arXiv subjects

Roberto Bosio

Publications and source records attributed to Roberto Bosio.

3 recordsLinked to original sources

arXiv2026

Seeing through water: diffuse image-based depth measurements in three-dimensional dam-break flows

In this work we present a dedicated experimental facility and an image-based method for measuring water depth in a radially spreading dam-break wave propagating over a horizontal plane. The facility consists of a prismatic reservoir containing a known volume of water dyed with a soluble colorant and equipped with a removable vertical breach whose geometry can be varied, and a 6.4 m x 3.4 m plane that can be inclined from 0{\deg} to 30{\deg}. The plane is enclosed within a light box providing highly uniform illumination through an array of 60 LED floodlights. Wave propagation is captured by two scientific CMOS cameras mounted on the ceiling of the light box, which record the spatial and temporal evolution of the dye-induced color intensity associated with the advancing water layer. Preliminary dry calibration tests were conducted to assess the spectral compatibility between the broadband white-LED emission, the CMOS sensor sensitivity, and the absorption properties of several dyes at different concentrations. This analysis identified the dye providing the highest attenuation within the effective spectral band of the imaging system, ensuring sensitivity to very small optical path lengths. Based on this characterization, a bi-exponential model is introduced to relate the normalized gray level to the optical path length. A series of dam-break experiments with five initial reservoir levels was performed to assess statistical repeatability. The high consistency observed across the repeated tests confirms the robustness of the measurement procedure. The validity of the reconstructed depth fields is further supported by independent estimates of the water volume released from the reservoir, obtained from an array of ultrasonic level sensors and from a calibrated analytical emptying model. Together, these comparisons confirm the reliability and accuracy of the proposed methodology.

physics.flu-dyn↗

arXiv2025

eIQ Neutron: Redefining Edge-AI Inference with Integrated NPU and Compiler Innovations

Neural Processing Units (NPUs) are key to enabling efficient AI inference in resource-constrained edge environments. While peak tera operations per second (TOPS) is often used to gauge performance, it poorly reflects real-world performance and typically rather correlates with higher silicon cost. To address this, architects must focus on maximizing compute utilization, without sacrificing flexibility. This paper presents the eIQ Neutron efficient-NPU, integrated into a commercial flagship MPU, alongside co-designed compiler algorithms. The architecture employs a flexible, data-driven design, while the compiler uses a constrained programming approach to optimize compute and data movement based on workload characteristics. Compared to the leading embedded NPU and compiler stack, our solution achieves an average speedup of 1.8x (4x peak) at equal TOPS and memory resources across standard AI-benchmarks. Even against NPUs with double the compute and memory resources, Neutron delivers up to 3.3x higher performance.

cs.AR↗

arXiv2024

SILVIA: Automated Superword-Level Parallelism Exploitation via HLS-Specific LLVM Passes for Compute-Intensive FPGA Accelerators

High-level synthesis (HLS) aims at democratizing custom hardware acceleration with highly abstracted software-like descriptions. However, efficient accelerators still require substantial low-level hardware optimizations, defeating the HLS intent. In the context of field-programmable gate arrays, digital signal processors (DSPs) are a crucial resource that typically requires a significant optimization effort for its efficient utilization, especially when used for sub-word vectorization. This work proposes SILVIA, an open-source LLVM transformation pass that automatically identifies superword-level parallelism within an HLS design and exploits it by packing multiple operations, such as additions, multiplications, and multiply-and-adds, into a single DSP. SILVIA is integrated in the flow of the commercial AMD Vitis HLS tool and proves its effectiveness by packing multiple operations on the DSPs without any manual source-code modifications on several diverse state-of-the-art HLS designs such as convolutional neural networks and basic linear algebra subprograms accelerators, reducing the DSP utilization for additions by 70 % and for multiplications and multiply-and-adds by 50 % on average.

cs.AR↗