The platform exploits the reconfigurable logic of Field-Programmable Gate Arrays to implement approximate multipliers directly in hardware, eliminating the need for LUT-based emulation on CPU/GPU platforms and thereby enabling efficient DNN inference while significantly reducing evaluation time on large datasets.
Abstract
Approximate multipliers can reduce hardware area and energy consumption in Deep Neural Network (DNN) inference; however, they introduce computational errors. Assessing the accuracy of numerous approximate multiplier designs across diverse DNN models and large-scale datasets remains challenging due to prohibitive evaluation times. This overhead primarily stems from the slow emulation of approximate multiplier behavior using look-up tables (LUTs) on CPU and GPU platforms. Moreover, the resulting accuracy degradation must be carefully quantified and, if necessary, mitigated (e.g., through retraining), further increasing the overall evaluation cost. To address these challenges, we propose FAME, an FPGA-based platform for evaluating approximate multipliers. The platform exploits the reconfigurable logic of Field-Programmable Gate Arrays (FPGAs) to implement approximate multipliers directly in hardware, eliminating the need for LUT-based emulation on CPU/GPU platforms and thereby enabling efficient DNN inference while significantly reducing evaluation time on large datasets. Furthermore, we introduce a pattern-guided DNN retraining technique to mitigate accuracy degradation induced by approximate multipliers. Specifically, retraining is guided by multiplier-specific patterns to effectively recover potential accuracy losses. We evaluate FAME using two DNN models, ResNet-18 and MobileNetV2, on the ImageNet dataset across 27 approximate multipliers. During inference, our approach achieves up to a 3.47x speedup in approximate multiplier evaluation compared to prior LUT-based emulation methods. Furthermore, the proposed retraining technique improves accuracy by up to 65.5% over existing retraining approaches for the evaluated multipliers. The code is publicly available at: https://github.com/gicLAB/FAME
Deep Neural Networks (DNNs) are critical to modern AI applications, yet their deployment on standard CPUs and GPUs is constrained by high power consumption and computational latency, particularly in resource-constrained edge environments. To address these limitations, this paper presents the design and implementation o...
P. V. G. K. Rao, Dudekula Raziya· 2026 International Conferenc...· 0 citations
This survey restricted the scope to FPGAs alone and organized 25 compression-hardware co-design case studies into a five-category taxonomy defined by which FPGA resources each strategy primarily reshapes, formalizing six open challenges: toolchain fragmentation, accuracy--efficiency characterization, automated mixed-pr...
Peter Forcha, H. Kajekusumadhar, Mbua Peter et al.· 0 citations
This paper proposes a metrics-driven predictive methodology for FPGA-based DNN acceleration using an OpenCL-based high-level synthesis framework that validates how predictive modeling can help guide FPGA design choices in a way that improves both efficiency and portability for signal and image-processing-oriented DNN a...
Soufien Gdaim, Hachem Bensalem, Ghyslain Gagnon· Journal of Signal Processing...· 0 citations
Deploying Transformer models on FPGA and System-on-Chip (SoC) platforms remains challenging due to their substantial computational complexity, large memory footprint, and high hardware resource requirements, particularly in multi-head attention and stacked encoder-decoder layers. This paper proposes a hardware-efficien...
Xuan Thao Tran, Thi Diem Tran· International Conference on...· 0 citations
A systematic review of FPGA-based DL deployment from a cross-layer perspective spanning model, compiler, architecture, runtime, and electronic design automation (EDA) is presented, highlighting that reliable cross-study comparison requires careful consideration of model configuration, precision, execution phase, batch...
Shuo Wang, Lei Chen, Chunsheng Tian et al.· Electronics· 0 citations
This work analyzes the Vitis AI compiler and proposes an XIR-level splitting framework that generates independently compilable .xmodel fragments while preserving the context required for DPU mapping, and restores correct DPU mapping by addressing boundary-context loss and incomplete dependency collection.
Federico Buccellato, Luca Mannini, C. De Sio· WiPiEC Journal - Works in Pr...· 0 citations
Related blog posts
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026