TileFuse: A Fused Mixed-Precision Kernel Library for Efficient Quantized LLM Inference on AMD NPUs

arXiv 2606.11357•8a8eb55bc2d651bd81b761444f1195057f25d441d53e879cf8a1095df4a6712b
AMD XDNA2AWQFP8ForeMoELLM hosting costsLLM inferenceMixture-of-ExpertsMoENPUsRL post-trainingRankGuardconcurrency-aware meteringdecentralized learningedge deploymentenergy efficiencyinfrastructure costload balancingmixed-precisionmodel poisoning defenseonline learning to rankperformance optimizationpoisoning attacksprocess mining on edge','edge computing','scheduling benchmarks'quantizationvllm-cost-meter

Paper metadata

arXiv ID
2606.11357
Version
Not specified by this published record
Category
Computer Science — Distributed, Parallel, and Cluster Computing (cs.DC)

The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.

Evidence and limitations

Source ID
arxiv_cs_dc
Record identifier
8a8eb55bc2d651bd81b761444f1195057f25d441d53e879cf8a1095df4a6712b
Enrichment time
2026-06-11T08:52:29Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.