• ETRI Journal
• /
• v.42 no.4
• /
• pp.491-504
• /
• 2020

We present AB9, a neural processor for inference acceleration. AB9 consists of a systolic tensor core (STC) neural network accelerator designed to accelerate artificial intelligence applications by exploiting the data reuse and parallelism characteristics inherent in neural networks while providing fast access to large on-chip memory. Complementing the hardware is an intuitive and user-friendly development environment that includes a simulator and an implementation flow that provides a high degree of programmability with a short development time. Along with a 40-TFLOP STC that includes 32k arithmetic units and over 36 MB of on-chip SRAM, our baseline implementation of AB9 consists of a 1-GHz quad-core setup with other various industry-standard peripheral intellectual properties. The acceleration performance and power efficiency were evaluated using YOLOv2, and the results show that AB9 has superior performance and power efficiency to that of a general-purpose graphics processing unit implementation. AB9 has been taped out in the TSMC 28-nm process with a chip size of 17 × 23 ㎟. Delivery is expected later this year.

• Korean Journal of Materials Research
• /
• v.4 no.2
• /
• pp.159-165
• /
• 1994

Two kinds of humidity sensor are made, one by anodizing pure aluminum and the other by evaporation Sn02 on the anodized pure alumia film, and their electrical characteristics are investigated in various humidity atmosphere. The change of surface resistance with humidity of $AI_2O_3/AI$ and $SnO_2-AI_2O_3/Al$ sensors are found to be $1.40 \times 10^{-2}\Omega$/RH and $1.56 \times 10^{-2}\Omega$/RH, respectively. The hysteresis phenomena associated with the irreversibility of surface resistance-humidity is less in $SnO_2-AI_2O_3/Al$ sensor than in $AI_2O_3/AI$. It is concluded that $SnO_2-AI_2O_3/Al$ film can be used as humidity sensor in room temperature region because temperature dependence of surface resistance of the film is found to be as $0.56 \times 10^{-2} \Omega /^{\circ}C$ in O~ $20^{\circ}C$ range, where as $2.50 \times 10^{-2} \Omega /^{\circ}C$ in 40-$50^{\circ}C$.

• Journal of radiological science and technology
• /
• v.46 no.4
• /
• pp.303-313
• /
• 2023

This study aimed to investigate the relevance of cardiovascular disease risk factors AI and AIP, divided into three groups, among 300 individuals who underwent health checkups at the hospital. Various variables such as Age, Sex, BMI, WC, TC, TG, HDL-C, LDL-C, FBS, HbA1C, SBP, DBP, HR, AI (TC/HDL-C), and AIP (log(TG/HDL-C)) were analyzed using statistical methods including frequency analysis, cross-tabulation, one-way ANOVA, Pearson's correlation analysis, and multiple linear regression analysis. The cross-analysis based on cardiovascular disease risk criteria revealed that men and individuals in their 50s had higher cardiovascular disease risk based on AI and AIP. Significant differences were observed in TG, TC, HDL-C, LDL-C, SBP, DBP, AI (TC/HDL-C), and AIP (log(TG/HDL-C)) according to AI criteria. For the AIP criteria, TG, TC, HDL-C, FBS, HbA1C, HR, AI (TC/HDL-C), and AIP (log(TG/HDL-C)) were identified as cardiovascular disease risk factors. FBS and HbA1c showed the highest positive correlation In the correlation analysis, followed by TC and LDL-C. The lowest positive correlation was observed between LDL-C and DBP. In terms of negative correlation, HDL-C and AI had the highest negative correlation, while LDL-C and TG showed the lowest negative correlation. Multiple regression analysis indicated that the AI and AIP risk criteria had explanatory powers of 73.6% and 72.5%, respectively. HDL-C had the greatest negative effect on the AI risk criterion, while TG had the most significant influence on the AIP risk criterion. In conclusion, while other serological variables are important, managing HDL-C and TG levels may help reduce the risk of cardiovascular disease.

• ETRI Journal
• /
• v.44 no.5
• /
• pp.859-874
• /
• 2022

This paper presents a low-power and lightweight human body communication (HBC) receiver with an embedded dummy electrode for improved signal acquisition. The clock data recovery (CDR) circuit in the receiver operates with a low supply voltage and utilizes a clock phase inversion scheme. The receiver is equipped with a main electrode and dummy electrode that strengthen the capacitive-coupled signal at the receiver frontend. The receiver CDR circuit exploits a clock inversion scheme to allow 0.9-V operation while achieving a shorter lock time than at 3.3-V operation. In experiments, a receiver chip fabricated using 130-nm complementary metal-oxide-semiconductor technology was demonstrated to successfully receive the transmitted signal when the transmitter and receiver are placed separately on each hand of the user while consuming only 4.98 mW at a 0.9-V supply voltage.

• Korean Journal of Materials Research
• /
• v.6 no.12
• /
• pp.1199-1212
• /
• 1996

본 연구는 주조성, 내압성, 내열성 등이 우수하여 군용 및 민수용 기계소재로 이용되고 있는 AI-Si-Mg계 AC4C 합금에 세라믹(AI2O3, AI2O3-TiC)을 강화시키는 복합재료제조에 관한 기초연구의 일환으로 수행하였다. 연구내용은 세라믹 강화재의 젖음성을 높이기 위하여 수소환원법에 의한 AI2O3입자의 Ni 피복과 기존의 프리폰 제조방법보다 간단하고 경제적인 자전연소합성법에 의해 AI2O3-TiC 다공성 pellet을 제조하여, 이들 강화재와 AC4C 기지금속을 이용하여 고대-compocasting 및 squeeze casting법으로 복합재료를 제조하고 미세조직, 계면생성물, 기계적 성질, 내마멸성 등의 특성을 조사하였다. 고대-compocasting법에 의해 제조된 AI2O3Ni 입자 강화 복합재료에서 강화재들은 응집체로 존재하지 않고 비교적 균일하게 분산되었고 AI2O3-TiC 강화재를 이용하여 squeeze casting으로 가압주조 하므로써 기지금속과 강화재의 젖음성이 향상되었다.

• Korean Journal of Materials Research
• /
• v.8 no.7
• /
• pp.596-600
• /
• 1998

Reaction bonded $AI_2O_3$(RBAO)-SiC ceramics were prepared from the Al/$AI_2O_3$/SiC powder mixture. The calcined alumina and fused alumina were used as the raw $AI_2O_3$ powder. The same sized (3mm) and differently sized (3mm + Smm) $Zr0_2$ balls were used in attrition milling of starting powders. The effect of $AI_2O_3$ powder shapes on the milling efficiency and reaction sintering was investigated. After attrition milling, isopressed compacts were preheated to 1l00$^{\circ}$C with a heating rate l.5$^{\circ}C$/min and then sintered at the temperature range 1500 to 1$600^{\circ}C$ for S hours with a heating rate 5$^{\circ}C$/min. The powder mixture with fused alumina was effectively comminuted than the mixtures with calcined alumina. And the same sized balls milled more effctive than the differently sized balls. The reaction sintering behavior differed not depending on the shapes of $AI_2O_3$ powder.

• Korean Journal of Materials Research
• /
• v.8 no.11
• /
• pp.1026-1030
• /
• 1998

Thermal properties of Fe- base amorpous alloys were investigated. $Fe_{80}P_6C_{12}B_{12}$ and $Fe_{73}P_{11}C_6B_4AI_4Ge_2$ amorphous alloys were fabricated by melt spinning method and thermal analysis was done by differential scanning calorimeter. After isothermal crystallization. the Avrami exponents of $Fe_{80}P_6C_{12}B_{12}$ and $Fe_{73}P_{11}C_6B_4AI_4Ge_2$ amorphous alloys were 1.8-2.2 and 2.5-4.0, respectively. It means the former alloy shows diffusion controlled growth and the latter one shows interface controlled growth. For $Fe_{80}P_6C_{12}B_{12}$ and $Fe_{73}P_{11}C_6B_4AI_4Ge_2$ amorphous alloys. the activation energies of isothermal crystallization was 353 and 371kJlmol. Also the activation energies of nucleation and growth were 301, 324kJlmol and 273. 30lkJ/mol, respectively. Thus $Fe_{73}P_{11}C_6B_4AI_4Ge_2$ amorphous alloy is considered to be more stable than $Fe_{73}P_{11}C_6B_4AI_4Ge_2$ amorphous alloy.

• ETRI Journal
• /
• v.44 no.5
• /
• pp.849-858
• /
• 2022

Dynamic voltage frequency scaling (DVFS) has been widely adopted for runtime power management of various processing units. In the case of neural processing units (NPUs), power management of neural network applications is required to adjust the frequency and voltage every layer to consider the power behavior and performance of each layer. Unfortunately, DVFS is inappropriate for layer-wise run-time power management of NPUs due to the long latency of voltage scaling compared with each layer execution time. Because the frequency scaling is fast enough to keep up with each layer, we propose a layerwise dynamic frequency scaling (DFS) technique for an NPU. Our proposed DFS exploits the highest frequency under the power limit of an NPU for each layer. To determine the highest allowable frequency, we build a power model to predict the power consumption of an NPU based on a real measurement on the fabricated NPU. Our evaluation results show that our proposed DFS improves frame per second (FPS) by 33% and saves energy by 14% on average, compared with DVFS.

• Journal of the Korean Society for Heat Treatment
• /
• v.15 no.2
• /
• pp.82-92
• /
• 2002

In the present investigation wear behavior of the 6061AI composites reinforced with 5, 10, 20% SiC particles for dry sliding against a SM45C counterface was studied as a function of load and sliding velocity. Sliding wear tests were conducted at two loads(19.6 and 49N) and three sliding velocities(0.2, 1 and 2 m/sec) at constant sliding distance of 4000 m using pin-on-disk machine under room temperature. Presence of SiC reinforcement particles in the composites has displayed a transition from mild to severe wear at relatively higher applied load and sliding velocity compare to that of the matrix metal. As the volume fraction of SiC particles increased, the transition moved to a more severe wear conditions. Eventually, mild wear prevailed at a most severe wear conditions in this study, that was 49N load and 2 m/sec sliding velocity in 20% SiC particle/6061AI composite.

• Korean Journal of Materials Research
• /
• v.6 no.9
• /
• pp.943-949
• /
• 1996

카올린에서 추출된 황산알루미늄 용액으로부터 촉매담체용 r-AI2O3분말 합성에 대하여 연구하였다. 황산알루미늄 용액을 교반중의 에탄올(ethanol)에 적하하여 단일상(single phase)의 AI2(So4)3 18H2O석출물을 제조하고 이 석출물의 하소로부터 r-AI2O3분말을 합성하였다. 이 분말을 100$0^{\circ}C$에서 2시간 하소하였을 때 열적안정성을 보였으나, 120$0^{\circ}C$-2시간 하소에서 $\alpha$-AI2O3 로 전이하였다. r-AI2O3의 열적안정성에 미치는 BaO 첨가의 영향을 조사 연구하였다. 첨가량은 r-AI2O3에 대해서 1.0-6.0wt%로 하였다. BaO를 4.0wt%첨가한 시료는 AI2O3-BaO.6AI2O3(hexa-aluminate)혼합상생성으로 인하여 120$0^{\circ}C$에서 r-AI2O3의 전이를 방지하는데 효과적이었다. BaO를 4.0wt%첨가한 시료와 BaO를 첨가하지 않은 시료에 대하여 r-AI2O3$\longrightarrow$ $\alpha$-AI2O3전이에 따른 비표면적을 조사하여 보았다. 120$0^{\circ}C$-2시간에서 BaO 4.0wt% 첨가한 시료와 BaO를 첨가하지 않은 시료의 비표면적은 각각 95$m^2$/g과 50$m^2$/g을 유지하였다.