Abstract:Extreme edge-AI systems, such as those in readout ASICs for radiation detection, must operate under stringent hardware constraints such as micron-level dimensions, sub-milliwatt power, and nanosecond-scale speed while providing clear accuracy advantages over traditional architectures. Finding ideal solutions means identifying optimal AI and ASIC design choices from a design space that has explosively expanded during the merger of these domains, creating non-trivial couplings which together act upon a small set of solutions as constraints tighten. It is impractical, if not impossible, to manually determine ideal choices among possibilities that easily exceed billions even in small-size problems. Existing methods to bridge this gap have leveraged theoretical understanding of hardware to f architecture search. However, the assumptions made in computing such theoretical metrics are too idealized to provide sufficient guidance during the difficult search for a practical implementation. Meanwhile, theoretical estimates for many other crucial metrics (like delay) do not even exist and are similarly variable, dependent on parameters of the process design kit (PDK). To address these challenges, we present a study that employs intelligent search using multi-objective Bayesian optimization, integrating both neural network search and ASIC synthesis in the loop. This approach provides reliable feedback on the collective impact of all cross-domain design choices. We showcase the effectiveness of our approach by finding several Pareto-optimal design choices for effective and efficient neural networks that perform real-time feature extraction from input pulses within the individual pixels of a readout ASIC.

Design flow for readout ASICs in High-energy Physics experiments

An Analog Front-end ASIC Design for 16-Channel SiPM Readout in PET Detectors

A multichannel front end ASIC for PMT readout in LHAASO WCDA

Design of Low Noise CMOS Charge Sensitive Amplifier ASIC

A High-Speed Low-Noise Front-End ASIC Prototype for Ionization Chambers of High-Intensity Beam Monitoring

Performance study of ASIC readout for electromagnetic calorimeter with injected charge

Design of A Monolithic Multi-Channel Front-End Readout Asic for Lyso/Sipm-Based Small-Animal Flat-Panel Pet Imaging

An Analogue Front-End ASIC Prototype Designed For PMT Signal Readout

Reusable Verification Components for High-Energy Physics readout ASICs

FATALIC: A novel CMOS front-end readout ASIC for the ATLAS Tile Calorimeter

Design and Testing of a Prototype ASIC for the MTPC Readout at CSNS Back-n White Neutron Source

Design of a Digital Readout Module for SCA ASIC Testing

Design of a Prototype Analog Front-End ASIC for the LGAD Readout

A Mixed-Signal Micromegas Readout ASIC Prototype With Automatic Charge Measurement Range Calibration

A compact and highly integrated 128-channel FPGA-based readout electronics for nuclear imaging application

Automated and Holistic Co-design of Neural Networks and ASICs for Enabling In-Pixel Intelligence

A Highly Linear FPGA-Based TDC and a Low-Power Multichannel Readout ASIC with a Shared SAR ADC for SiPM Detectors

A Hardware Co-design Workflow for Scientific Instruments at the Edge

Application of the VMM ASIC for SiPM-based calorimetry

Testing of Front-End Readout Prototype ASICs Designed for WCDA in LHAASO

Development of a Low Power and High Integration Readout ASIC for Time Projection Chambers in 65 Nm CMOS