Stable Diffusion has become one of the mainstream image synthesis algorithms. The mainstream computing platform for Stable Diffusion is GPU. However, the deployment of Stable Diffusion on GPU still faces the problems ...
详细信息
Recent advances in developing beyond von Neumann architectures have moved the memristive devices to the forefront as one of the key enablers to realizing memristive computing-in-memory(m CIM)structures, which shows a ...
详细信息
Recent advances in developing beyond von Neumann architectures have moved the memristive devices to the forefront as one of the key enablers to realizing memristive computing-in-memory(m CIM)structures, which shows a great promise to boost the energy-efficiency and the performance of artificial intelligence(AI) chips. In this study, by considering the interactions between devices, circuits, and systems in the m CIM design, we propose several cross-layer design techniques, including(1) the BL-SL interactive forming protection(BSIFP) circuit that can reduce the voltage drop on the selected transistor, suppress the current overshoot by 65.96%, and improve the bit-cell density by more than 10.19%,(2) the clamping transistor trimming scheme(CTTS) to prevent the multiply-and-accumulate(MAC) signal margin degradation from chip-to-chip resistance variations, and(3) dynamic input-parallelism and output-precision(DIPOP) that can reduce the energy cost by 22.92% in a typical inference task with negligible accuracy loss. The results demonstrate the significant role of the cross-layer-interactive approach and provide a preliminary guideline for highly-efficient m CIM design.
A 6-Gb/s half rate current mode logic (CML) transmitter has been designed in TSMC 28nm CMOS technology, which employs a 3-tap 3-bit feed forward equalizer (FFE), an analog duty cycle correction module (DCC) for half r...
详细信息
The COordinate Rotation DIgital Computer (CORDIC) simplifies the elementary function using bit-shift operation and addition. However, the iteration increases with the accuracy and causes a long latency. In this paper,...
详细信息
As a key biomarker for noninvasive diagnosis of diabetes,the selective detection of trace acetone in exhaled gas using a portable and low-cost device remains a great *** metal oxide(SMO)based gas sensors have drawn si...
详细信息
As a key biomarker for noninvasive diagnosis of diabetes,the selective detection of trace acetone in exhaled gas using a portable and low-cost device remains a great *** metal oxide(SMO)based gas sensors have drawn signification attention due to their potential in miniaturization,user-friendliness,high cost-effectiveness and selective real-time detection for noninvasive clinical ***,we propose a one-pot solvent evaporation induced tricomponent co-assembly strategy to design a novel ordered mesoporous SMO of silica-implanted WO_(3)(Si O_(2)/WO_(3))as sensing materials for trace acetone *** controlled co-assembly of silicon and tungsten precursors and amphiphilic diblock copolymer poly(ethylene oxide)-block-polystyrene(PEO-b-PS),and the subsequent thermal treatment enable the local lattice disorder of WO_(3)induced by the amorphous silica and the formation of ordered mesoporous Si O_(2)/WO_(3)hybrid walls with a unique metastableε-phase WO_(3)*** obtained mesoporous SiO_(2)/WO_(3)composites possess highly crystalline framework with large uniform pore size(12.0-13.3 nm),high surface area(99-113 m^(2)/g)and pore volume(0.17-0.23 cm^(3)/g).Typically,the asfabricated gas sensor based on mesoporous 2.5%Si O_(2)/WO_(3)exhibits rapid response/recovery rate(5/17 s),superior sensitivity(R_(air)/R_(gas)=105 for 50 ppm acetone),as well as high selectivity towards *** limit of detection is as low as 0.25 ppm,which is considerably lower than the thresh value of acetone concentration(>1.1 ppm)in the exhaled breath of diabetic patients,demonstrating its great prospect in real-time monitoring in diabetes ***,the mesoporous 2.5%Si O_(2)/WO_(3)sensor is integrated into a wireless sensing module connected to a smart phone,providing a convenient real-time detection of acetone.
Topography recognition is essential for environmental science and ecological research. Conventionally, the deep-learning-based method will improve the recognition accuracy, but the large model size limits the applicat...
详细信息
A 0.9V high performance 3GHz charge pump phase-locked loop (CP PLL) has been designed in TSMC 28nm CMOS technology, which features high accuracy charge pump (CP) and low phase noise LC voltage-controlled oscillator (L...
详细信息
In the field of computer vision, the acquired dataset usually contains a certain number of outliers and noise, which leads to errors in the estimated mathematical model. RANSAC estimates model parameters by randomly s...
详细信息
Finding distinctions and connections between multiple visual targets through the detection of keypoints has become one of the research hot-spots in the field of computer vision. SIFT has received wide recognition and ...
详细信息
Despite the effort of analog circuit design automation, currently complex analog circuit design still requires extensive manual iterations, making it labor intensive and time-consuming. Recently, reinforcement learnin...
详细信息
ISBN:
(纸本)9798350323481
Despite the effort of analog circuit design automation, currently complex analog circuit design still requires extensive manual iterations, making it labor intensive and time-consuming. Recently, reinforcement learning (RL) algorithms have been demonstrated successfully for the analog circuit design optimization. However, a robust and highly efficient RL method to design analog circuits with complex design space has not been fully explored yet. In this work, inspired by multiagent planning theory as well as human expert design practice, we propose a multiagent based RL (MA-RL) framework to tackle this issue. Particularly, we (i) partition the complex analog circuits into several sub-blocks based on topology information and effectively reduce the complexity of design search space; (ii) leverage MA-RL for the circuit optimization, where each agent corresponds to a single sub-block, and the interactions between agents delicately mimic the best design tradeoffs between circuit sub-blocks by human experts; (iii) introduce the multiagent twin-delayed techniques to further boost training stability and accomplish higher performances. Experiments on two different analog circuit topologies and knowledge transfers between two technology nodes are demonstrated. It's shown that MA-RL framework can achieve the best FoM for complex analog circuits design. This work shines the light for future large scale analog circuit system design automation.
暂无评论