Endurance-aware clustering-based mining algorithm for non-volatile phase-change memory
Ming-Chang Yang, Cheng-Chin Tu, Yuan-Hao Chang, Pei-Lun Suei, Tei‐Wei Kuo · 2014
The explosively growing amounts of data let many big data applications face the difficulty in maintaining all of the enormous runtime information in main memory. This paper considers a new memory architecture constructed by the emerging non-volatile memory (NVM) technologies, such as phase-change memory (PCM), to exploit the coexistent advantages for being main memory and secondary storage, so that the high demands of memory space can be overcome without sacrificing the efficiency for the big data applications. This paper chooses the clustering-based mining algorithms as the target applications and exploits the special asymmetric access patterns of the clustering-based mining strategies to further resolve the potential weak endurance problem of NVM. The experiments were conducted based on various datasets to evaluate the efficacy of the proposed scheme, and the results are very encouraging.