llm-kb
← Back to research
Paper

STEEL: Sparsity-Aware Fused Attention for Energy-Efficient Long-Sequence Inference on AMD's XDNA NPU

A new paper introduces STEEL, a sparsity-aware fused attention mechanism designed for AMD's XDNA NPU, enabling energy-efficient long-sequence inference on laptop-class SoCs. The approach addresses the challenge of mapping attention mechanisms onto NPUs while maintaining low power consumption, which is critical for agentic workloads that require on-device processing for reliability and privacy.

0 engagement·1 source·Fri, Jul 10, 2026, 01:09 PM
The paper, titled 'STEEL: Sparsity-Aware Fused Attention for Energy-Efficient Long-Sequence Inference on AMD's XDNA NPU,' was published on arXiv on July 10, 2026. It highlights the growing need for energy-efficient inference in laptop-class systems-on-chip (SoCs) due to the adoption of LLM-based agents in OS workflows. Cloud offloading, while common, introduces reliability and privacy concerns for agentic workloads. The proposed STEEL mechanism leverages sparsity to fuse attention operations, optimizing for AMD's XDNA NPU architecture. This work is significant for practitioners seeking to deploy LLM agents on local hardware without sacrificing performance or energy efficiency.

Entities

STEEL(tool)AMD(company)XDNA NPU(tool)sparsity-aware fused attention(concept)

Related

PaperFri, Jul 10, 2026, 04:02 AM

Survey on Green Development of Large Models: Resource-Efficient Architectures and Hardware-Software Co-Design

A comprehensive survey published on arXiv reviews strategies for reducing computational costs and energy consumption of large AI models, covering efficient architectures (attention optimization, linear-complexity models, sparsification) and full-stack hardware-software co-design. The paper provides a systematic overview of recent advances in green AI development.

0 engagement·1 source·arxiv
arXiv
PaperThu, Jul 9, 2026, 11:40 PM

SATS: Sensitivity-Aware Thresholding for MLP Activation Sparsification in LLMs

A new arxiv paper proposes Sensitivity-Aware Thresholding for Sparsity (SATS), a method to calibrate layerwise gate thresholds for MLP activation sparsification using a local sensitivity proxy instead of activation percentiles. The approach aims to reduce computation during LLM inference while preserving model quality.

0 engagement·1 source·arxiv
arXiv
PaperFri, Jul 10, 2026, 08:04 AM

Study evaluates energy, performance, and accuracy trade-offs across vLLM configurations

A new arxiv paper presents a large-scale controlled study of three vLLM configuration options—attention kernel type, prefix caching, and chunked prefill—examining their impact on energy consumption, performance, and output quality. The work addresses a gap in understanding how inference engine configuration affects these trade-offs in production LLM deployments.

0 engagement·1 source·arxiv
arXiv
PaperFri, Jul 10, 2026, 03:31 PM

Study reveals text generation, not vision, is the energy bottleneck in edge VLM inference

A systematic energy profiling study of on-device Vision-Language Models (VLMs) across five models, four resolutions, and two hardware platforms (NVIDIA RTX 3070 and Jetson Orin NX) overturns the common assumption that visual processing dominates energy cost. The authors find that text generation is the true bottleneck, accounting for the majority of energy consumption during inference on edge devices.

0 engagement·1 source·arxiv
arXiv
PaperSun, Jul 12, 2026, 08:46 PM

Flash-MSA: Sparse Attention Kernels Enable Million-Token Training

A new paper introduces Flash-MSA, a sparse attention kernel that reduces memory and computation for long-context transformers, enabling training on sequences up to one million tokens. The method achieves up to 8x speedup over FlashAttention-2 on 128K-length sequences while maintaining model quality.

10 engagement·3 sources·hackernews, rss, arxiv