We develop an end-to-end workflow for the training and implementation of co-designed neural networks (NNs) for efficient field-programmable gate array (FPGA). Our approach leverages Hessian-aware quantization (HAWQ) of NNs, the Quantized Open Neural Network Exchange (QONNX) intermediate representation, and the hls4ml tool flow for transpiling NNs into FPGA firmware. This makes efficient NN...
To increase the science rate for high data rates/volumes, Thomas Jefferson National Accelerator Facility (JLab) has partnered with Energy Sciences Network (ESnet) to define and implement an edge to compute cluster data event load balancing architecture with hardware accelerated elements to address compression, fragmentation, dynamically switched destination transport and reassembly at the...
Proton computed tomography (pCT) is a novel three-dimensional medical imaging technique proposed for the pre-treatment diagnosis of patients undergoing proton therapy. The two main components of our pCT prototype are a proton tracker and a calorimeter. A proton tracker allows precise tracking of proton trajectories, while a proton calorimeter provides accurate measurement of their energies....
Effective control of experiment flow and real-time monitoring are vital to the growing bandwidth and complexity of data acquisition systems. Unfortunately, platforms like EPICS and LabVIEW are inadequate for handling high data rates. To address this, we create a system that connects a high-speed processing component with a visualization component through a video streaming interface. We further...
The next era of LHC experiments will provide an unprecedented volume of data, aiming to achieve a tenfold increase in integrated luminosity. Processing these data presents formidable computing challenges. In the case of the LHCb detector, a fully software based trigger has been employed in its current design, which processes events at ~30MHz. Currently, GPU-based compute acceleration is...
Track finding in high-density environments is a key challenge for experiments at modern accelerators. In this presentation we describe the performance obtained running machine learning models studied for the ATLAS Muon High Level Trigger. These models are designed for hit position reconstruction and track pattern recognition with a tracking detector, on a commercially available Xilinx FPGA:...
