Searcharxiv⌕ Search

arXiv subjects

Thomas B. Preußer

Publications and source records attributed to Thomas B. Preußer.

9 recordsLinked to original sources
arXiv2026

Arithmetic Packing on Wide Integer Datapaths in DSP Primitives of Modern FPGA Devices

Deep Neural Networks increasingly employ low-precision quantization to reduce computational requirements. While FPGAs are well suited for workloads with heterogeneous precisions, their dedicated digital signal processing (DSP) slices only feature fixed-width datapaths that are significantly underutilized by low-bitwidth arithmetic. While previous approaches have already introduced the packing of multiple values onto the same wide DSP datapath, they either only support specific fixed bitwidths or are wasteful regarding the use of additional support logic external to the DSP. This paper proposes an efficient method to dynamically pack multiple (un-)signed inputs with arbitrary bitwidths into a wide multiplier path by leveraging the DSP's internal pre-adder. Building on this, we present two distinct architectures, one optimized for matrix-vector multiplications and the other for convolutions. Our implementations are integrated into AMD's FINN framework. With these optimizations, we reduce the LUT utilization by 21% and increase the FPS/DSP by 36% for the UltraNet model compared to the FINN reference.

cs.AR↗
arXiv2025

FINN-GL: Generalized Mixed-Precision Extensions for FPGA-Accelerated LSTMs

Recurrent neural networks (RNNs), particularly LSTMs, are effective for time-series tasks like sentiment analysis and short-term stock prediction. However, their computational complexity poses challenges for real-time deployment in resource constrained environments. While FPGAs offer a promising platform for energy-efficient AI acceleration, existing tools mainly target feed-forward networks, and LSTM acceleration typically requires full custom implementation. In this paper, we address this gap by leveraging the open-source and extensible FINN framework to enable the generalized deployment of LSTMs on FPGAs. Specifically, we leverage the Scan operator from the Open Neural Network Exchange (ONNX) specification to model the recurrent nature of LSTM computations, enabling support for mixed quantisation within them and functional verification of LSTM-based models. Furthermore, we introduce custom transformations within the FINN compiler to map the quantised ONNX computation graph to hardware blocks from the HLS kernel library of the FINN compiler and Vitis HLS. We validate the proposed tool-flow by training a quantised ConvLSTM model for a mid-price stock prediction task using the widely used dataset and generating a corresponding hardware IP of the model using our flow, targeting the XCZU7EV device. We show that the generated quantised ConvLSTM accelerator through our flow achieves a balance between performance (latency) and resource consumption, while matching (or bettering) inference accuracy of state-of-the-art models with reduced precision. We believe that the generalisable nature of the proposed flow will pave the way for resource-efficient RNN accelerator designs on FPGAs.

cs.LG↗
arXiv2024

Shedding the Bits: Pushing the Boundaries of Quantization with Minifloats on FPGAs

Post-training quantization (PTQ) is a powerful technique for model compression, reducing the numerical precision in neural networks without additional training overhead. Recent works have investigated adopting 8-bit floating-point formats(FP8) in the context of PTQ for model inference. However, floating-point formats smaller than 8 bits and their relative comparison in terms of accuracy-hardware cost with integers remains unexplored on FPGAs. In this work, we present minifloats, which are reduced-precision floating-point formats capable of further reducing the memory footprint, latency, and energy cost of a model while approaching full-precision model accuracy. We implement a custom FPGA-based multiply-accumulate operator library and explore the vast design space, comparing minifloat and integer representations across 3 to 8 bits for both weights and activations. We also examine the applicability of various integerbased quantization techniques to minifloats. Our experiments show that minifloats offer a promising alternative for emerging workloads such as vision transformers.

cs.CV↗
arXiv2021

MicroRec: Efficient Recommendation Inference by Hardware and Data Structure Solutions

Deep neural networks are widely used in personalized recommendation systems. Unlike regular DNN inference workloads, recommendation inference is memory-bound due to the many random memory accesses needed to lookup the embedding tables. The inference is also heavily constrained in terms of latency because producing a recommendation for a user must be done in about tens of milliseconds. In this paper, we propose MicroRec, a high-performance inference engine for recommendation systems. MicroRec accelerates recommendation inference by (1) redesigning the data structures involved in the embeddings to reduce the number of lookups needed and (2) taking advantage of the availability of High-Bandwidth Memory (HBM) in FPGA accelerators to tackle the latency by enabling parallel lookups. We have implemented the resulting design on an FPGA board including the embedding lookup step as well as the complete inference process. Compared to the optimized CPU baseline (16 vCPU, AVX2-enabled), MicroRec achieves 13.8~14.7x speedup on embedding lookup alone and 2.5$~5.4x speedup for the entire recommendation inference in terms of throughput. As for latency, CPU-based engines needs milliseconds for inferring a recommendation while MicroRec only takes microseconds, a significant advantage in real-time recommendation systems.

cs.AR↗
arXiv2020

HyperLogLog Sketch Acceleration on FPGA

Data sketches are a set of widely used approximated data summarizing techniques. Their fundamental property is sub-linear memory complexity on the input cardinality, an important aspect when processing streams or data sets with a vast base domain (URLs, IP addresses, user IDs, etc.). Among the many data sketches available, HyperLogLog has become the reference for cardinality counting (how many distinct data items there are in a data set). Although it does not count every data item (to reduce memory consumption), it provides probabilistic guarantees on the result, and it is, thus, often used to analyze data streams. In this paper, we explore how to implement HyperLogLog on an FPGA to benefit from the parallelism available and the ability to process data streams coming from high-speed networks. Our multi-pipelined high-cardinality HyperLogLog implementation delivers 1.8x higher throughput than an optimized HyperLogLog running on a dual-socket Intel Xeon E5-2630 v3 system with a total of 16 cores and 32 hyper-threads.

cs.DC↗
arXiv2020

Using DSP Slices as Content-Addressable Update Queues

Content-Addressable Memory (CAM) is a powerful abstraction for building memory caches, routing tables and hazard detection logic. Without a native CAM structure available on FPGA devices, their functionality must be emulated using the structural primitives at hand. Such an emulation causes significant overhead in the consumption of the underlying resources, typically general-purpose fabric and on-chip block RAM (BRAM). This often motivates mitigating trade-offs, such as the reduction of the associativity of memory caches. This paper describes a technique to implement the hazard resolution in a memory update queue that hides the off-chip memory readout latency of read-modify-write cycles while guaranteeing the delivery of the full memory bandwidth. The innovative use of DSP slices allows them to assume and combine the functions of (a) the tag and data storage, (b) the tag matching, and (c) the data update in this key-value storage scenario. The proposed approach provides designers with extra flexibility by adding this resource type as another option to implement CAM.

cs.AR↗
arXiv2018

Inference of Quantized Neural Networks on Heterogeneous All-Programmable Devices

Neural networks have established as a generic and powerful means to approach challenging problems such as image classification, object detection or decision making. Their successful employment foots on an enormous demand of compute. The quantization of network parameters and the processed data has proven a valuable measure to reduce the challenges of network inference so effectively that the feasible scope of applications is expanded even into the embedded domain. This paper describes the making of a real-time object detection in a live video stream processed on an embedded all-programmable device. The presented case illustrates how the required processing is tamed and parallelized across both the CPU cores and the programmable logic and how the most suitable resources and powerful extensions, such as NEON vectorization, are leveraged for the individual processing steps. The crafted result is an extended Darknet framework implementing a fully integrated, end-to-end solution from video capture over object annotation to video output applying neural network inference at different quantization levels running at 16~frames per second on an embedded Zynq UltraScale+ (XCZU3EG) platform.

cs.NE↗
arXiv2018

Generic and Universal Parallel Matrix Summation with a Flexible Compression Goal for Xilinx FPGAs

Bit matrix compression is a highly relevant operation in computer arithmetic. Essentially being a multi-operand addition, it is the key operation behind fast multiplication and many higher-level operations such as multiply-accumulate, the computation of the dot product or the implementation of FIR filters. Compressor implementations have been constantly evolving for greater efficiency both in general and in the context of concrete applications or specific implementation technologies. This paper is building on this history and describes a generic implementation of a bit matrix compressor for Xilinx FPGAs, which does not require a generator tool. It contributes FPGA-oriented metrics for the evaluation of elementary parallel bit counters, a systematic analysis and partial decomposition of previously proposed counters and a fully implemented construction heuristic with a flexible compression target matching the device capabilities. The generic implementation is agnostic of the aspect ratio of the input matrix and can be used for multiplication the same way as it can be for single-column population count operations.

cs.AR↗