FPGA加速的SpeckleNN模型实现实时X射线衍射模式分类，功耗降低7.8倍。
FPGA-Accelerated SpeckleNN with SNL for Real-time X-ray Single-Particle Imaging
- 用FPGA部署优化后的SpeckleNN，参数量减少98.8%至64.6K
- 在KCU1500上实现45.015μs延迟，比GPU快8.9倍、省电7.8倍
- 支持动态权重加载，可快速部署新训练模型，适合实时自适应场景
我们基于SLAC神经网络库（SNL）在FPGA上实现了专用于实时X射线单粒子成像（SPI）中散斑图案分类的SpeckleNN模型。该硬件针对高通量X射线自由电子激光设施（如LCLS）的近探测器推理进行优化。为适配FPGA约束，我们对SpeckleNN进行了优化，参数量从560万降至6.46万（减少98.8%），准确率达90%，同时将潜在空间维度从128压缩至50。部署于KCU1500 FPGA时，使用了71%的DSP、75%的LUT和48%的FF，平均功耗为9.4W。在200 MHz下，推理延迟为45.015 μs。相比之下，相同推理在NVIDIA A100 GPU上耗电约73W，延迟为400 μs。FPGA版本相较GPU实现8.9倍速度提升和7.8倍功耗降低。关键创新包括模型专用化与通过SNL实现的动态权重加载，避免了频繁的FPGA重新综合，支持（重）训练模型的快速连续部署。这些改进使SpeckleNN适用于实时自适应分类与高效散斑模式剔除，显著提升XFEL实验效率并增强对动态变化环境的适应能力。
原文摘要 · Abstract (English)
We implement a specialized version of our SpeckleNN model for real-time speckle pattern classification in X-ray Single-Particle Imaging (SPI) using the SLAC Neural Network Library (SNL) on an FPGA. This hardware is optimized for inference near detectors in high-throughput X-ray free-electron laser (XFEL) facilities like the Linac Coherent Light Source (LCLS). To fit FPGA constraints, we optimized SpeckleNN, reducing parameters from 5.6M to 64.6K (98.8% reduction) with 90% accuracy. We also compressed the latent space from 128 to 50 dimensions. Deployed on a KCU1500 FPGA, the model used 71% of DSPs, 75% of LUTs, and 48% of FFs, with an average power consumption of 9.4W. The FPGA achieved 45.015us inference latency at 200 MHz. On an NVIDIA A100 GPU, the same inference consumed ~73W and had a 400us latency. Our FPGA version achieved an 8.9x speedup and 7.8x power reduction over the GPU. Key advancements include model specialization and dynamic weight loading through SNL, eliminating time-consuming FPGA re-synthesis for fast, continuous deployment of (re)trained models. These innovations enable real-time adaptive classification and efficient speckle pattern vetoing, making SpeckleNN ideal for XFEL facilities. This implementation accelerates SPI experiments and enhances adaptability to evolving conditions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。