Metraipolis › archive › dc_32e956490e6b
FlashQuant: Sparse-Dense Fusion for Memory-Efficient Outlier-Aware LLM Inference — arXiv:2608.15531
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.15531
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T19:02:06.874Z
- SHA-256 of the archived text
f8e63566ded94187d9e271e4f6b232c283bcd2355023dd3b531a41e65af3d651- Stable URL of this page
https://metraipolis.com/documento/dc_32e956490e6b
How to cite
FlashQuant: Sparse-Dense Fusion for Memory-Efficient Outlier-Aware LLM Inference — arXiv:2608.15531 — arXiv (en), https://arxiv.org/abs/2608.15531 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_32e956490e6b (Metraipolis archive dc_32e956490e6b, sha256 f8e63566ded94187d9e271e4f6b232c283bcd2355023dd3b531a41e65af3d651)
BibTeX
@misc{dc32e956490e6b,
title = {FlashQuant: Sparse-Dense Fusion for Memory-Efficient Outlier-Aware LLM Inference — arXiv:2608.15531},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.15531},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_32e956490e6b; sha256 f8e63566ded94187d9e271e4f6b232c283bcd2355023dd3b531a41e65af3d651},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15531. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Low-bit quantization reduces the memory footprint and computational cost of large language model (LLM) inference. However, high-magnitude outlier weights can induce substantial quantization errors and degrade model accuracy. Outlier-aware quantization addresses this issue by retaining outliers in high precision while quantizing the remaining weights, resulting in a low-bit dense GEMM path and a high-precision sparse SpMM path. Existing implementations execute these paths in separate GPU kernels, despite their shared activations and outputs, thereby missing opportunities for intra-operator reuse and incurring redundant global-memory accesses. This inefficiency is particularly pronounced in memory-bound decoding workloads. We propose FlashQuant, a content-sharing execution framework for outlier-aware W4A16 decoding. FlashQuant fuses the dense GEMM and sparse outlier SpMM paths into a single GPU kernel, enabling on-chip reuse of activation and output tiles across heterogeneous computations. It introduces three key techniques: sparse-dense tiling, which aligns outlier processing with dense GEMM tiles; Tile-COO outlier encoding, which enables efficient sparse access and reduces shared-memory bank conflicts; and pipelined scheduling, which overlaps computation with data movement. Experiments show that FlashQuant reduces outlier-processing overhead, achieving $2.74\times - 4.18\times$ speedup over cuBLAS BF16 and up to $1.53\times$ speedup over the strongest unfused outlier-aware baseline. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-08-18 in cs.DC). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC). Pagina del lavoro: https://arxiv.org/abs/2608.15531 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T18:07:16.070Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).