HAP: Hybrid-memory-Aware Partition in shared Last-Level Cache
HAP: Hybrid-memory-Aware Partition in shared Last-Level Cache
复制标题
DOI：
10.1109/iccd.2014.6974658
复制
发表时间：
2014-12
期刊：
影响因子：
--
通讯作者：
Wei Wei-Wei;D. Jiang;Jin Xiong;Mingyu Chen
中科院分区：
文献类型：
--
作者：
Wei Wei-Wei;D. Jiang;Jin Xiong;Mingyu Chen
数据中心服务器需要大容量的主内存来同时运行多个工作负载。然而，DRAM的可扩展性和功耗限制了其构建大容量存储器的能力。新兴的非易失性存储器（如PCM和STT-RAM）提供比DRAM更好的可扩展性和更低的功耗泄漏。特别是由DRAM和NVM组成的混合存储器能够发挥不同存储介质的优势。然而，nvm有一些缺点，比如相对较长的读写延迟。如果丢失的数据位于NVM中，则共享最后一级缓存（LLC）的缓存丢失会导致较长的延迟。当前的LLC策略管理缓存空间，而不知道底层的异构媒体。如果大量丢失的数据来自NVM，这将导致缓存性能下降。考虑到非对称缓存丢失成本，我们首先提出了一个新的性能指标-TMPKI，它可以准确地反映混合存储器上的LLC性能。然后，我们提出了一种基于TMPKI的混合内存感知缓存分区技术（HAP）来动态调整DRAM和NVM数据的缓存空间。实验结果表明，相对于传统的LRU策略，HAP提高了54.3%（平均19.6%）的性能，而它只增加了少量的存储开销（0.2%）。
Data-center servers require large capacity main memory to run multiple workloads simultaneously. However, the scalability and power consumption of DRAM limit its capability of constructing large capacity memory. Emerging non-volatile memories (e.g. PCM and STT-RAM) provide better scalability and lower power leakage than DRAM. Especially, hybrid memory consisting of DRAM and NVM is able to exploit advantages of different memory medias. However, NVMs have a few drawbacks, such as relatively longer read and write latency. Cache miss at the shared last level cache (LLC) suffers from longer latency if the missing data resides in NVM. Current LLC policies manage the cache space without being aware of the underlying heterogeneous medias. This results in cache performance degradation if a large number of missing data come from NVM. Taking the asymmetric cache miss cost into account, we first propose a new performance metric -TMPKI, which can exactly reflect the LLC performance on the top of hybrid memories. Then we propose a hybrid memory aware cache partitioning technique (HAP) to dynamically adjust the cache spaces for DRAM and NVM data based on TMPKI. Experimental results show that HAP improves performance against the traditional LRU policy by up to 54.3% (19.6% on average) while it incurs a little storage overhead (0.2%).