An adaptive frequency synthesizer architecture reducing reference sidebands

An adaptive phase-locked loop (PLL) frequency synthesizer architecture for reducing reference sidebands at the output of the frequency synthesizer is described. The architecture combines two tuning loops: one is the main loop for locking the PLL frequency synthesizer and operating all the time, the other one is auxiliary loop for reducing reference sidebands and operating only when the main loop is closely locked. A 1.8V 1GHz fully integrated CMOS dual-loop frequency synthesizer is designed in a 0.18um CMOS process. The suppression of the reference sidebands of the proposed frequency synthesizer is 13.8dB more than that of the general frequency synthesizer.

Novel compact SOI-based reconfigurable optical add/drop multiplexer using microring resonators - art. no. 60190L

Novel compact design for 4-channel SOI-based reconfigurable optical add/drop multiplexer using microring resonators is presented and analyzed. Microring resonators have two important attributes as a key new technology for future optical communications, namely functionality and compactness. Functionality refers to the fact that a wide range of desirable filter characteristics can be synthesized by coupling multiple rings. Compactness refers the fact that ring resonators with radii about 30 mu m can lead to large scale integration of devices with densities on the order of 10(4) similar to 10(5) devices per square centimeter. A 4-channel reconfigurable optical add/drop multiplexer comprises a grid-like array of ridge waveguides which perpendicularly cross through each other. SOI-based resonators consisted of multiple rings at each of the cross-grid nodes serve as the wavelength selective switch, and they can switch an optical signal between two ports by means of tuning refractive index of one of the rings. The thermo-optic coefficient of silicon is 1.86x 10(-4) /K. Thus a temperature rise of 27K will increase the refractive index by 5 x 10(-3), which is enough to cause the switching of our designed microring resonators. The thermo-optic effect is used to suppress the resonator power transfer, rather than to promote loss. Thus, the input signal only suffers small attenuation and simultaneously low crosstalk can be achieved by using multiple rings.

Fabrication of low cost Si-based tunable high performance resonant cavity enhanced photodetectors

Low cost Si-based tunable InGaAs RCE photodetectors operating at 1.3similar to1.6 mum were fabricated using sol-gel bonding. A tuning range of 14.5 nm, a quantum efficiency of 44% at 1476 nm and a 3-dB bandwidth of 1.8 GHz were obtained.

A Novel 0.72-6.2GHz Continuously-Tunable Delta Sigma Fractional-N Frequency Synthesizer

This paper presents a wideband Delta Sigma-based fractional-N synthesizer with three integrated quadrature VCOs for multiple-input multiple-output (MIMO) wireless communication applications. It continuously covers a wide range frequency from 0.72GHz to 6.2GHz that is suitable for multiple communication standards. The synthesizer is designed in 0.13-um RE CMOS process. The dual clock full differential multi-modulus divide (MMD) with low power consumption can operate over 9GHz under the worst condition. In the whole range frequency from 0.72GHz to 6.2GHz, the maximal tuning range of the QVCOs reaches 33.09% and their phase noise is -119d8/Hz similar to 124d8/Hz @1MHz. Its current is less than 12mA at a 1.2V voltage supply when it operates at the highest frequency of 6.2GHz.

A broadband external cavity tunable InAs/GaAs quantum dot laser by utilizing only the ground state emission

A broadband external cavity tunable laser is realized by using a broad-emitting spectral InAs/GaAs quantum dot (QD) gain device. A tuning range of 69 nm with a central wavelength of 1056 nm, is achieved at a bias of 1.25 kA/cm(2) only by utilizing the light emission from the ground state of QDs. This large tunable range only covers the QD ground-state emission and is related to the inhomogeneous size distribution of QDs. No excited state contributes to the tuning bandwidth. The application of the QD gain device to the external cavity tunable laser shows its immense potential in broadening the tuning bandwidth. By the external cavity feedback, the threshold current density can be reduced remarkably compared with the free-running QD gain device.

The Fabrication of Eight-Channel DFB Laser Array Using Sampled Gratings

An eight-channel monolithically integrated complex-coupled distributed-feedback laser array based on sampled gratings has been designed and fabricated. Selective lasing at different wavelengths is obtained. The frequency separation between each adjacent channel is about 200 GHz. The typical threshold current is between 30 and 40 mA. The optical output power of each channel is about 10 mW at an injection current of 100 mA. The continuous tuning of emission wavelength with injected currents is also demonstrated.

Broadband external cavity tunable quantum dot lasers with low injection current density

Broadband grating-coupled external cavity laser, based on InAs/GaAs quantum dots, is achieved. The device has a wavelength tuning range from 1141.6 nm to 1251.7 nm under a low continuous-wave injection current density (458 A/cm(2)). The tunable bandwidth covers consecutively the light emissions from both the ground state and the 1st excited state of quantum dots. The effects of cavity length and antireflection facet coating on device performance are studied. It is shown that antireflection facet coating expands the tuning bandwidth up to similar to 150 nm, accompanied by an evident increase in threshold current density, which is attributed to the reduced interaction between the light field and the quantum dots in the active region of the device.

Delay of the excited state lasing of 1310 nm InAs/GaAs quantum dot lasers by facet coating

In this letter, we present a facet coating design to delay the excited state (ES) lasing for 1310 nm InAs/GaAs quantum dot lasers. The key point of our design is to ensure that the mirror loss of ES is larger than that of the ground state by decreasing the reflectivity of the ES. In the facet coating design, the central wavelength is at 1480 nm, and the high- and low-index materials are Ta2O5 and SiO2, respectively. Compared with the traditional Si/SiO2 facet coating with a central wavelength of 1310 nm, we have found that with the optimal design the turning temperature of the ES lasing has been delayed from 90 to 100 degrees C for the laser diodes with cavity length of 1.2 mm. Furthermore, the characteristic temperature (T-0) of the laser diodes is also improved.

Photoexcited charge current for the presence of pure spin current

The asymmetric spin distribution in k space caused by the pure spin current (PSC) can introduce a photoexcited charge current (PECC). This provides us a practical scheme for direct detection of PSC. We demonstrate theoretically that the PECC related to the PSC depends sensitively on the wave vector and spin orientation of the carriers, more important, the helicity dependence of this PECC provides us a way to refine it from the helicity independent background current by tuning the polarized laser beams from left to right circular polarization.

All-optical clock recovery using a ridge width varied two-section partly gain-coupled DFB self-pulsation laser

A 1.55 mu m InGaAsP-InP partly gain-coupled two-section DFB self-pulsation laser (SPL) with a varied ridge width has been fabricated. The laser produces self-pulsations with a frequency tuning range of more than 135 GHz. All-optical clock recovery from 40 Gb/s degraded data streams has been demonstrated. Successful lockings of the device at frequencies of 30 GHz, 40 GHz, 50 GHz, and 60 GHz to a 10 GHz sidemode injection are also conducted, which demonstrates the capability of the device for all-optical clock recovery at different frequencies. This flexibility of the device is highly desired for practical uses. Crown Copyright

Stable single-mode distributed feedback quantum cascade laser with surface metal grating

A design of single-mode distributed feedback quantum cascade lasers (DFB-QCLs) with surface metal grating is described. A rigorous modal expansion theory is adopted to analyse the interaction between the waveguide mode and the surface plasmon wave for different grating parameters. A stable single-mode operation can be obtained in a wide range of grating depths and duty cycles. The single-mode operation of surface metal grating DFB-QCLs at room temperature for lambda = 8.5 mu m is demonstrated. The device shows a side-mode suppression ratio of above 20 dB. A linear tuning of wavelength with temperature indicates the stable single-mode operation without mode hopping.

SpMV的自动性能优化实现技术及其应用研究

在科学计算中,稀疏矩阵向量乘(SpMV)是一个十分重要且经常被大量调用的计算内核.由于SpMV一般实现算法的浮点计算和存储访问次数比率非常低,且其存储访问模式极为不规则,其实际运行性能往往很低.通过采用寄存器分块算法和启发式分块大小选择算法,将稀疏矩阵分成小的稠密分块,重用保存在寄存器中向量x元素,可以提高该计算内核的性能.剖析和总结了OSKI软件包所采用的若干关键优化技术,并进行了实际应用性能测试.测试表明,在实际应用这些优化技术的过程中,应用程序对SpMV的调用次数要达到上百次的量级,才能抵消由于应用这些性能优化技术所带来的额外时间开销,取得性能加速效果.在Pentium4和AMD Athlon平台上,测试了10个矩阵,其平均加速比分别达到了1.69和1.48.

循环优化序列自动定制方法研究

现代计算机在体系结构和应用场景复杂性的增长使得程序性能的增长、保持程序性能的可移植性以及程序开发效率的提升越来越困难，程序自动调优(auto-tuning) 是解决此问题的一个可行途径。本文研究一种通过为循环嵌套自动定制循环优化序列来进行程序自动调优的方法，提高了被编译程序的性能和可移植性，也有助于提升开发效率。 在综述相关工作的基础上，本文首先分析了循环优化序列自动定制研究需要解决的几个关键问题：循环优化的形式化描述及实施、循环优化序列的性能评估、循环优化序列的自动定制方法以及相关的原型工具。本文围绕这几个问题进行研究。 本文采用了Polyhedron模型来描述循环优化，并基于此实施了循环优化的仿射变换。这一工作基于URUK/WRaP-IT开源软件包实现，并作出了若干重要改进。借助Polyhedron模型，本文还给出了部分常见循环优化之间的使能(enable)关系。循环优化之间的相互影响较为普遍，一个循环优化的执行可能会破坏后续优化的实施条件，也可能不破坏甚至创造相应条件。这一关系可以用于过滤循环优化序列定制过程中产生的无效序列从而加快定制速度。本文分析了两个循环优化以先后次序执行时，循环嵌套内所有依赖距离向量均得以保持的条件，给出了它们之间的使能关系。并且给出了使能关系在Polyhedron模型下的计算方法，以及在序列搜索过程中通过维护依赖距离向量状态计算使能关系的方法。 对于循环优化序列性能评估方法，本文采用了高速缓存失效率简化方程(CMES方程)以动态给出循环嵌套中的高速缓存失效率数据，提出了LoopCost模型以静态分析方式发现循环嵌套中对性能影响最大的维度，这两种评估方法分别适用于不同的序列定制方法。 给出了Cache 失效率简化方程(CMES, Cache Miss Equation Simplified)在Polyhedron 模型下的求解方法。CMES方程可在编译时给出循环嵌套中的Cache失效率数据，不需要实际编译运行目标程序，从而避免相关文献中使用运行时信息来评价优化序列带来的巨大编译开销。给出了CMES方程各个参数从循环嵌套的Polyhedron模型中提取的方法，并给出了具体计算过程。 提出了一种可发现循环嵌套中对性能影响最大维度的循环优化性能评估模型LoopCost。相关文献中所使用的性能评估模型可给出一个循环嵌套的整体性能，但无法给出哪个维度对其性能影响最大；而大多数循环优化实际上是对某一个维度所做的变换，因此这些评估模型不能直接指导优化序列的搜索，只能是在一个迭代过程中提供性能信息。LoopCost模型基于数据依赖关系对一个循环嵌套估算出其每个维度使用的cache行数，作为衡量局部性优劣的指标，适用于紧嵌套循环和非紧嵌套循环，并且其复杂度与循环嵌套个数和引用个数成线性关系。通过对SPEC CPU 2006中5,703个循环嵌套的实验，LoopCost模型对其中98.3%的循环嵌套都能给出正确的性能评估数据。 对于循环优化序列自动定制方法，本文首先提出了一种基于CMES方程的Custom方法，使用迭代方式搜索可行优化序列；为了进一步提升优化效率，又提出了一种基于LoopCost模型的快速定制方法FastCustom。 Custom方法使用CMES方程评估优化序列，使用遗传算法作为序列搜索方法。在遗传算法中的各个算子中利用循环优化的enable关系加速搜索速度。该方法可以避免使用运行时刻信息评估性能所带来的巨大编译开销，又可避免相关工作中利用Polyhedron模型中语句实例执行次序矩阵的性质来搜索带来的局限性，适用的循环优化范围广，编译效率高，同时能取得与相关工作基本相当的加速比。 FastCustom方法优化效率高，并且与Custom方法不同，适用于所有场合。它基于LoopCost模型寻找对性能影响最大的维度，并根据一个启发式策略给出一个循环优化序列。该方法编译速度快，优化效果与Custom方法基本相当。与Custom相比，该方法需要给出启发式策略，因此适用的循环优化范围不如Custom，适用于迭代编译无法进行的情形。 对于循环优化序列自动定制原型工具，本文设计并实现了原型工具CPOLO。该工具改进了URUK/WRaP-IT开源软件包，使其能够支持较为复杂的程序；增强了代码重生成的效率；并且实现了本文所提出的优化序列性能评估模型和自动定制方法。 最后，总结了全文并探讨了进一步的工作方向。

Tunable Ag surface-plasmon-resonance wavelength and its application on the photochromic behavior of TiO2-Ag films

Surface and bulk plasmon resonance of noble metal particles play an essential role in the multicolor photochromism of semiconductor systems containing noble metal particles, Here we examined several key parameters affecting surface plasmon resonance wavelength (SPRW) of Ag particles and investigated the relation between surface plasmon and photochromic reaction wavelength. From the transmission spectra of sandwiched (TiO2/Ag/TiO2) and overcoated (Ag/TiO2) films deposited on quartz substrates at room temperature by rf helicon magnetron sputtering, we demonstrated that the SPRW can be made tunable by changing the surrounding media and thickness of the metal layer. The coloration and bleaching in visible light region due to photochromism were clearly observed for the films inserted with a 0.55 nm Ag layer.