NAND Flash Memory With Multiple Page Sizes for High-Performance Storage Devices

Kim, Jin‐Young; Park, Sang Hoon; Seo, Hyeokjun; Song, Ki–Whan; Yoon, Sungroh; Chung, Eui-Young

doi:10.1109/tvlsi.2015.2409055

Cited by 8 publications

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In cases where the size of a single example exceeds the page size, frequent data fragmentation is inevitable, which will eventually affect the overall performance. The effectiveness of using multiple page sizes has already been reported for conventional SSDs [42]. We may borrow this idea to further optimize ISP-ML.…”

Section: Opportunities For Future Researchmentioning

confidence: 95%

Near-Data Processing for Differentiable Machine Learning Models

Choe,

Lee,

Nam

et al. 2016

Preprint

Self Cite

View full text Add to dashboard Cite

Near-data processing (NDP) refers to augmenting memory or storage with processing power. Despite its potential for acceleration computing and reducing power requirements, only limited progress has been made in popularizing NDP for various reasons. Recently, two major changes have occurred that have ignited renewed interest and caused a resurgence of NDP. The first is the success of machine learning (ML), which often demands a great deal of computation for training, requiring frequent transfers of big data. The second is the popularity of NAND flash-based solid-state drives (SSDs) containing multicore processors that can accommodate extra computation for data processing. In this paper, we evaluate the potential of NDP for ML using a new SSD platform that allows us to simulate instorage processing (ISP) of ML workloads. Our platform (named ISP-ML) is a full-fledged simulator of a realistic multi-channel SSD that can execute various ML algorithms using data stored in the SSD. To conduct a thorough performance analysis and an in-depth comparison with alternative techniques, we focus on a specific algorithm: stochastic gradient descent (SGD), which is the de facto standard for training differentiable models such as logistic regression and neural networks. We implement and compare three SGD variants (synchronous, Downpour, and elastic averaging) using ISP-ML, exploiting the multiple NAND channels to parallelize SGD. In addition, we compare the performance of ISP and that of conventional in-host processing, revealing the advantages of ISP. Based on the advantages and limitations identified through our experiments, we further discuss directions for future research on ISP for accelerating ML.

show abstract

Section: Opportunities For Future Researchmentioning

confidence: 95%