TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Exploiting Byte-Accessibility of SSDs Within a Unified Memory-Storage Hierarchy [pdf]

55 点作者 lainon将近 6 年前

1 comment

twotwotwo将近 6 年前
I <i>think</i> the sources of the claimed benefits here are 1) the OS can stay out of some reads entirely, because PCIe devices directly answering requests to read&#x2F;write memory is a thing, and 2) for small reads [or writes, as arcticbull notes below] you don&#x27;t have to send the whole page between the SSD and the host system, freeing up bandwidth.<p>The gains measured in the paper are using emulation, so, like, they didn&#x27;t really hack SSD firmware. There&#x27;s got to be some cost on the controller side to service these new requests. And they seem to be accounting for this, but of course taking away some overheads doesn&#x27;t take away the underlying limits of the storage medium or controller.<p>A couple things make something like this more interesting in the future: PCIe 4 doubles bandwidth and is supposed to be coming in some AMD chipsets soon (and the PCIe 5 spec has been finalized and, once it&#x27;s really shipping, provides another doubling), and of course Intel&#x2F;Micron&#x27;s new storage medium has zippier read&#x2F;writes than Flash.<p>I wonder whether arrangements like this over PCIe end up a middle ground between using swap&#x2F;mmap&#x27;ing vs. way pricier setups with nonvolatile DIMMs.
评论 #20070748 未加载
评论 #20070750 未加载