一句话：揭示 I/O 设备经单一 CPU 发起的非均匀 DMA（NUDMA）会像 NUMA 一样拖慢应用性能。
在多 CPU 服务器中，内存模块与所连 CPU 形成非均匀内存访问（NUMA）架构；由于非本地访问比本地访问慢，NUMA 会降低应用性能。摘要指出，当 I/O 设备也通过单一 CPU 连接内存并发出非均匀 DMA（NUDMA）操作时，会出现类似的访问延迟不均衡，从而以类似 NUMA 的方式损害应用性能。给出的摘要只描述了 NUDMA 问题及其与 NUMA 的类比，未说明 IOctopus 采用的具体方法、实现或评估结果。
In a multi-CPU server, memory modules are local to the CPU to which they are connected, forming a nonuniform memory access (NUMA) architecture. Because non-local accesses are slower than local accesses, the NUMA architecture might degrade application performance. Similar slowdowns occur when an I/O device issues nonuniform DMA (NUDMA) operations, as the device is connected to memory via a single CPU. NUDMA effects therefore degrade application performance similarly to NUMA effects.
DOI 原文 ·
@inproceedings{paperbot5464,
title = {IOctopus},
author = {Igor Smolyar and Alex Markuze and Boris Pismenny and Haggai Eran and Gerd Zellweger and Austin Bolen and Liran Liss and Adam Morrison and Dan Tsafrir},
booktitle = {ASPLOS},
year = {2020},
doi = {10.1145/3373376.3378509}
}