FINN-R

By Michaela Blott, Thomas B. Preußer, Nicholas J. Fraser, Giulio Gambardella, Kenneth O'Brien, Yaman Umuroglu, Miriam Leeser, Kees A. Vissers. ACM Transactions on Reconfigurable Technology and Systems, 2018

375 citations recorded by the source corpus.

Convolutional Neural Networks have rapidly become the most successful machine-learning algorithm, enabling ubiquitous machine vision and intelligent decisions on even embedded computing systems. While the underlying arithmetic is structurally simple, compute and memory requirements are challenging. One of the promising opportunities is leveraging reduced-precision representations for inputs, activations, and model parameters. The resulting scalability in performance, power efficiency, and storage footprint provides interesting design compromises in exchange for a small reduction in accuracy. FPGAs are ideal for exploiting low-precision inference engines leveraging custom precisions to achieve the required numerical accuracy for a given application. In this article, we describe the second generation of the FINN framework, an end-to-end tool that enables design-space exploration and automates the creation of fully customized inference engines on FPGAs. Given a neural network description, the tool optimizes for given platforms, design targets, and a specific precision. We introduce formalizations of resource cost functions and performance predictions and elaborate on the optimization algorithms. Finally, we evaluate a selection of reduced precision neural networks ranging from CIFAR-10 classifiers to YOLO-based object detection on a range of platforms including PYNQ and AWS F1, demonstrating new unprecedented measured throughput at 50 TOp/s on AWS F1 and 5 TOp/s on embedded devices.

Subfields: Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Hardware and Architecture

View Paper: https://www.semanticscholar.org/paper/8daa392b62968206885f3e84c3d96cf04d48a5b4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Kurt Keutzer
2. Zhen Dong
3. Bichen Wu
4. Yaohui Cai
5. Qijing Huang
6. Dequan Wang
7. Luca Benini
8. John Wawrzynek

In Electrical and Electronic Engineering

1. Zhen Dong
2. Kurt Keutzer
3. Bichen Wu
4. Tian Li
5. Kees A. Vissers
6. Luca Benini
7. Giulio Gambardella
8. Michaela Blott

In Hardware and Architecture

1. Kurt Keutzer
2. Luca Benini
3. Giovanni De Micheli
4. Srinivasan Murali

Citations in the corpus

• Efficient Acceleration of Deep Learning Inference on Resource-Constrained Edge Devices: A Review (2023, methodology): “Xilinx provides quantization support through the FINN-R framework for inference realization on FPGA [413].”
• Synetgy: Algorithm-hardware Co-design for ConvNet Accelerators on Embedded FPGAs (2018, background): “VGG-SVD[39] AlexNet[48] VGG16[49] VGG16 [40] DoReFa[42] FINN-R [41] Ours”
• Memory-Driven Mixed Low Precision Quantization For Enabling Deep Network Inference On Microcontrollers (2019, methodology): “The work [1] introduced FINN-R to quantize and deploy a generic model into constrained FPGA architectures.”
• CoDeNet: Efficient Deployment of Input-Adaptive Object Detection on Embedded FPGAs (2021, methodology): “As shown in Figure 8 and Table 5, compared to the results from FINN-R [2] [28], the state-of-the-art embedded FPGA accelerator design targeting VOC, our configuration a and b (with single-batch inference latency of 31ms and 37ms respectively) achieve both higher accuracy, higher framerate, and lower”