scholarly journals RRAM-Based CAM Combined With Time-Domain Circuits for Hyperdimensional Computing

2021 ◽  
Author(s):  
Yasmin Halawani ◽  
Dima Kilani ◽  
Eman Hassan ◽  
Huruy Tesfai ◽  
Hani Saleh ◽  
...  
Keyword(s):  
Time Domain ◽  
High Speed ◽  
Clock Cycle ◽  
Content Addressable Memory ◽  
The Time Domain ◽  
Analog Voltage ◽  
Voltage Saturation ◽  
Reduction In Area ◽  
Remarkable Reduction ◽  
Analog Adder

Abstract Content addressable memory (CAM) for search and match operations demands high speed and low power for near real-time decision-making across many critical domains. Resistive RAM-based in-memory computing has high potential in realizing an efficient static CAM for artificial intelligence tasks, especially on resource-constrained platforms. This paper presents an XNOR-based RRAM-CAM with a time-domain analog adder for efficient winning class computation. The CAM compares two operands, one voltage and the second one resistance, and outputs a voltage proportional to the similarity between the input query and the pre-stored patterns. Processing the summation of the output similarity voltages in the time-domain helps avoid voltage saturation, variation, and noise dominating the analog voltage-based computing. After that, to determine the winning class among the multiple classes, a digital realization is utilized to consider the class with the longest pulse width as the winning class. As a demonstrator, hyperdimensional computing for efficient MNIST classification is considered.The proposed design uses 65nm CMOS foundry technology and realistic data for RRAM with total area of 0.0077 mm2 , consumes 13.6 pJ of energy per 1k query within 10 ns clock cycle for 10 classes. It shows a reduction of ∼ 31× in area and ∼ 3× in energy consumption compared to fully digital ASIC implementation using 65nm foundry technology. The proposed design exhibits a remarkable reduction in area and energy compared to two of the state-of-the-art RRAM designs.

scholarly journals RRAM-based CAM combined with time-domain circuits for hyperdimensional computing

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Yasmin Halawani ◽  
Dima Kilani ◽  
Eman Hassan ◽  
Huruy Tesfai ◽  
Hani Saleh ◽  
...  
Keyword(s):  
Time Domain ◽  
High Speed ◽  
Clock Cycle ◽  
Content Addressable Memory ◽  
The Time Domain ◽  
Analog Voltage ◽  
Voltage Saturation ◽  
Reduction In Area ◽  
Remarkable Reduction ◽  
Analog Adder

AbstractContent addressable memory (CAM) for search and match operations demands high speed and low power for near real-time decision-making across many critical domains. Resistive RAM (RRAM)-based in-memory computing has high potential in realizing an efficient static CAM for artificial intelligence tasks, especially on resource-constrained platforms. This paper presents an XNOR-based RRAM-CAM with a time-domain analog adder for efficient winning class computation. The CAM compares two operands, one voltage and the second one resistance, and outputs a voltage proportional to the similarity between the input query and the pre-stored patterns. Processing the summation of the output similarity voltages in the time-domain helps avoid voltage saturation, variation, and noise dominating the analog voltage-based computing. After that, to determine the winning class among the multiple classes, a digital realization is utilized to consider the class with the longest pulse width as the winning class. As a demonstrator, hyperdimensional computing for efficient MNIST classification is considered. The proposed design uses 65 nm CMOS foundry technology and realistic data for RRAM with total area of 0.0077 mm2, consumes 13.6 pJ of energy per 1 k query within 10 ns clock cycle. It shows a reduction of ~ 31 × in area and ~ 3 × in energy consumption compared to fully digital ASIC implementation using 65 nm foundry technology. The proposed design exhibits a remarkable reduction in area and energy compared to two of the state-of-the-art RRAM designs.

scholarly journals 15 Years MR-encephalography

2020 ◽  
Author(s):  
Juergen Hennig ◽  
Vesa Kiviniemi ◽  
Bruno Riemenschneider ◽  
Antonia Barghoorn ◽  
Burak Akin ◽  
...  
Keyword(s):  
Time Domain ◽  
High Speed ◽  
Direct Detection ◽  
Principal Component ◽  
Isotropic Resolution ◽  
High Acceleration Factor ◽  
Spiral Trajectory ◽  
Time Domain Data ◽  
High Acceleration ◽  
The Time Domain

Abstract Objective This review article gives an account of the development of the MR-encephalography (MREG) method, which started as a mere ‘Gedankenexperiment’ in 2005 and gradually developed into a method for ultrafast measurement of physiological activities in the brain. After going through different approaches covering k-space with radial, rosette, and concentric shell trajectories we have settled on a stack-of-spiral trajectory, which allows full brain coverage with (nominal) 3 mm isotropic resolution in 100 ms. The very high acceleration factor is facilitated by the near-isotropic k-space coverage, which allows high acceleration in all three spatial dimensions. Methods The methodological section covers the basic sequence design as well as recent advances in image reconstruction including the targeted reconstruction, which allows real-time feedback applications, and—most recently—the time-domain principal component reconstruction (tPCR), which applies a principal component analysis of the acquired time domain data as a sparsifying transformation to improve reconstruction speed as well as quality. Applications Although the BOLD-response is rather slow, the high speed acquisition of MREG allows separation of BOLD-effects from cardiac and breathing related pulsatility. The increased sensitivity enables direct detection of the dynamic variability of resting state networks as well as localization of single interictal events in epilepsy patients. A separate and highly intriguing application is aimed at the investigation of the glymphatic system by assessment of the spatiotemporal patterns of cardiac and breathing related pulsatility. Discussion MREG has been developed to push the speed limits of fMRI. Compared to multiband-EPI this allows considerably faster acquisition at the cost of reduced image quality and spatial resolution.

scholarly journals Development of a High-Speed Digitizer to Time Resolve Nanosecond Fluorescence Pulses

2012 ◽  
Vol 10 (2) ◽  
Author(s):  
E. Moreno-García ◽  
R. Galicia-Mejía ◽  
D. Jiménez-Olarte ◽  
J. M. de la Rosa Vázquez ◽  
S. Stolik-Isakina
Keyword(s):  
Time Domain ◽  
Digital Signal Processor ◽  
High Speed ◽  
High Performance ◽  
Input Pulse ◽  
Control Program ◽  
Sampling Technique ◽  
Digital Signal ◽  
Time Resolved ◽  
The Time Domain

The development of a high-speed digitizer system to measure time-domain voltage pulses in nanoseconds range is presented in this work. The digitizer design includes a high performance digital signal processor, a high-bandwidth analog-to-digital converter of flash-type, a set of delay lines, and a computer to achieve the time-domain measurements. A program running on the processor applies a time-equivalent sampling technique to acquire the input pulse. The processor communicates with the computer via a serial port RS-232 to receive commands and to transmit data. A control program written in LabVIEW 7.1 starts an acquisition routine in the processor. The program reads data from processor point by point in each occurrence of the signal, and plots each point to recover the time-resolved input pulse after n occurrences. The developed prototype is applied to measure fluorescence pulses from a homemade spectrometer. For this application, the LabVIEW program was improved to control the spectrometer, and to register and plot time-resolved fluorescence pulses produced by a substance. The developed digitizer has 750 MHz of analog input bandwidth, and it is able to resolve 2 ns rise-time pulses with 150 ps of resolution and a temporal error of 2.6 percent.

scholarly journals A time-domain Kirchhoff formula for the convective acoustic wave equation

2016 ◽  
Vol 472 (2187) ◽  
pp. 20150689 ◽  
Author(s):  
Ghader Ghorbaniasl ◽  
Leonidas Siozos-Rousoulis ◽  
Chris Lacor
Keyword(s):  
Wave Equation ◽  
Time Domain ◽  
High Speed ◽  
Flow Region ◽  
Medium Effect ◽  
Moving Frame ◽  
Moving Medium ◽  
Mean Flow ◽  
The Time Domain ◽  
Moving Media

Kirchhoff’s integral method allows propagated sound to be predicted, based on the pressure and its derivatives in time and space obtained on a data surface located in the linear flow region. Kirchhoff’s formula for noise prediction from high-speed rotors and propellers suffers from the limitation of the observer located in uniform flow, thus requiring an extension to arbitrarily moving media. This paper presents a Kirchhoff formulation for moving surfaces in a uniform moving medium of arbitrary configuration. First, the convective wave equation is derived in a moving frame, based on the generalized functions theory. The Kirchhoff formula is then obtained for moving surfaces in the time domain. The formula has a similar form to the Kirchhoff formulation for moving surfaces of Farassat and Myers, with the presence of additional terms owing to the moving medium effect. The equation explicitly accounts for the influence of mean flow and angle of attack on the radiated noise. The formula is verified by analytical cases of a monopole source located in a moving medium.

Ship Motion Computations Using a High Froude Number Time Domain Strip Theory

2006 ◽  
Vol 50 (01) ◽  
pp. 15-30
Author(s):  
D. S. Holloway ◽  
M. R. Davis
Keyword(s):  
Froude Number ◽  
Time Domain ◽  
High Speed ◽  
Equations Of Motion ◽  
Fluid Motion ◽  
Three Dimensional ◽  
Viscous Effects ◽  
Strip Theory ◽  
Large Amplitude Motions ◽  
The Time Domain

High-speed strip theories are discussed, and a time domain formulation making use of a fixed reference frame for the two-dimensional fluid motion is described in detail. This, and classical (low-speed) strip theory, are compared with the experimental results of Wellicome et al. (1995) up to a Froude number of 0.8, as well as with our own test data for a semi-SWATH, demonstrating the marked improvement of the predictions of the former at high speeds, while the need to account for modest viscous effects at these speeds is also argued. A significant contribution to time domain computations is a method of stabilizing the integration of the ship's equations of motion, which are inherently unstable due to feedback from implicit added mass components of the hydrodynamic force. The time domain high-speed theory is recommended as a practical alternative to three-dimensional methods. It also facilitates the investigation of large-amplitude motions with stern or bow emergence and forms a simulation base for the investigation of ride control systems and local or global loads.

A Comprehensive Review of Energy Efficient Content Addressable Memory Circuits for Network Applications

2016 ◽  
Vol 25 (04) ◽  
pp. 1630002 ◽  
Author(s):  
Syed Iftekhar Ali ◽  
Md Shafiqul Islam ◽  
Mohammad Rakibul Islam
Keyword(s):  
Energy Efficient ◽  
High Speed ◽  
Clock Cycle ◽  
Parallel Search ◽  
Search Activity ◽  
Content Addressable Memory ◽  
Network Applications ◽  
Reduction Techniques ◽  
Network Routers ◽  
Memory Circuits

Content addressable memory (CAM) can perform high-speed table look-up with bit level masking capability. This feature makes CAMs extremely attractive for high-speed packet forwarding and classification in network routers. High-speed look-up implies all the CAM word entries to be accessed and compared with a search word to find a suitable match in a single clock cycle. This parallel search activity requires large energy consumption which needs to be reduced. In this paper, a review of the energy reduction techniques of CAM is presented. A comparative study of some popular techniques has been made with the help of simulations carried out in this work and published results.

Sign in / Sign up

Export Citation Format

Share Document