5 papers
From Human Guidance to Autonomy: Agent Skill System for End-to-End LLM Deployment on Spatial NPUs
Jiajie Li, Erwei Wang, Zhiru Zhang +1
Spatial neural processing units (NPUs) provide an energy-efficient platform for edge LLM inference, but efficiently deploying an LLM end-to-end on such hardware remains labor-inten…
Fleet: Hierarchical Task-based Abstraction for Megakernels on Multi-Die GPUs
Sangeeta Chowdhary, Ryan Swann, Sean Siddens +7
Modern GPUs adopt chiplet-based designs with multiple private cache hierarchies, but current programming models (CUDA/HIP) expose a flat execution hierarchy that cannot express chi…
Power Aware Dynamic Reallocation For Inference
Yiwei Jiang, Sangeeta Chowdhary, Nathaniel Morris +3
Disaggregation has emerged as a powerful strategy for optimizing large language model (LLM) inference by separating compute-intensive prefill and memory-bound decode phases across…
From Loop Nests to Silicon: Mapping AI Workloads onto AMD NPUs with MLIR-AIR
Erwei Wang, Samuel Bayliss, Andra Bisca +19
General-purpose compilers abstract away parallelism, locality, and synchronization, limiting their effectiveness on modern spatial architectures. As modern computing architectures…
Efficiency, Expressivity, and Extensibility in a Close-to-Metal NPU Programming Interface
Erika Hunhoff, Joseph Melber, Kristof Denolf +8
Accelerators such as neural processing units (NPUs) deliver an enticing balance of performance and efficiency compared to general purpose compute architectures. However, effectivel…