Quickly understanding the resource usage of a model is a common workflow for anyone designing an FPGA targeted model. Often the developer won't have access to vivado or a PC capable of running it. A website where the user can upload a trained model, hls4ml can run on the model, produce firmware estimates and a package of the associated firmware could solve this issue. A web interface GUI could...
The Global Event Processor (GEP) FPGA is an area-constrained, performance-critical element of the Large Hadron Collider's (LHC) ATLAS experiment. It needs to very quickly determine which small fraction of detected events should be retained for further processing, and which other events will be discarded. This system involves a large number of individual processing tasks, brought together...
For the HL-LHC upgrade of the ATLAS TDAQ system, a heterogeneous computing farm deploying GPUs and/or FPGAs is under study, together with the use of modern machine learning algorithms such as Graph Neural Networks (GNNs). We present a study on the reconstruction of tracks in the ATLAS Inner Tracker using GNNs on FPGAs for the Event Filter system. We explore each of the steps in a GNN-based...
With hls4ml now deployed online at CMS, we may look back on the deployment experience with a view to improvements for the years ahead. In particular, the compile-time fixing of network architecture and parameters has some implications. In emulation, model updates are tied to CMS Software release schedules and cannot be updated on changing conditions. Likewise in firmware, models are baked into...
Highly granular pixel detectors allow for increasingly precise measurements of charged particle tracks. At the High Luminosity Large Hadron Collider, data rates from the pixel detector exceed those feasible for integration in the Level 1 trigger. However, the shape of charge clusters deposited in the pixel sensors can be used to determine the physical properties of the traversing particle,...