FINN+ Workshop (Deploying Fast and Energy Efficient Neural Networks to FPGAs)

Wednesday Dec 10, 2025, 9:00 AM → 11:30 AM Europe/Berlin

X0.101 (Paderborn University, Building X)

FINN+ is an experimental, open-source framework that optimises deep neural network inference on field-programmable gate arrays (FPGAs).

Designed for quantized neural networks, it automatically generates customised dataflow-style architectures, resulting in highly efficient FPGA accelerators with high throughput and low latency.

The workshop has two main objectives: to understand the concepts of FINN+ and to use FINN+ in practical, hands-on sessions on the FPGAs of Noctua 2.

The workshop is therefore divided into three parts:

1. When and why use FINN+? (30 minutes)
2. Guided hands-on tutorial: Deploying a neural network to FPGAs (90 minutes)
3. Going beyond a single FPGA. Multi-FPGA neural network inference (about 30 minutes).

The workshop is a hybrid event, combining in-person and virtual participation.