一句话：提出面向推荐模型嵌入表推理的 SSD 近数据处理内存系统，将端到端推理延迟降低 2 倍。
神经个性化推荐模型广泛用于搜索、社交媒体和娱乐等数据中心应用。当前先进模型包含数十亿参数的大规模嵌入表，需要大内存容量；而大容量快速 DRAM 成本高昂，传统 SSD 虽容量大一个数量级，但读延迟和带宽较差，会降低推理性能。RecSSD 是一种为神经推荐推理定制的近数据处理 SSD 内存系统，通过将计算靠近 SSD 数据来减少数据搬运并优化推理。作者在八个行业代表性模型上评估，相比使用商用现货 SSD，端到端模型推理延迟降低 2 倍。
Neural personalized recommendation models are used across a wide variety of datacenter applications including search, social media, and entertainment. State-of-the-art models comprise large embedding tables that have billions of parameters requiring large memory capacities. Unfortunately, large and fast DRAM-based memories levy high infrastructure costs. Conventional SSD-based storage solutions offer an order of magnitude larger capacity, but have worse read latency and bandwidth, degrading inference performance. RecSSD is a near data processing based SSD memory system customized for neural recommendation inference that reduces end-to-end model inference latency by 2× compared to using COTS SSDs across eight industry-representative models.
DOI 原文 ·
@inproceedings{paperbot6101,
title = {RecSSD: near data processing for solid state drive based recommendation inference},
author = {Mark Wilkening and Udit Gupta and Samuel Hsia and Caroline Trippel and Carole-Jean Wu and David Brooks and Gu-Yeon Wei},
booktitle = {ASPLOS},
year = {2021},
doi = {10.1145/3445814.3446763}
}