No Arabic abstract
In this paper, we propose a Spin-Torque (ST) based sensing scheme that can enable energy efficient multi-bit long distance interconnect architectures. Current-mode interconnects have recently been proposed to overcome the performance degradations associated with conventional voltage mode Copper (Cu) interconnects. However, the performance of current mode interconnects are limited by analog current sensing transceivers and equalization circuits. As a solution, we propose the use of ST based receivers that use Magnetic Tunnel Junctions (MTJ) and simple digital components for current-to-voltage conversion and do not require analog transceivers. We incorporate Spin-Hall Metal (SHM) in our design to achieve high speed sensing. We show both single and multi-bit operations that reveal major benefits at higher speeds. Our simulation results show that the proposed technique consumes only 3.93-4.72 fJ/bit/mm energy while operating at 1-2 Gbits/sec; which is considerably better than existing charge based interconnects. In addition, Voltage Controlled Magnetic Anisotropy (VCMA) can reduce the required current at the sensor. With the inclusion of VCMA, the energy consumption can be further reduced to 2.02-4.02 fJ/bit/mm
Optical interconnects have attracted significant research interest for use in short-reach board-level optical communication links in supercomputers and data centres. Multimode polymer waveguides in particular constitute an attractive technology for on-board optical interconnects as they provide high bandwidth, offer relaxed alignment tolerances, and can be cost-effectively integrated onto standard printed circuit boards (PCBs). However, the continuing improvements in bandwidth performance of optical sources make it important to investigate approaches to develop high bandwidth polymer waveguides. In this paper, we present dispersion studies on a graded-index (GI) waveguide in siloxane materials designed to deliver high bandwidth over a range of launch conditions. Bandwidth-length products of >70 GHzxm and ~65 GHzxm are observed using a 50/125 um multimode fibre (MMF) launch for input offsets of +/- 10 um without and with the use of a mode mixer respectively; and enhanced values of >100 GHzxm are found under a 10x microscope objective launch for input offsets of ~18 x 20 um^2. The large range of offsets is within the -1 dB alignment tolerances. A theoretical model is developed using the measured refractive index profile of the waveguide, and general agreement is found with experimental bandwidth measurements. The reported results clearly demonstrate the potential of this technology for use in high-speed board-level optical links, and indicate that data transmission of 100 Gb/s over a multimode polymer waveguide is feasible with appropriate refractive index engineering.
Optical interconnects play a key role in the implementation of high-speed short-reach communication links within high-performance electronic systems. Multimode polymer waveguides in particular are strong candidates for use in passive optical backplanes as they can be cost-effectively integrated onto standard PCBs. Various optical backplanes using this technology and featuring a large number of multimode polymer waveguide components have been recently demonstrated. The optimisation of the loss performance of these complex waveguide layouts becomes particularly important at high data rates (>=25 Gb/s) due to the associated stringent power budget requirements. Moreover, launch conditions have to be carefully considered in such systems due to the highly-multimoded nature of this waveguide technology. In this paper therefore, we present thorough loss and bandwidth studies on siloxane-based multimode waveguides and waveguide components (i.e. bends and crossings) that enable the implementation of passive optical backplanes. The performance of these components is experimentally investigated under different launch conditions for different waveguide profiles that can be readily achieved through fabrication. Useful design rules on the use of waveguide bends and crossings are derived for each waveguide type. It is shown that the choice of waveguide parameters depends on the particular waveguide layout, assumed launch conditions and desired link bandwidth. As an application of these studies, the obtained results are employed to optimise the loss performance of a 10-card shuffle router and enable >=40 Gb/s data transmission.
This paper reports the demonstration of high-speed PAM-4 transmission using a 1.5-{mu}m single-mode vertical cavity surface emitting laser (SM-VCSEL) over multicore fiber with 7 cores over different distances. We have successfully generated up to 70 Gbaud 4-level pulse amplitude modulation (PAM-4) signals with a VCSEL in optical back-to-back, and transmitted 50 Gbaud PAM-4 signals over both 1-km dispersion-uncompensated and 10-km dispersion-compensated in each core, enabling a total data throughput of 700 Gbps over the 7-core fiber. Moreover, 56 Gbaud PAM-4 over 1-km has also been shown, whereby unfortunately not all cores provide the required 3.8 $times$ 10 $^{-3}$ bit error rate (BER) for the 7% overhead-hard decision forward error correction (7% OH HDFEC). The limited bandwidth of the VCSEL and the adverse chromatic dispersion of the fiber are suppressed with pre-equalization based on accurate end-to-end channel characterizations. With a digital post-equalization, BER performance below the 7% OH-HDFEC limit is achieved over all cores. The demonstrated results show a great potential to realize high-capacity and compact short-reach optical interconnects for data centers.
Spin-orbit torques (SOTs), which rely on spin current generation from charge current in a nonmagnetic material, promise an energy-efficient scheme for manipulating magnetization in magnetic devices. A critical topic for spintronic devices using SOTs is to enhance the charge to spin conversion efficiency. Besides, the current-induced spin polarization is usually limited to in-plane, whereas out-of-plane spin polarization could be favored for efficient perpendicular magnetization switching. Recent advances in utilizing two important classes of van der Waals materials$-$topological insulators and transition-metal dichalcogenides$-$as spin sources to generate SOT shed light on addressing these challenges. Topological insulators such as bismuth selenide have shown a giant SOT efficiency, which is larger than those from three-dimensional heavy metals by at least one order of magnitude. Transition-metal dichalcogenides such as tungsten telluride have shown a current-induced out-of-plane spin polarization, which is allowed by the reduced symmetry. In this review, we use symmetry arguments to predict and analyze SOTs in van der Waal material-based heterostructures. We summarize the recent progress of SOT studies based on topological insulators and transition-metal dichalcogenides and show how these results are in line with the symmetry arguments. At last, we identify unsolved issues in the current studies and suggest three potential research directions in this field.
Probabilistic inference from real-time input data is becoming increasingly popular and may be one of the potential pathways at enabling cognitive intelligence. As a matter of fact, preliminary research has revealed that stochastic functionalities also underlie the spiking behavior of neurons in cortical microcircuits of the human brain. In tune with such observations, neuromorphic and other unconventional computing platforms have recently started adopting the usage of computational units that generate outputs probabilistically, depending on the magnitude of the input stimulus. In this work, we experimentally demonstrate a spintronic device that offers a direct mapping to the functionality of such a controllable stochastic switching element. We show that the probabilistic switching of Ta/CoFeB/MgO heterostructures in presence of spin-orbit torque and thermal noise can be harnessed to enable probabilistic inference in a plethora of unconventional computing scenarios. This work can potentially pave the way for hardware that directly mimics the computational units of Bayesian inference.