• Proceedings of the Materials Research Society of Korea Conference
• /
• 1998.08a
• /
• pp.97.3-97
• /
• 1998

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.15 no.5
• /
• pp.476-484
• /
• 2015

Due to the recent advances in non-volatile memory technologies such as PCM, a new memory hierarchy of computer systems is expected to appear. In this paper, we explore the performance of PCM-based swap systems and discuss how this system can be managed efficiently. Specifically, we introduce three management techniques. First, we show that the page fault handling time can be reduced by attaching PCM on DIMM slots, thereby eliminating the software stack overhead of block I/O and the context switch time. Second, we show that it is effective to reduce the page size and turn off the read-ahead option under the PCM swap system where the page fault handling time is sufficiently small. Third, we show that the performance is not degraded even with a small DRAM memory under a PCM swap device; this leads to the reduction of DRAM's energy consumption significantly compared to HDD-based swap systems. We expect that the result of this paper will lead to the transition of the legacy swap system structure of "large memory - slow swap" to a new paradigm of "small memory - fast swap."

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.11
• /
• pp.4246-4267
• /
• 2020

As the network goes deep into all aspects of people's lives, the number and the complexity of network traffic is increasing, and traffic classification becomes more and more important. How to classify them effectively is an important prerequisite for network management and planning, and ensuring network security. With the continuous development of deep learning, more and more traffic classification begins to use it as the main method, which achieves better results than traditional classification methods. In this paper, we provide a comprehensive review of network traffic classification based on deep learning. Firstly, we introduce the research background and progress of network traffic classification. Then, we summarize and compare traffic classification based on deep learning such as stack autoencoder, one-dimensional convolution neural network, two-dimensional convolution neural network, three-dimensional convolution neural network, long short-term memory network and Deep Belief Networks. In addition, we compare traffic classification based on deep learning with other methods such as based on port number, deep packets detection and machine learning. Finally, the future research directions of network traffic classification based on deep learning are prospected.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.05a
• /
• pp.237-240
• /
• 2003

Efficient memory management is one of the most crucial requirement of lava virtual machine. In Java, memory is allocated everytime when a new instance of class is created or when a method is called. The allocated memory is freed when the instance is no longer used, or when the called method is returned. In this paper we have examined the memory management scheme applied to the simpleRTJ lava virtual machine. The simpleRTJ has such a distinguished characteristic in its memory management scheme that the size of all instances are forced to be the same and the size of stack frames of all methods be the same, respectively. We present in this paper the scheme thoroughly and analyze its anticipated performance qualitatively.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2012.02a
• /
• pp.408-408
• /
• 2012

It is well known that magnetic random access memory (MRAM) is nonvolatile memory devices using ferromagnetic materials. MRAM has the merits such as fast access time, unlimited read/write endurance and nonvolatility. Although DRAM has many advantages containing high storage density, fast access time and low power consumption, it becomes volatile when the power is turned off. Owing to the attractive advantages of MRAM, MRAM is being spotlighted as an alternative device in the future. MRAM consists of magnetic tunnel junction (MTJ) stack and complementary metal- oxide semiconductor (CMOS). MTJ stacks are composed of various magnetic materials. FePt thin films are used as a pinned layer of MTJ stack. Up to date, an inductively coupled plasma reactive ion etching (ICPRIE) method of MTJ stacks showed better results in terms of etch rate and etch profile than any other methods such as ion milling, chemical assisted ion etching (CAIE), reactive ion etching (RIE). In order to improve etch profiles without redepositon, a better etching process of MTJ stack needs to be developed by using different etch gases and etch parameters. In this research, influences of $O_2$ gas on the etching characteristics of FePt thin films were investigated. FePt thin films were etched using ICPRIE in $CH_4/O_2/Ar$ gas mix. The etch rate and the etch selectivity were investigated in various $O_2$ concentrations. The etch profiles were studied in varying etch parameters such as coil rf power, dc-bias voltage, and gas pressure. TiN was employed as a hard mask. For observation etch profiles, field emission scanning electron microscopy (FESEM) was used.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2011.02a
• /
• pp.134-134
• /
• 2011

High-k dielectric materials such as $HfO_2$, $ZrO_2$ and $Al_2O_3$ increase gate capacitance and reduce gate leakage current in MOSFET structures. This behavior suggests that high-k materials will be promise candidates to substitute as a tunnel barrier. Furthermore, stack structure of low-k and high-k tunnel barrier named variable oxide thickness (VARIOT) is more efficient.[1] In this study, we fabricated the $WSi_2$ nanocrystals nonvolatile memory device with $SiO_2/HfO_2/Al_2O_3$ tunnel layer. The $WSi_2$ nano-floating gate capacitors were fabricated on p-type Si (100) wafers. After wafer cleaning, the phosphorus in-situ doped poly-Si layer with a thickness of 100 nm was deposited on isolated active region to confine source and drain. Then, on the gate region defined by using reactive ion etching, the barrier engineered multi-stack tunnel layers of $SiO_2/HfO_2/Al_2O_3$ (2 nm/1 nm/3 nm) were deposited the gate region on Si substrate by using atomic layer deposition. To fabricate $WSi_2$ nanocrystals, the ultrathin $WSi_2$ film with a thickness of 3-4 nm was deposited on the multi-stack tunnel layer by using direct current magnetron sputtering system [2]. Subsequently, the first post annealing process was carried out at $900^{\circ}C$ for 1 min by using rapid thermal annealing system in nitrogen gas ambient. The 15-nm-thick $SiO_2$ control layer was deposited by using ultra-high vacuum magnetron sputtering. For $SiO_2$ layer density, the second post annealing process was carried out at $900^{\circ}C$ for 30 seconds by using rapid thermal annealing system in nitrogen gas ambient. The aluminum gate electrodes of 200-nm thickness were formed by thermal evaporation. The electrical properties of devices were measured by using a HP 4156A precision semiconductor parameter analyzer with HP 41501A pulse generator, an Agillent 81104A 80MHz pulse/pattern generator and an Agillent E5250A low leakage switch mainframe. We will discuss the electrical properties for application next generation non-volatile memory device.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2010.06a
• /
• pp.358-358
• /
• 2010

To improve the programming/erasing speed and leakage current of multiple dielectric stack tunnel barrier engineering (TBE) Non-volatile memory, We propose a new concept called staggered structure of TBE memory. In this study, We fabricated staggered structure capacitor on $Si_3N_4$ stacked HfAlO and measured C-V curve that can observe tunneling characteristic of this device as various annealing temperature compared with that of single layer $SiO_2$ capacitor.

• Journal of the Korean Crystal Growth and Crystal Technology
• /
• v.12 no.3
• /
• pp.161-164
• /
• 2002

Polycrystalline and epitaxial heterostructure films of $La_{0.5}Sr_{0.5}CoO_{3}/Pb(Nb_{0.04}Zr_{0.28}Ti_{0.68})O_{3}/La_{0.5}Sr_{0.5}CoO_{3}$ (LSCO/PNZT/LSCO) capacitors were evaluated in terms of low voltage and high speed operation in high density memory, using TiN/Pt conducting barrier combination. Structural studies for a high density ferroelectric memory process flow, which requires the integration of conducting barrier layers to connect the drain of the pass-gate transistor to the bottom electrode of the ferroelectric stack, indicate complete phase purity (i.e. fully perovskite) in both epitaxial and polycrystalline materials. The polycrystalline capacitors show lower remnant polarization and coercive voltages. However, the retention, and high-speed characteristics are similar, indicating minimal influence of crystalline quality on the ferroelectric properties.

• KIISE Transactions on Computing Practices
• /
• v.22 no.10
• /
• pp.497-503
• /
• 2016

Emerging next-generation storage technologies such as non-volatile memory will help eliminate almost all of the storage latency that has plagued previous storage devices. In conventional storage systems, the latency of slow storage devices dominates access latency; hence, software efficiency is not critical. With low-latency storage, software costs can quickly dominate memory latency. Hence, researchers have proposed the memory mapped file I/O to avoid the software overhead. Mapping a file into the user memory space enables users to access the file directly. Therefore, it is possible to avoid the complicated I/O stack. This minimizes the number of user/kernel mode switchings. In addition, there is no data copy between kernel and user areas. Despite of the benefits in the memory mapped file I/O, its overhead still needs to be addressed, as the existing mechanism for the memory mapped file I/O is designed for slow block devices. In this paper, we identify the overheads of the memory mapped file I/O via experiments.

• The Proceeding of the Korean Institute of Electromagnetic Engineering and Science
• /
• v.24 no.2
• /
• pp.51-57
• /
• 2013

The current expanding mobile markets incessantly demands small form factor, low power consumption and high aggregate throughput for silicon-level integration such as memory to logic system. One of emerging solution for meeting this high market demand is 3D through silicon stacking (TSS) technology. Main challenges to bring 3D TSS technology to the volume production level are establishing a cost effective supply chain and building a reliable manufacturing processes. In addition, this technology inherently help increase number of IOs and shorten interconnect length. With those benefits, however, potential signal and power integrity risks are also elevated; increase in PDN inductance, channel loss on substrate, crosstalk and parasitic capacitance. This paper will report recent progress of wide IO memory to high count TSV logic device assembly development work. 28 nm node TSV test vehicles were fabricated by the foundry and assembled. Successful integration of memory wide IO chip with less than a millimeter package thickness form factor was achieved. For this successful integration, we discussed potential signal and power integrity challenges. This report demonstrated functional wide IO memory to 28 nm logic device assembly using 3D package architecture with such a thin form factor.