arXiv subjects

Hang Gu

Publications and source records attributed to Hang Gu.

6 recordsLinked to original sources

arXiv2026

UnionSparse: An Index-Efficient Sparsity Framework for Low-Bit Sparse LLM Inference on Edge

Edge LLM inference combines sparsity and low-bit quantization to meet device memory, latency, and power limits. Yet quantization shrinks weight payloads without proportionally reducing sparse metadata, so index traffic and nonzero extraction become critical SpMM bottlenecks. We introduce the Payload-to-Metadata Ratio (PMR) and show that improving PMR raises effective compute intensity in decoding. We present UnionSparse, an index-efficient framework that combines Index-Efficient Bitmap Encoding (IE-BME) with a SpMM kernel using Low-Bit Shared-Memory Parallel Decoding (LSPD). IE-BME amortizes metadata and aligns sparse traversal with fragment assembly, while LSPD improves small-batch execution. Under W4A4 quantization and 30%--70% sparsity, UnionSparse outperforms FlashLLM and SpInfer by 2.30x and 1.43x, and CUTLASS and cuBLAS Tensor Core by 1.56x and 3.46x, respectively. These results establish payload-extraction efficiency as a first-order concern for low-bit sparse inference on edge GPUs. Source code is available at: https://github.com/Victor-Alen/UnionSparse.

cs.DC↗

arXiv2026

Accelerating OpenPangu Inference on NPU via Speculative Decoding

To mitigate the Memory Wall bottleneck encountered by Large Language Models (LLMs) during inference on \textbf{NPU} hardware, and addressing the scarcity of native support for mainstream speculative decoding algorithms on domestic infrastructure, this study presents an end-to-end speculative inference acceleration scheme for OpenPangu-7B.

cs.DC↗

arXiv2026

Hermes: A Unified High-Performance NTT Architecture with Hybrid Dataflow

Fully Homomorphic Encryption (FHE) relies heavily on the Number Theoretic Transform (NTT), making NTT a major performance bottleneck due to its intensive polynomial computations. Hybrid Homomorphic Encryption (HHE), which integrates arithmetic and logic FHE, further requires support for multiple NTT lengths. However, existing accelerators mainly optimize NTT throughput and do not provide unified support for HHE. This paper presents Hermes, a unified high-performance NTT architecture based on hybrid dataflow. Hermes exploits parallelism along both temporal and spatial dimensions and incorporates a fully pipelined on-chip computing core. A conflict-free on-chip fragmentation algorithm is introduced to resolve bank conflicts and enable burst HBM access, while an efficient dataflow improves computational intensity through data reuse, reducing bandwidth demand. Experimental results show that Hermes supports multiple NTT lengths and achieves up to 13.6x and 1.3x higher throughput than state-of-the-art GPU and FPGA accelerators, respectively. Our source code is available at https://anonymous.4open.science/r/Hermes_conf-4E6F.

cs.AR↗

arXiv2025

ActionFlow: A Pipelined Action Acceleration for Vision Language Models on Edge

Vision-Language-Action (VLA) models have emerged as a unified paradigm for robotic perception and control, enabling emergent generalization and long-horizon task execution. However, their deployment in dynamic, real-world environments is severely hin dered by high inference latency. While smooth robotic interaction requires control frequencies of 20 to 30 Hz, current VLA models typi cally operate at only 3-5 Hz on edge devices due to the memory bound nature of autoregressive decoding. Existing optimizations often require extensive retraining or compromise model accuracy. To bridge this gap, we introduce ActionFlow, a system-level inference framework tailored for resource-constrained edge plat forms. At the core of ActionFlow is a Cross-Request Pipelin ing strategy, a novel scheduler that redefines VLA inference as a macro-pipeline of micro-requests. The strategy intelligently batches memory-bound Decode phases with compute-bound Prefill phases across continuous time steps to maximize hardware utilization. Furthermore, to support this scheduling, we propose a Cross Request State Packed Forward operator and a Unified KV Ring Buffer, which fuse fragmented memory operations into efficient dense computations. Experimental results demonstrate that ActionFlow achieves a 2.55x improvement in FPS on the OpenVLA-7B model without retraining, enabling real-time dy namic manipulation on edge hardware. Our work is available at https://anonymous.4open.science/r/ActionFlow-1D47.

cs.AI↗

arXiv2012

Crossing on hyperbolic lattices

We divide the circular boundary of a hyperbolic lattice into four equal intervals, and study the probability of a percolation crossing between an opposite pair, as a function of the bond occupation probability p. We consider the {7,3} (heptagonal), enhanced or extended binary tree (EBT), the EBT-dual, and {5,5} (pentagonal) lattices. We find that the crossing probability increases gradually from zero to one as p increases from the lower p_l to the upper p_u critical values. We find bounds and estimates for the values of p_ l and p_u for these lattices, and identify the self-duality point p* corresponding to where the crossing probability equals 1/2. Comparison is made with recent numerical and theoretical results.

cond-mat.dis-nn↗

arXiv2009

Universal condition for critical percolation thresholds of kagome-like lattices

Lattices that can be represented in a kagome-like form are shown to satisfy a universal percolation criticality condition, expressed as a relation between P_3, the probability that all three vertices in the triangle connect, and P_0, the probability that none connect. A linear approximation for P_3(P_0) is derived and appears to provide a rigorous upper bound for critical thresholds. A numerically determined relation for P_3(P_0) gives thresholds for the kagome, site-bond honeycomb, (3-12^2), and "stack-of-triangle" lattices that compare favorably with numerical results.

cond-mat.dis-nn↗