Inference for state-of-the-art deep neural networks is computationally expensive, making them difficult to deploy on constrained hardware environments. An efficient way to reduce this complexity is to quantize the weight parameters and/or activations during training by approximating their distributions with a limited entry codebook. For very low-precisions, such as binary or ternary networks with 1-8-bit activations, the information loss from quantization leads to significant accuracy degradation due to large gradient mismatches between the forward and backward functions. In this paper, we introduce a quantization method to reduce this loss by learning a symmetric codebook for particular weight subgroups. These subgroups are determined based on their locality in the weight matrix, such that the hardware simplicity of the low-precision representations is preserved. Empirically, we show that symmetric quantization can substantially improve accuracy for networks with extremely low-precision weights and activations. We also demonstrate that this representation imposes minimal or no hardware implications to more coarse-grained approaches. Source code is available at https://www.github.com/julianfaraone/SYQ.

Research has shown that convolutional neural networks contain significant redundancy, and high classification accuracy can be obtained even when weights and activations are reduced from floating point to binary values. In this paper, we present Finn, a framework for building fast and flexible FPGA accelerators using a flexible heterogeneous streaming architecture. By utilizing a novel set of optimizations that enable efficient mapping of binarized neural networks to hardware, we implement fully connected, convolutional and pooling layers, with per-layer compute resources being tailored to user-provided throughput requirements. On a ZC706 embedded FPGA platform drawing less than 25 W total system power, we demonstrate up to 12.3 million image classifications per second with 0.31 µs latency on the MNIST dataset with 95.8% accuracy, and 21906 image classifications per second with 283 µs latency on the CIFAR-10 and SVHN datasets with respectively 80.1% and 94.9% accuracy. To the best of our knowledge, ours are the fastest classification rates reported to date on these benchmarks.

It is well known that many types of artificial neural networks, including recurrent networks, can achieve a high classification accuracy even with low-precision weights and activations. The reduction in precision generally yields much more efficient hardware implementations in regards to hardware cost, memory requirements, energy, and achievable throughput. In this paper, we present the first systematic exploration of this design space as a function of precision for Bidirectional Long Short-Term Memory (BiLSTM) neural network. Specifically, we include an in-depth investigation of precision vs. accuracy using a fully hardware-aware training flow, where during training quantization of all aspects of the network including weights, input, output and in-memory cell activations are taken into consideration. In addition, hardware resource cost, power consumption and throughput scalability are explored as a function of precision for FPGA-based implementations of BiLSTM, and multiple approaches of parallelizing the hardware. We provide the first open source HLS library extension of FINN [1] for parameterizable hardware architectures of LSTM layers on FPGAs which offers full precision flexibility and allows for parameterizable performance scaling offering different levels of parallelism within the architecture. Based on this library, we present an FPGA-based accelerator for BiLSTM neural network designed for optical character recognition, along with numerous other experimental proof points for a Zynq UltraScale+ XCZU7EV MPSoC within the given design space.

Despite network monitoring and testing being critical for computer networks, current solutions are both extremely expensive and inflexible. Into this lacuna we launch the Open Source Network Tester (OSNT), a fully open-source traffic generator and capture system. Our prototype implementation on the NetFPGA-10G supports 4×10Gbps traffic generation across all packet sizes and traffic capture is supported up to 2×10Gbps with naïve host software. Our system implementation provides methods for scaling and coordinating multiple generator/capture systems and supports 6.25ns timestamp resolution with clock drift and phase coordination maintained by a GPS input. Additionally, our approach has demonstrated lower-cost than comparable commercial systems while achieving comparable levels of precision and accuracy; all within an open-source framework extensible with new features to support new applications, while permitting validation and review of the implementation.

Binarized neural networks (BNNs) are gaining interest in the deep learning community due to their significantly lower computational and memory cost. They are particularly well suited to reconfigurable logic devices, which contain an abundance of fine-grained compute resources and can result in smaller, lower power implementations, or conversely in higher classification rates. Towards this end, the Finn framework was recently proposed for building fast and flexible field programmable gate array (FPGA) accelerators for BNNs. Finn utilized a novel set of optimizations that enable efficient mapping of BNNs to hardware and implemented fully connected, non-padded convolutional and pooling layers, with per-layer compute resources being tailored to user-provided throughput requirements. However, FINN was not evaluated on larger topologies due to the size of the chosen FPGA, and exhibited decreased accuracy due to lack of padding. In this paper, we improve upon Finn to show how padding can be employed on BNNs while still maintaining a 1-bit datapath and high accuracy. Based on this technique, we demonstrate numerous experiments to illustrate flexibility and scalability of the approach. In particular, we show that a large BNN requiring 1.2 billion operations per frame running on an ADM-PCIE-8K5 platform can classify images at 12 kFPS with 671 µs latency while drawing less than 41 W board power and classifying CIFAR-10 images at 88.7% accuracy. Our implementation of this network achieves 14.8 trillion operations per second. We believe this is the fastest classification rate reported to date on this benchmark at this level of accuracy.

TCP/IP is the predominant communication protocol in modern networks but also one of the most demanding. Consequently, TCP/IP offload is becoming increasingly popular with standard network interface cards. TCP/IP Offload Engines have also emerged for FPGAs, and are being offered by vendors such as Intilop, Fraunhofer HHI, PLDA and Dini Group. With the target application being high-frequency trading, these implementations focus on low latency and support a limited session count. However, many more applications beyond high-frequency trading can potentially be accelerated inside an FPGA once TCP with high session count is available inside the fabric. This way, a network-attached FPGA on ingress and egress to a CPU can accelerate functions such as encryption, compression, memcached and many others in addition to running the complete network stack. This paper introduces a novel architecture for a 10 Gbps linerate TCP/IP stack for FPGAs that can scale with the number of sessions and thereby addresses these new applications. We prototyped the design on a VC709 development board, demonstrating compatibility with existing network infrastructure, operating at full 10 Gbps throughput full-duplex while supporting 10,000 sessions. Finally, the design has been described primarily using high-level synthesis, which accelerates development time and improves maintainability.

Current High-Frequency Trading (HFT) platforms are typically implemented in software on computers with highperformance network adapters. The high and unpredictable latency of these systems has led the trading world to explore alternative "hybrid" architectures with hardware acceleration. In this paper, we survey existing solutions and describe how FPGAs are being used in electronic trading to approach the goal of zero latency. We present an FPGA IP library which implements networking, I/O, memory interfaces and financial protocol parsers. The library provides pre-built infrastructure which accelerates the development and verification of new financial applications. We have developed an example financial application using the IP library on a custom 1U FPGA appliance. The application sustains 10Gb/s Ethernet line rate with a fixed end-toend latency of 1µs -up to two orders of magnitude lower than comparable software implementations.

Deployment of deep neural networks for applications that require very high throughput or extremely low latency is a severe computational challenge, further exacerbated by inefficiencies in mapping the computation to hardware. We present a novel method for designing neural network topologies that directly map to a highly efficient FPGA implementation. By exploiting the equivalence of artificial neurons with quantized inputs/outputs and truth tables, we can train quantized neural networks that can be directly converted to a netlist of truth tables, and subsequently deployed as a highly pipelinable, massively parallel FPGA circuit. However, the neural network topology requires careful consideration since the hardware cost of truth tables grows exponentially with neuron fan-in. To obtain smaller networks where the whole netlist can be placed-and-routed onto a single FPGA, we derive a fan-in driven hardware cost model to guide topology design, and combine high sparsity with low-bit activation quantization to limit the neuron fan-in. We evaluate our approach on two tasks with very high intrinsic throughput requirements in high-energy physics and network intrusion detection. We show that the combination of sparsity and low-bit activation quantization results in high-speed circuits with small logic depth and low LUT cost, demonstrating competitive accuracy with less than 15 ns of inference latency and throughput in the hundreds of millions of inferences per second.