A novel scheme for the design of an ultra-compact and high-performance optical switch is proposed and investigated numerically. Based on a standard silicon(Si) photonic stripe waveguide, a section of hyperbolic metama...A novel scheme for the design of an ultra-compact and high-performance optical switch is proposed and investigated numerically. Based on a standard silicon(Si) photonic stripe waveguide, a section of hyperbolic metamaterials(HMM) consisting of 20-pair alternating vanadium dioxide (VO_2)∕Si thin layers is inserted to realize the switching of fundamental TE mode propagation. Finite-element-method simulation results show that, with the help of an HMM with a size of 400 nm × 220 nm × 200 nm(width × height × length), the ON/OFF switching for fundamental TE mode propagation in an Si waveguide can be characterized by modulation depth(MD) of5.6 d B and insertion loss(IL) of 1.25 dB. It also allows for a relatively wide operating bandwidth of 215 nm maintaining MD > 5 dB and IL < 1.25 dB. Furthermore, we discuss that the tungsten-doped VO_2 layers could be useful for reducing metal-insulator-transition temperature and thus improving switching performance. In general, our findings may provide some useful ideas for optical switch design and application in an on-chip all-optical communication system with a demanding integration level.展开更多
Modern recommendation systems are widely used in modern data centers.The random and sparse embedding lookup operations are the main performance bottleneck for processing recommendation systems on traditional platforms...Modern recommendation systems are widely used in modern data centers.The random and sparse embedding lookup operations are the main performance bottleneck for processing recommendation systems on traditional platforms as they induce abundant data movements between computing units and memory.ReRAM-based processing-in-memory(PIM)can resolve this problem by processing embedding vectors where they are stored.However,the embedding table can easily exceed the capacity limit of a monolithic ReRAM-based PIM chip,which induces off-chip accesses that may offset the PIM profits.Therefore,we deploy the decomposed model on-chip and leverage the high computing efficiency of ReRAM to compensate for the decompression performance loss.In this paper,we propose ARCHER,a ReRAM-based PIM architecture that implements fully yon-chip recommendations under resource constraints.First,we make a full analysis of the computation pattern and access pattern on the decomposed table.Based on the computation pattern,we unify the operations of each layer of the decomposed model in multiply-and-accumulate operations.Based on the access observation,we propose a hierarchical mapping schema and a specialized hardware design to maximize resource utilization.Under the unified computation and mapping strategy,we can coordinatethe inter-processing elements pipeline.The evaluation shows that ARCHER outperforms the state-of-the-art GPU-based DLRM system,the state-of-the-art near-memory processing recommendation system RecNMP,and the ReRAM-based recommendation accelerator REREC by 15.79×,2.21×,and 1.21× in terms of performance and 56.06×,6.45×,and 1.71× in terms of energy savings,respectively.展开更多
Laser-assisted gas nitriding of selective Ti-6Al-4V surfaces has been achieved during laser powder bed fusion fabrication by exchanging the argon build gas environment with nitrogen.Systematic variation of processing ...Laser-assisted gas nitriding of selective Ti-6Al-4V surfaces has been achieved during laser powder bed fusion fabrication by exchanging the argon build gas environment with nitrogen.Systematic variation of processing parameters allowed microdendritic Ti N surface coatings to be formed having thicknesses ranging from a few tens of microns to several hundred microns,with TiN dendrite microstructure volume fractions ranging from 0.6 to 0.75;and corresponding Vickers microindentation hardness values ranging from^7.5 GPa–9.5 GPa.Embedded TiN hard layers ranging from 50μm to 150μm thick were also fabricated in the laser-beam additively manufactured Ti-6Al-4V alloy producing prototype,hybrid,planar composites having alternating,ductile Ti-6Al-4V layers with a hardness of^4.5 GPa and a stiff,TiN layer with a hardness of^8.5 GPa.The results demonstrate prospects for fabricating novel,additively manufactured components having selective,hard,wear and corrosion resistant coatings along with periodic,planar or complex metal matrix composite regimes exhibiting superior toughness and related mechanical properties.展开更多
This paper＂ describes and analyses the impact of the Ti layer, which is embedded between the insulator and top electrode, on the programming characteristic of the A1-HfO2-AI antifuse. The programming voltage of the a...This paper＂ describes and analyses the impact of the Ti layer, which is embedded between the insulator and top electrode, on the programming characteristic of the A1-HfO2-AI antifuse. The programming voltage of the antiiikse with 120 A HfO2 is properly reduced from 5.5 to 4.6 V with an embedded Ti layer. Low on-state resistance （-19Ω） and low programming voltage （4.6 V） is demonstrated in the embedded Ti antifhse with 120 A H fO2 while keeping sufficient off-state reliability. The antifuse embedded with a Ti layer between the insulator and top electrode has been developed and has potential in field programmable devices.展开更多
基金Ministry of Science and Technology of the People's Republic of China(MOST)(2016YFA0301300)National Natural Science Foundation of China(NSFC)(61275201,61372037)+2 种基金Beijing University of Posts and Telecommunications(BUPT)Excellent Ph.D.Students Foundation(CX2016204)Fundamental Research Funds for the Central Universities(2016RC24)Beijing Excellent Ph.D.Thesis Guidance Foundation(20131001301)
文摘A novel scheme for the design of an ultra-compact and high-performance optical switch is proposed and investigated numerically. Based on a standard silicon(Si) photonic stripe waveguide, a section of hyperbolic metamaterials(HMM) consisting of 20-pair alternating vanadium dioxide (VO_2)∕Si thin layers is inserted to realize the switching of fundamental TE mode propagation. Finite-element-method simulation results show that, with the help of an HMM with a size of 400 nm × 220 nm × 200 nm(width × height × length), the ON/OFF switching for fundamental TE mode propagation in an Si waveguide can be characterized by modulation depth(MD) of5.6 d B and insertion loss(IL) of 1.25 dB. It also allows for a relatively wide operating bandwidth of 215 nm maintaining MD > 5 dB and IL < 1.25 dB. Furthermore, we discuss that the tungsten-doped VO_2 layers could be useful for reducing metal-insulator-transition temperature and thus improving switching performance. In general, our findings may provide some useful ideas for optical switch design and application in an on-chip all-optical communication system with a demanding integration level.
基金This work was supported by the National Key R&D Program of China(No.2022YFB4501403)the National Natural Science Foundation of China(Grant Nos.62322205,62072195,61825202,and 61832006)the Zhejiang Lab(No.2022PI0AC02).
文摘Modern recommendation systems are widely used in modern data centers.The random and sparse embedding lookup operations are the main performance bottleneck for processing recommendation systems on traditional platforms as they induce abundant data movements between computing units and memory.ReRAM-based processing-in-memory(PIM)can resolve this problem by processing embedding vectors where they are stored.However,the embedding table can easily exceed the capacity limit of a monolithic ReRAM-based PIM chip,which induces off-chip accesses that may offset the PIM profits.Therefore,we deploy the decomposed model on-chip and leverage the high computing efficiency of ReRAM to compensate for the decompression performance loss.In this paper,we propose ARCHER,a ReRAM-based PIM architecture that implements fully yon-chip recommendations under resource constraints.First,we make a full analysis of the computation pattern and access pattern on the decomposed table.Based on the computation pattern,we unify the operations of each layer of the decomposed model in multiply-and-accumulate operations.Based on the access observation,we propose a hierarchical mapping schema and a specialized hardware design to maximize resource utilization.Under the unified computation and mapping strategy,we can coordinatethe inter-processing elements pipeline.The evaluation shows that ARCHER outperforms the state-of-the-art GPU-based DLRM system,the state-of-the-art near-memory processing recommendation system RecNMP,and the ReRAM-based recommendation accelerator REREC by 15.79×,2.21×,and 1.21× in terms of performance and 56.06×,6.45×,and 1.71× in terms of energy savings,respectively.
基金provided through the MSI STEM Research&Development Consortium sponsored by the U.S.Army via cooperative agreement#W911SR-14-2-0001 project number 0025。
文摘Laser-assisted gas nitriding of selective Ti-6Al-4V surfaces has been achieved during laser powder bed fusion fabrication by exchanging the argon build gas environment with nitrogen.Systematic variation of processing parameters allowed microdendritic Ti N surface coatings to be formed having thicknesses ranging from a few tens of microns to several hundred microns,with TiN dendrite microstructure volume fractions ranging from 0.6 to 0.75;and corresponding Vickers microindentation hardness values ranging from^7.5 GPa–9.5 GPa.Embedded TiN hard layers ranging from 50μm to 150μm thick were also fabricated in the laser-beam additively manufactured Ti-6Al-4V alloy producing prototype,hybrid,planar composites having alternating,ductile Ti-6Al-4V layers with a hardness of^4.5 GPa and a stiff,TiN layer with a hardness of^8.5 GPa.The results demonstrate prospects for fabricating novel,additively manufactured components having selective,hard,wear and corrosion resistant coatings along with periodic,planar or complex metal matrix composite regimes exhibiting superior toughness and related mechanical properties.
文摘This paper＂ describes and analyses the impact of the Ti layer, which is embedded between the insulator and top electrode, on the programming characteristic of the A1-HfO2-AI antifuse. The programming voltage of the antiiikse with 120 A HfO2 is properly reduced from 5.5 to 4.6 V with an embedded Ti layer. Low on-state resistance （-19Ω） and low programming voltage （4.6 V） is demonstrated in the embedded Ti antifhse with 120 A H fO2 while keeping sufficient off-state reliability. The antifuse embedded with a Ti layer between the insulator and top electrode has been developed and has potential in field programmable devices.
