Research Article
Using Emulation Software to Predict the Performance of Algorithms on NVRAM
@INPROCEEDINGS{10.4108/icst.simutools.2014.254796, author={Jana Traue and J\o{}rg Nolte and Philipp Engel and Reinhardt Karnapke}, title={Using Emulation Software to Predict the Performance of Algorithms on NVRAM}, proceedings={Seventh International Conference on Simulation Tools and Techniques}, publisher={ICST}, proceedings_a={SIMUTOOLS}, year={2014}, month={8}, keywords={emulation non-volatile memory}, doi={10.4108/icst.simutools.2014.254796} }- Jana Traue
Jörg Nolte
Philipp Engel
Reinhardt Karnapke
Year: 2014
Using Emulation Software to Predict the Performance of Algorithms on NVRAM
SIMUTOOLS
ICST
DOI: 10.4108/icst.simutools.2014.254796
Abstract
Currently, new storage technologies which unite the latency and byte-addressability of DRAM with the persistence of disks are being developed. This non-volatile memory (NV-RAM) may start a software revolution. Traditionally, software was developed for two levels of storage and NVRAM reduces the hierarchy to a single-level store. Current research projects are already exploring the potential of NVRAM, but they face a challenge when they want to evaluate the performance: The new hardware is not yet available. In this paper, we discuss why benchmark results which are gained on existing DRAM are insufficient for a prediction of the performance on NVRAM. Either existing instructions have to be changed or new ones have to be introduced. We further show that the bochs emulator can be used to build systems which resemble NVRAM, to predict the NVRAM’s consequences, and it even allows a comparison of algorithms for NVRAM.