在资源受限的FPGA上实现超长卷积，突破内存瓶颈。
Enabling Long FFT Convolutions on Memory-Constrained FPGAs via Chunking
- 将FFT卷积分块处理，结合重叠相加重建
- 在Alveo U200上实现45万长度的卷积运算
- 适合边缘设备部署长序列模型的开发者
长上下文推理需求推动了Transformer和自注意力之外的神经网络架构发展，其中Hyena通过使用FFT实现因果1D卷积成为热门选择。长卷积能高效混合全局上下文，但中间结果占用内存超过FPGA的2-3MB块存储容量。本文提出一种分块FFT卷积方法，通过分块与重叠相加重建，在Alveo U200 FPGA（2.8MB BRAM）上实现了45万长度序列与45万长度滤波器的卷积运算。结果显示吞吐量随分块大小线性增长，最长序列下性能仅下降7%，表明精心设计的内存管理可在不牺牲性能的前提下，将长上下文算子部署于边缘FPGA。
原文摘要 · Abstract (English)
The need for long-context reasoning has led to alternative neural network architectures besides Transformers and self-attention, a popular model being Hyena, which employs causal 1D-convolutions implemented with FFTs. Long convolutions enable efficient global context mixing, but requirements for intermediate results exceed the 2-3 MB Block RAM capacity of FPGAs. We present a chunked FFT convolution approach enabling 450K length sequence by 450K length filter convolutions on an Alveo U200 FPGA with 2.8 MB BRAM through chunking and overlap-add reconstruction. We find that throughput scales proportionally with chunk size while degrading minimally by 7% for our longest sequences, demonstrating that careful memory management enables deployment of long-context primitives on edge FPGAs without sacrificing performance.
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。