paperbot · PL 论文追踪

全部论文 待读 收藏

P2 Hardware-Software Co-design to Mitigate DRAM Refresh Overheads

硬件-软件协同设计以缓解 DRAM 刷新开销
ASPLOS Proceedings of the Twenty-Second International Conference on Architectural Support for Programming Languages and Operating Systems2017匹配度 0/10 · 论文研究DRAM刷新开销的硬件-软件协同优化，属于计算机体系结构方向，与用户关注的程序语言、静态分析、LLM/Agent、浮点程序分析及贝叶斯程序分析均无直接关联。引用 33 inbox
Jagadish B. Kotra, Narges Shahidi, Zeshan A. Chishti, Mahmut T. Kandemir
语言与体系结构
crossref: 2026-09-12dblp: 2026-09-12

AI 速览

一句话：提出硬件-软件协同设计以缓解 DRAM 刷新带来的性能开销。

DRAM 单元需周期性刷新以保持数据完整性；容量增大使每次刷新命令要刷新的行数增多，刷新周期时间 tRFC 增加，刷新成为显著性能瓶颈。现代 DRAM 通常在 rank 级刷新，移动 LPDDR 支持按 bank 刷新：rank 级刷新会阻塞同一 rank 内所有 bank 的按需请求，按 bank 刷新则允许其他 bank 继续服务。摘要指出典型保持时间为数毫秒，低于 85°C 时 64ms，高于 85°C 时 32ms。标题表明论文采用硬件-软件协同设计缓解这些刷新开销，但摘要未给出具体方法、实验设置或量化结果。

问题
DRAM 周期性刷新带来性能瓶颈；rank 级刷新会阻塞整个 rank 的按需请求，高容量 DRAM 下 tRFC 增大加剧该问题。
方法
摘要未展开具体方法；从标题看，论文采用硬件-软件协同设计来降低或隐藏 DRAM 刷新开销。
评估
摘要未报告评估方法或实验结果；仅给出 DRAM 保持时间等背景参数。
原文摘要（Abstract）

DRAM cells need periodic refresh to maintain data integrity. With high capacity DRAMs, DRAM refresh poses a significant performance bottleneck as the number of rows to be refreshed (and hence the refresh cycle time, tRFC) with each refresh command increases. Modern day DRAMs perform refresh at a rank-level, while LPDDRs used in mobile environments support refresh at a per-bank level. Rank-level refresh degrades the performance significantly since none of the banks in a rank can serve the on-demand requests. Per-bank refresh alleviates some of the performance bottlenecks as the other banks in a rank are available for on-demand requests. Typical DRAM retention time is in the order several of milliseconds, viz, 64msec for environments operating in temperatures below 85 deg C and 32msec for environments operating above 85 deg C.

链接与引用

DOI 原文 · DBLP

BibTeX 
@inproceedings{KotraSCK17,
  title = {Hardware-Software Co-design to Mitigate DRAM Refresh Overheads},
  author = {Jagadish B. Kotra and Narges Shahidi and Zeshan A. Chishti and Mahmut T. Kandemir},
  booktitle = {ASPLOS},
  year = {2017},
  doi = {10.1145/3037697.3037724}
}