Abstract:
Flash memory in the last decade became dominant technology for storing data both for personal electronic devices and enterprise products: servers, network storages and data centers. Initially flash memory technology assumed storing only one data bit in a cell (SLC-memory). Further development of production technology of flash memory and inegration of more powerfull error correction codes in flash memory modules made it possible to store 2 bits (MLC-memory) and even 3-bits (TLC-memory) of data. Increase of number of bits stored in each cell leads to signifi cant increase of probability of error during read operation. Furthermore whith increase of programm-erase cycles electric charges stored in flash memory cells have tendency to decrease. This process also leads to increase of probability of error during read operation. Finally this process of charge decrease leads to flash memory failure when error correction code cannot fix all read errors. Rank method of storing data in a cells is resilient to process of graduated increase of cell charges. Moreover this method theoretiacally has higher area efficiency in comparison with conventional SLC/MLC/TLC memory. First in this paper brief description of conventional flash memory is given. Read operation is analyzed, basic model of read errors is described. Then rank method of storing data is presented. Estimated amount of memory stored by this method in comparison with conventional flash memory. Introduced Kendall-Tau distance between permutations. By means of this distance estimated area of rank memory taking in account technological limitations. Finally it's analyzed area efficiency of rank flash memory in comparison with conventional flash memory. Explicitly described cases when rank memory has higher area efficiency than conventional SLC/MLC/TLC memory.
Keywords:rank flash memory, SLC/MLC/TLC flash memory, flash memory area efficiency.
Fulltext:
PDF file (539 kB)