• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.6B
• /
• pp.325-333
• /
• 2007

An OS requires the device driver to control hardware IPs at application level. Development of a device driver requires specific acknowledge for target hardware and OS. In this paper, we present a system which generates a device driver together with an interface circuit. In the proposed system, an efficient device driver is generated by selecting a basic device driver skeleton, a function module code, and a header file table from the pre-constructed library and an interface circuit is constructed such that the generated device driver operates correctly. The proposed system is evaluated by generating a TFT-LCD device driver on the ARM922T core with 3.5 inch Samsung TFT-LCD in ARM-Linux environment. Experiment result shows that the writing time on the LCD is decreased by 1.12% and the compiled code size is increased by 0.17% compared to the manually generated one. The automatically generated device driver has no performance degradation in the latency of hardware control at the application program level. The system development time can be reduced using the proposed device driver generation system.

• The Journal of the Korea Contents Association
• /
• v.15 no.4
• /
• pp.9-16
• /
• 2015

On modern operating systems such as Linux, virtual memory is a general way to provide a large address space to applications by using main memory and storage devices. Recently, storage devices have been improved in terms of latency and bandwidth, and it is expected that applications with large memory show high-performance if next-generation storage devices are considered. However, due to the overhead of virtual memory subsystem, the paging system can not exploit the performance of next-generation storage devices. In this study, we propose several optimization techniques to extend memory with next-generation storage devices. The techniques are to allocate block addresses of storage devices for write-back operations as well as to configure the system parameters, and we implement the techniques on Linux 3.14.3. Our evaluation through using multiple benchmarks shows that our system has 3 times (/24%) better performance on average than the baseline system in the micro(/macro)-benchmark.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.16 no.3
• /
• pp.7-12
• /
• 2016

DRAM is commonly used as a smartphone memory medium, but extending its capacity is challenging due to DRAM's large battery consumption and density limit. Meanwhile, smartphone applications such as social network services need increasingly large memory, resulting in long latency due to additional storage accesses. To alleviate this situation, we adopt emerging nonvolatile memory (NVRAM) as smartphone's buffer cache and propose an efficient management scheme. The proposed scheme stores all dirty data in NVRAM, thereby reducing the number of storage accesses. Moreover, it separately exploits read and write histories of data accesses, leading to more efficient management of volatile and nonvolatile buffer caches, respectively. Trace-driven simulations show that the proposed scheme improves I/O performances significantly.

• Journal of Advanced Navigation Technology
• /
• v.14 no.5
• /
• pp.767-774
• /
• 2010

Due to advantages of NAND flash memory such as non-volatility, low access latency, low energy consumption, light weight, small size and shock resistance, it has become a better alternative over traditional magnetic disk drives, and has been widely used. Traditional DBMSs including mobile DBMSs may run on flash memory without any modification by using Flash Translation Layer (FTL), which emulates a random access block device to hide the characteristics of flash memory such as "erase-before-update". However, most existing FTLs are optimized for file systems, not for DBMSs, and traditional DBMSs are not aware of them. Also, traditional DBMSs do not consider the characteristics of flash memory. In this paper, we propose a flash-conscious storage system for DBMSs that utilizes flash memory as a main storage medium, and carefully put the characteristics of flash memory into considerations. The proposed flash-conscious storage system exploits log records to avoid costly update operations. It is shown that the proposed storage system outperforms the state.

• Journal of KIISE:Databases
• /
• v.29 no.3
• /
• pp.159-167
• /
• 2002

Recently, a major issue in the research of metadata is the standardization of metadata format. The new extension capability of metadata in the standardization requires some changes - storing and managing dynamic data consistently. In this paper, we define the characteristics of new metadata and propose a concurrency control called Two Version Latch (2VL). 2VL uses a latch and maintains two versions. Maintaining two versions using latch minimizes conflicts between read operation and write operation. The removal of unnecessary lock holding minimizes refresh latency. Therefore, this algorithm presents fast response time and recent data retrieval in read operation execution. As a result of the performance evaluation, the 2VL algorithm is shown to be better than other algorithms in metadata management system.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.6
• /
• pp.1-12
• /
• 2024

Deep learning with faster and more accurate results requires large amounts of storage space and large computations. Accordingly, many studies are using hardware accelerators for quick and accurate calculations. However, the performance bottleneck is due to data movement between the hardware accelerators and the CPU. In this paper, we propose a data prefetch strategy that can efficiently reduce such operational bottlenecks. The core idea of the data prefetch strategy is to predict the data needed for the next task and upload it to local memory while the hardware accelerator (Matrix Multiplication Unit, MMU) performs a task. This strategy can be enhanced by using a dual buffer to perform read and write operations simultaneously. This reduces latency and execution time of data transfer. Through simulations, we demonstrate a 24% improvement in the performance of hardware accelerators by maximizing parallel processing with dual buffers and bottlenecks between memories with data prefetch.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.48 no.8
• /
• pp.33-44
• /
• 2011

In this paper, we presents a hybrid on-chip bus architecture based on the AMBA 3.0 AXI protocol for MPSoC with high performance and low power. Among AXI channels, data channels with a lot of traffic are designed by crossbar-switch architecture for massively parallel processing. On the other hand, addressing and write-response channels having a few of traffic is handled by shared-bus architecture due to the overheads of (areas, interconnection wires and power consumption) reduction. In experiments, the comparisons are carried out in terms of time, space and power domains for the verification of proposed hybrid on-chip bus architecture. For $16{\times}16$ bus configuration, the hybrid on-chip bus architecture has almost similar performance in time domain with respect to crossbar on-chip bus architecture, as the masters's latency is differenced about 9% and the total execution time is only about 4%. Furthermore, the hybrid on-chip bus architecture is very effective on the overhead reduction, such as it reduced about 47% of areas, and about 52% of interconnection wires, as well as about 66% of dynamic power consumption. Thus, the presented hybrid on-chip bus architecture is shown to be very effective for the MPSoC interconnection design aiming at high performance and low power.