arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.22293cs.ARcs.LGhep-ex
收藏

HGQ-LUT:用于DNN推理的快速LUT感知训练和高效架构

HGQ-LUT: Fast LUT-Aware Training and Efficient Architectures for DNN Inference

California Institute of Technology, Pasadena, CA, USA(加州理工学院);University of Bristol, UK(布里斯托大学);Imperial College London, UK(伦敦帝国理工学院);SLAC National Accelerator Laboratory, Menlo Park, CA, USA(SLAC国家加速器实验室)

Chang Sun, Zhiqiang Que, Bakhtiar Zadeh, Qibin Liu, Kevin H. Alvarez, Wayne Luk, Maria Spiropulu

AI总结:

HGQ-LUT通过引入LUT-Dense和LUT-Conv层及精细化量化,实现高效硬件效率和加速训练,支持混合架构的一体化设计与验证。

摘要:

基于查找表(LUT)的神经网络可通过直接映射算术运算到逻辑原语,在FPGA上实现超低延迟和优异的硬件效率。然而,最先进的LUT感知训练(LAT)方法在实践中仍面临诸多挑战:训练速度比传统网络慢多个数量级,需要复杂的手动调优以实现硬件效率,并缺乏端到端的工作流程。本文提出HGQ-LUT,集成在https://github.com/calad0i/HGQ2中,是一种新的LAT方法,实现了最先进的硬件效率,并在现代GPU上将训练速度提高了超过100倍。HGQ-LUT引入了LUT-Dense和LUT-Conv层,这些层在训练期间使用常规且加速器高效的张量操作实现,然后被编译为逻辑LUTs用于硬件。通过将这些层与精细化的元素级异构量化(包括零位剪枝)和LUT感知的资源代理相结合,HGQ-LUT能够在不进行手动位宽调优的情况下自动探索精度-资源权衡。我们进一步将HGQ-LUT集成到开源工具链中,使混合架构的一体化设计、编译和位精确验证成为可能,这些特性使基于LAT的DNN在实际部署中变得实用,例如在CERN大型强子对撞机的实验中。

英文摘要:

Lookup-table (LUT) based neural networks can deliver ultra-low latency and excellent hardware efficiency on FPGAs by mapping arithmetic operations directly onto the logic primitives. However, state-of-the-art LUT-aware training (LAT) approaches remain difficult to use in practice: they are often orders of magnitude slower to train than conventional networks, require non-trivial manual tuning for hardware efficiency, and lack an end-to-end workflow. This work presents HGQ-LUT, integrated in https://github.com/calad0i/HGQ2, a new LAT approach that achieves state-of-the-art hardware efficiency while accelerating training by over 100 times on modern GPUs. HGQ-LUT introduces LUT-Dense and LUT-Conv layers that are implemented with regular, accelerator-efficient tensor operations during training, which are then compiled into logic LUTs for hardware. By combining these layers with fine-grained, element-wise heterogeneous quantization (including zero-bit pruning) and a LUT-aware resource surrogate, HGQ-LUT enables the automatic exploration of accuracy-resource trade-offs without manual bit-width tuning. We further integrate HGQ-LUT into open-source toolchains, enabling unified design, compilation, and bit-exact verification of hybrid architectures that mix LUT-based with conventional arithmetic blocks. These features make LAT-based DNNs practical for real-world deployment, such as at the CERN Large Hadron Collider's experiments.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑