computing;Graphics processing units;Data structures;Libraries;Hardware;Sparse matrices;Optimization;Mathematics of computing;SpMM;Matrix Multiplication;Tensor Cores},
  doi={10.1109/SC41406.2024.00060}}

Labini, Paolo Sylos, Jurco, Andrej, Ceccarello, Matteo, Guarino, Stefano, Mastrostefano, Enrico, Vella, Flavio. Scaling Expected Force: Efficient Identification of Key Nodes in Network-Based Epidemic Models. 2024 32nd Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP). (2024)

@INPROCEEDINGS{10495558,
  author={Labini, Paolo Sylos and Jurco, Andrej and Ceccarello, Matteo and Guarino, Stefano and Mastrostefano, Enrico and Vella, Flavio},
  booktitle={2024 32nd Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP)}, 
  title={Scaling Expected Force: Efficient Identification of Key Nodes in Network-Based Epidemic Models}, 
  year={2024},
  volume={},
  number={},
  pages={98-107},
  keywords={Measurement;Epidemics;Force measurement;Scalability;Computational modeling;Force;Graphics processing units;Epidemic;SIR;Big Data;Expected Force;Graph Centrality;Network;Parallel Computing},
  doi={10.1109/PDP62718.2024.00021}}

Santos, Fernando Fernandes Dos, Carro, Luigi, Vella, Flavio, Rech, Paolo. Assessing the Impact of Compiler Optimizations on GPUs Reliability. ACM Trans. Archit. Code Optim.. (2024)

@article{10.1145/3638249,
author = {Santos, Fernando Fernandes Dos and Carro, Luigi and Vella, Flavio and Rech, Paolo},
title = {Assessing the Impact of Compiler Optimizations on GPUs Reliability},
year = {2024},
issue_date = {June 2024},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {21},
number = {2},
issn = {1544-3566},
url = {https://doi.org/10.1145/3638249},
doi = {10.1145/3638249},
abstract = {Graphics Processing Units (GPUs) compilers have evolved in order to support general-purpose programming languages for multiple architectures. NVIDIA CUDA Compiler (NVCC) has many compilation levels before generating the machine code and applies complex optimizations to improve performance. These optimizations modify how the software is mapped in the underlying hardware; thus, as we show in this article, they can also affect GPU reliability. We evaluate the effects on the GPU error rate of the optimization flags applied at the NVCC Parallel Thread Execution (PTX) compiling phase by analyzing two NVIDIA GPU architectures (Kepler and Volta) and two compiler versions (NVCC 10.2 and 11.3). We compare and combine fault propagation analysis based on software fault injection, hardware utilization distribution obtained with application-level profiling, and machine instructions radiation-induced error rate measured with beam experiments. We consider eight different workloads and 144 combinations of compilation flags, and we show that optimizations can impact the GPUs’ error rate of up to an order of magnitude. Additionally, through accelerated neutron beam experiments on a NVIDIA Kepler GPU, we show that the error rate of the unoptimized GEMM (-O0 flag) is lower than the optimized GEMM’s (-O3 flag) error rate. When the performance is evaluated together with the error rate, we show that the most optimized versions (-O1 and -O3) always produce a higher amount of correct data than the unoptimized code (-O0).},
journal = {ACM Trans. Archit. Code Optim.},
month = feb,
articleno = {26},
numpages = {22},
keywords = {Graphics processing units, reliability, neutron-induced errors, error rate, reliability}
}

Vallero, Marzio, Dri, Emanuele, Giusto, Edoardo, Montrucchio, Bartolomeo, Rech, Paolo. Understanding Logical-Shift Error Propagation in Quanvolutional Neural Networks. IEEE Transactions on Quantum Engineering. (2024)

@ARTICLE{Vallero2024Understanding,
  author={Vallero, Marzio and Dri, Emanuele and Giusto, Edoardo and Montrucchio, Bartolomeo and Rech, Paolo},
  journal={IEEE Transactions on Quantum Engineering}, 
  title={Understanding Logical-Shift Error Propagation in Quanvolutional Neural Networks}, 
  year={2024},
  volume={5},
  number={},
  pages={1-14},
  keywords={Qubit;Circuit faults;Reliability;Quantum computing;Quantum circuit;Neural networks;Integrated circuit reliability;Fault tolerance;Machine learning;Fault injection;fault tolerance;quantum computing (QC);quantum machine learning (QML);reliability evaluation},
  doi={10.1109/TQE.2024.3372880}}

Vallero, Marzio, Casagranda, Gioele, Vella, Flavio, Rech, Paolo. On the Efficacy of Surface Codes in Compensating for Radiation Events in Superconducting Devices. SC24: International Conference for High Performance Computing, Networking, Storage and Analysis. (2024)

@INPROCEEDINGS{Vallero2024Efficacy,
  author={Vallero, Marzio and Casagranda, Gioele and Vella, Flavio and Rech, Paolo},
  booktitle={SC24: International Conference for High Performance Computing, Networking, Storage and Analysis}, 
  title={On the Efficacy of Surface Codes in Compensating for Radiation Events in Superconducting Devices}, 
  year={2024},
  volume={},
  number={},
  pages={1-15},
  keywords={Computers;Superconducting devices;Codes;Sensitivity;Error analysis;Qubit;Stability analysis;Topology;Tuning;Resilience;surface code;quantum error correction;quantum reliability;quantum fault injection},
  doi={10.1109/SC41406.2024.00075}}

2023

Besta, Maciej, Renc, Pawel, Gerstenberger, Robert, Sylos Labini, Paolo, Ziogas, Alexandros, Chen, Tiancheng, Gianinazzi, Lukas, Scheidl, Florian, Szenes, Kalman, Carigiet, Armon, Iff, Patrick, Kwasniewski, Grzegorz, Kanakagiri, Raghavendra, Ge, Chio, Jaeger, Sammy, W\k{a. High-Performance and Programmable Attentional Graph Neural Networks with Global Tensor Formulations. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. (2023)

@inproceedings{10.1145/3581784.3607067,
author = {Besta, Maciej and Renc, Pawel and Gerstenberger, Robert and Sylos Labini, Paolo and Ziogas, Alexandros and Chen, Tiancheng and Gianinazzi, Lukas and Scheidl, Florian and Szenes, Kalman and Carigiet, Armon and Iff, Patrick and Kwasniewski, Grzegorz and Kanakagiri, Raghavendra and Ge, Chio and Jaeger, Sammy and W\k{a}s, Jaros\l{}aw and Vella, Flavio and Hoefler, Torsten},
title = {High-Performance and Programmable Attentional Graph Neural Networks with Global Tensor Formulations},
year = {2023},
isbn = {9798400701092},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3581784.3607067},
doi = {10.1145/3581784.3607067},
abstract = {Graph attention models (A-GNNs), a type of Graph Neural Networks (GNNs), have been shown to be more powerful than simpler convolutional GNNs (C-GNNs). However, A-GNNs are more complex to program and difficult to scale. To address this, we develop a novel mathematical formulation, based on tensors that group all the feature vectors, targeting both training and inference of A-GNNs. The formulation enables straightforward adoption of communication-minimizing routines, it fosters optimizations such as vectorization, and it enables seamless integration with established linear algebra DSLs or libraries such as GraphBLAS. Our implementation uses a data redistribution scheme explicitly developed for sparse-dense tensor operations used heavily in GNNs, and fusing optimizations that further minimize memory usage and communication cost. We ensure theoretical asymptotic reductions in communicated data compared to the established message-passing GNN paradigm. Finally, we provide excellent scalability and speedups of even 4--5x over modern libraries such as Deep Graph Library.},
booktitle = {Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis},
articleno = {66},
numpages = {16},
keywords = {graph attention models, graph neural networks, sparse-dense tensor operations},
location = {Denver, CO, USA},
series = {SC '23}
}

Turchet, Luca, Vella, Flavio, Fiore, Sandro Luigi. The potential of high-performance computing for the Internet of Sounds. 2023 4th International Symposium on the Internet of Sounds. (2023)

@INPROCEEDINGS{10335270,
  author={Turchet, Luca and Vella, Flavio and Fiore, Sandro Luigi},
  booktitle={2023 4th International Symposium on the Internet of Sounds}, 
  title={The potential of high-performance computing for the Internet of Sounds}, 
  year={2023},
  volume={},
  number={},
  pages={1-7},
  keywords={Industries;High performance computing;Music;Distributed databases;Sensors;Internet of Things;Internet of Sounds;High-Performance Computing;audio processing},
  doi={10.1109/IEEECONF59510.2023.10335270}}

Bernaschi, Massimo, Celestini, Alessandro, Vella, Flavio, D'Ambra, Pasqua. A Multi-GPU Aggregation-Based AMG Preconditioner for Iterative Linear Solvers. IEEE Transactions on Parallel and Distributed Systems. (2023)

@ARTICLE{10155151,
  author={Bernaschi, Massimo and Celestini, Alessandro and Vella, Flavio and D'Ambra, Pasqua},
  journal={IEEE Transactions on Parallel and Distributed Systems}, 
  title={A Multi-GPU Aggregation-Based AMG Preconditioner for Iterative Linear Solvers}, 
  year={2023},
  volume={34},
  number={8},
  pages={2365-2376},
  keywords={Sparse matrices;Graphics processing units;Scalability;Linear systems;Iterative methods;Convergence;Linear accelerators;GPU accelerators;heterogeneous computing;iterative sparse linear solvers;parallel numerical algorithms;scalability},
  doi={10.1109/TPDS.2023.3287238}}

2022

Labini, Paolo Sylos, Bernaschi, Massimo, Nutt, Werner, Silvestri, Francesco, Vella, Flavio. Blocking Sparse Matrices to Leverage Dense-Specific Multiplication. 2022 IEEE/ACM Workshop on Irregular Applications: Architectures and Algorithms (IA3). (2022)

@INPROCEEDINGS{10027183,
  author={Labini, Paolo Sylos and Bernaschi, Massimo and Nutt, Werner and Silvestri, Francesco and Vella, Flavio},
  booktitle={2022 IEEE/ACM Workshop on Irregular Applications: Architectures and Algorithms (IA3)}, 
  title={Blocking Sparse Matrices to Leverage Dense-Specific Multiplication}, 
  year={2022},
  volume={},
  number={},
  pages={19-24},
  keywords={Deep learning;Tensors;Costs;Conferences;Focusing;Computer architecture;Software;Reordering;clustering;sparse matrices;tensor cores},
  doi={10.1109/IA356718.2022.00009}}

Besta, Maciej, Miglioli, Cesare, Labini, Paolo Sylos, Tětek, Jakub, Iff, Patrick, Kanakagiri, Raghavendra, Ashkboos, Saleh, Janda, Kacper, Podstawski, Michał, Kwaśniewski, Grzegorz, Gleinig, Niels, Vella, Flavio, Mutlu, Onur, Hoefler, Torsten. ProbGraph: High-Performance and High-Accuracy Graph Mining with Probabilistic Set Representations. SC22: International Conference for High Performance Computing, Networking, Storage and Analysis. (2022)

@INPROCEEDINGS{10046121,
  author={Besta, Maciej and Miglioli, Cesare and Labini, Paolo Sylos and Tětek, Jakub and Iff, Patrick and Kanakagiri, Raghavendra and Ashkboos, Saleh and Janda, Kacper and Podstawski, Michał and Kwaśniewski, Grzegorz and Gleinig, Niels and Vella, Flavio and Mutlu, Onur and Hoefler, Torsten},
  booktitle={SC22: International Conference for High Performance Computing, Networking, Storage and Analysis}, 
  title={ProbGraph: High-Performance and High-Accuracy Graph Mining with Probabilistic Set Representations}, 
  year={2022},
  volume={},
  number={},
  pages={1-17},
  keywords={Matched filters;Runtime;Data analysis;Instruction sets;High performance computing;Clustering algorithms;Filtering algorithms;Approximate Graph Mining;Approximate Graph Pattern Matching;Approximate Triangle Counting;Approximate Community Detection;Approximate Graph Clustering;Bloom Filters;MinHash;K Minimum Values;High-Performance Graph Computations;Graph Sketching},
  doi={10.1109/SC41404.2022.00048}}