Nov 2026· IEEE Transactions on Circuits and Systems Part 1: Regular Papers· pp. 1-14· 0 citations· 33 references
Computer Science
TL;DR
SparseForge is presented, a hardware-aware sparse SNN acceleration framework that jointly optimizes neuron pruning, sparse dataflow, and accelerator microarchitecture and develops a mask-aware sparse processing unit that supports compressed spike decoding, sparse implicit im2col conversion, neuron-mask-gated psum updates, and decoupled sparse execution.
Abstract
Spiking Neural Networks (SNNs) provide an event-driven paradigm for energy-efficient edge computing. However, operation-level sparsity in SNNs does not necessarily eliminate the hardware costs associated with redundant output neurons, including neuron-state maintenance, partial-sum (psum) accesses, and on-chip storage requirements. This paper presents SparseForge, a hardware-aware sparse SNN acceleration framework that jointly optimizes neuron pruning, sparse dataflow, and accelerator microarchitecture. At the algorithm level, we introduce a gated leaky integrate-and-fire (G-LIF) neuron to learn differentiable neuron masks with binary spike semantics. A dual regularization scheme is further developed to align pruning with both synaptic-operation reduction and psum buffer capacity reduction. At the dataflow level, we design a sparse Column-Wise Product (CWP) dataflow with a variant of incremental spike encoding to improve weight-access regularity and reduce sparse-index overhead under dynamic spike activity. At the architecture level, we develop a mask-aware sparse processing unit (SPU) that supports compressed spike decoding, sparse implicit im2col conversion, neuron-mask-gated psum updates, and decoupled sparse execution. Experimental results across multiple SNN backbones and datasets show that SparseForge retains 1.75%--77.99% of neurons and 0.16%--62.36% of synaptic operations, with a configurable accuracy--efficiency trade-off. The buffer-capacity-aware regularization reduces the required psum buffer capacity by up to 96.18\%. FPGA implementations on a Xilinx ZCU9EG achieve 296.94--619.49 GOPS/W across the evaluated workloads. Across matched network and dataset settings, SparseForge provides 1.24--4.17 higher FPS and 1.50--7.18 higher GOPS/W than representative FPGA-based SNN accelerators, while reducing EDP by 1.70--19.13, with benchmark-dependent accuracy--efficiency trade-offs.
While Spiking Neural Networks (SNNs) offer a promising path toward energy-efficient edge intelligence, conventional Binary representations often suffer from high inference latency and discretization errors. Multi-level (graded) spike models address these limitations by increasing information density per pulse, yet they...
Wen-Fei Song, A. Castagnetti, Pierre-Emmanuel Novac et al.· Neuromorphic Computing and E...· 0 citations
Spiking neural networks (SNNs) are promising for energy-efficient artificial intelligence due to their event-driven sparse computation, and neuron pruning can further reduce redundant operations. However, in practical hardware accelerators, spike sparsity and pruning-induced sparsity produce highly irregular sparse wor...
San-Hong Fu, Kai-Nang Wang, Bo Wang et al.· International Conference on...· 0 citations
The increasing scale of spiking neural networks (SNNs) poses significant challenges for deployment on resource-constrained neuromorphic hardware, necessitating lightweight and learnable structural solutions. Interestingly, biological neural systems employ an efficient organizational strategy—hierarchical structural reo...
Jiang-Rong Shen, Shu-Xian Dong, Qi Xu et al.· Proceedings of the 34th ACM...· 0 citations
Energy-constrained aerial platforms require accurate object detection under strict power and latency budgets, yet conventional inference remains bottlenecked by multiply-accumulate (MAC) operations. While spiking neural networks (SNNs) offer bio-inspired efficiency, many SNN detectors retain hybrid MACs in intermediate...
Deep spiking neural networks (SNNs) have gained significant attention in the current landscape of neural computing, primarily due to their remarkable energy efficiency and capability to perform complex computational tasks. This article presents the hardware implementation of a residual SNN (RSNN) architecture featuring...
Saras Mani Mishra, Souraja Kundu, Naorem Yaipharenba Meitei et al.· IEEE Transactions on Very La...· 0 citations
Deep neural networks continue to grow in parameter count, driving up training and inference cost on GPUs. Sparse neural networks and Dynamic Sparse Training (DST) promise to reduce these costs, but most implementations rely on binary masks over dense tensors and recover little of the theoretical compute, memory, or ene...
Jafar Badour, Maurice van Keulen, E. Mocanu· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.