发表机构
Indian Institute of Technology Madras（印度理工学院马德拉斯分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
NeuroFlex提出首个元素级ANN-SNN协同执行加速器，通过扩展整数精确等价性实现零精度损失的模式切换，结合成本引导调度器，将PE利用率提升至97-99%，EDP降低57-67%，吞吐量提高16-19%。
AI 中文摘要
稀疏DNN加速器专门针对ANN或SNN执行进行优化，当层内工作负载特征变化时，会留下能量或延迟开销。在层或块粒度切换模式的混合加速器设计存在PE利用率低的问题，因为当一种核心类型激活时，另一种核心类型空闲。NeuroFlex是首个将每个输出元素独立分配给ANN或SNN执行模式且零精度损失的加速器。我们将整数精确的ANN-SNN等价性从层扩展到单个输出元素，从而实现无转换误差的模式切换。一个离线成本引导调度器根据每个元素的边际能量-延迟权衡对其进行评分，并在PE间打包工作，实现了97-99%的PE利用率，而层级混合仅为40-45%。NeuroFlex相比强ANN-only基线将EDP降低了57-67%，相比双稀疏SNN-only基线提供了高达2.5倍的加速。我们的成本引导调度器在视觉、语言和Transformer工作负载上相比随机元素分配将吞吐量提高了16-19%。
英文摘要
Sparse DNN accelerators specialize in ANN or SNN execution, leaving energy or latency on the table when workload characteristics vary within a layer. Hybrid accelerator designs that switch modes at layer or tile granularity suffer from low PE utilization since one core type idles whenever the other is active. NeuroFlex is the first accelerator to assign every output element independently to ANN or SNN execution mode with zero accuracy loss. We extend integer-exact ANN-SNN equivalence from layers to individual output elements, thereby enabling mode switching with no conversion error. An offline cost-guided scheduler scores each element by its marginal energy-delay trade-off and packs work across PEs, achieving 97-99% PE utilization compared to 40-45% for layer-wise hybrids. NeuroFlex reduces EDP by 57-67% over a strong ANN-only baseline and delivers up to 2.5x speedup over a dual-sparse SNN-only baseline. Our cost-guided scheduler improves throughput by 16-19% over random element assignment across vision, language, and transformer workloads.
CommentsThis entry duplicates arXiv:2511.05215 and was submitted as a new article in error instead of as a replacement. The updated manuscript is now posted as a version of arXiv:2511.05215, which is the canonical record. Please cite arXiv:2511.05215; two identifiers for one paper split its citations and version history
