一句话：提出 FAST 全栈搜索框架，联合搜索硬件数据通路、软件调度与编译优化，为特定深度学习工作负载定制加速器并提升性能功耗比。
深度学习模型快速演进，为面向数据中心规模特定工作负载定制推理加速器带来机会。论文提出 FAST（Full-stack Accelerator Search Technique），一个覆盖硬件-软件栈关键设计决策的加速器搜索框架，联合搜索硬件数据通路、软件调度以及算子融合、张量填充等编译 pass。作者分析 EfficientNet、BERT 等视觉与 NLP 模型的瓶颈，并用 FAST 生成针对性加速器。结果显示，面向单一工作负载优化的加速器在全部基准上 Perf/TDP 平均较 TPU-v3 提升 3.7 倍；面向一组工作负载服务的加速器平均提升 2.4 倍。投资回报分析表明，FAST 生成的加速器对中等规模数据中心部署具有潜在实用性。
The rapidly-changing deep learning landscape presents a unique opportunity for building inference accelerators optimized for specific datacenter-scale workloads. We propose Full-stack Accelerator Search Technique (FAST), a hardware accelerator search framework that defines a broad optimization environment covering key design decisions within the hardware-software stack, including hardware datapath, software scheduling, and compiler passes such as operation fusion and tensor padding. In this paper, we analyze bottlenecks in state-of-the-art vision and natural language processing (NLP) models, including EfficientNet and BERT, and use FAST to design accelerators capable of addressing these bottlenecks. FAST-generated accelerators optimized for single workloads improve Perf/TDP by 3.7× on average across all benchmarks compared to TPU-v3. A FAST-generated accelerator optimized for serving a suite of workloads improves Perf/TDP by 2.4× on average compared to TPU-v3. Our return on investment analysis shows that FAST-generated accelerators can potentially be practical for moderate-sized datacenter deployments.
DOI 原文 · arXiv · PDF（开放获取） · DBLP
@inproceedings{ZhangHSPLGM22,
title = {A full-stack search technique for domain optimized deep learning accelerators},
author = {Dan Zhang and Safeen Huda and Ebrahim Songhori and Kartik Prabhu and Quoc Le and Anna Goldie and Azalia Mirhoseini},
booktitle = {ASPLOS},
year = {2022},
doi = {10.1145/3503222.3507767}
}