-
Giacomo Tenaglia (CERN)13/11/2026, 14:00
-
René Scholz13/11/2026, 14:05
This talk presents the latest developments and current status of IHP's openPDK, openADK, and open-silicon MPW activities, highlighting ongoing efforts to enable productive chip design using open-source EDA tools.
A particular focus is placed on recent improvements in analog and RF design flows, including the integration of open-source electromagnetic simulators, as well as the development...
Go to contribution page -
Frank Gürkaynak13/11/2026, 14:35
Modern IC design is a complex process that involves several different specialized electronic design automation tools resulting in ICs that cost many millions to manufacture. Is it possible that we can use openly available tools for meaningful designs? Can open-source tools come close to or even surpass proprietary tools in terms of the performance of the chips designed using them? What is...
Go to contribution page -
Mathias Schindler13/11/2026, 15:05
We need to talk about the health of open source: which projects are thriving, which are struggling, and what risks the ecosystem may be facing. But how much of that picture can we actually observe, and how much is inference, anecdote or narrative? Drawing on GitHub’s perspective on the ecosystem, I ask what available signals there are and what they can and cannot tell us about open source in a...
Go to contribution page -
Lucas Nussbaum13/11/2026, 16:00
AI-assisted software engineering is rapidly and fundamentally altering how code is written, reviewed, and sustained. Beyond technical workflows, it is profoundly challenging the social dimensions of software development—forcing communities to reconsider how they evaluate, govern, and integrate AI-generated contributions. These questions have become central to heated debates across Free...
Go to contribution page -
Tomas Roun (CERN)13/11/2026, 16:30
A quick look into how CPython is managing the influx of LLM-generated (good and bad) PRs, what we've learned so far and what the next steps might be.
Go to contribution page -
Elena de la Fuente Garcia13/11/2026, 16:35
The interaction between particle beams and their surrounding accelerator components can generate electromagnetic wakefields that affect beam stability and compromise component integrity through beam-induced heating. Accurately modelling these interactions through beam-coupling impedance is essential for the design and operation of particle accelerators.
In alignment with CERN's...
Go to contribution page -
Benoit Denkinger13/11/2026, 16:40
Modern System-on-Chip projects mix RTL in several HDLs, SystemC models, UVM testbenches, generated register files and the embedded firmware that runs on the chip. These pieces are usually handled by separate, often hand-written scripts. SoCMake is an open-source (LGPL-3.0) CMake library that brings all of this under one build system. Rather than inventing yet another hardware build tool, it...
Go to contribution page -
Anastasiia Petrovych (CERN)13/11/2026, 16:45
We present PQuantML, an open-source library for end-to-end hardware-aware model compression that enables the training and deployment of compact, high-performance neural networks on resource-constrained hardware in physics and beyond. PQuantML abstracts away the low-level details of compression by letting users compress models with a simple configuration file and an API call. It enables the use...
Go to contribution page -
Dimitrios Danopoulos (CERN)13/11/2026, 16:50
We present AIE4ML, a framework that enables the compilation and optimization of neural networks for low-latency execution on AMD Versal AI Engine architectures. It includes a library of over 20 modular kernels covering essential neural network computations, enabling the implementation of diverse models, such as MLPs, ResNets, Deep Sets, and Transformers. The framework automates the translation...
Go to contribution page -
Sanjiban Sengupta (CERN, University of Manchester)13/11/2026, 16:55
SOFIE (System for Optimized Fast Inference code Emit), developed by the ML4EP Project at CERN, translates trained machine learning models into self-contained, low-latency C++ code that is portable across hardware architectures and highly optimized, while only being dependent to BLAS libraries.
SOFIE achieves portability across heterogeneous architectures by leveraging the abstract...
Go to contribution page -
13/11/2026, 17:00
-
13/11/2026, 17:20
The OSPO needs your input: we do this for you, to help open-source projects. Please come with your ideas and concerns.
Go to contribution page
After the event, you can share your feedback by sending an email to Open.Source@cern.ch or at the forum. -
Choose timezone
Your profile timezone: