Time-Domain Segmentation based Massively Parallel Simulation Bichen Wu Dept. Micro/nano electronics Tsinghua Univ., Beijing, China

Slides:



Advertisements
Similar presentations
DCSP-3: Fourier Transform Jianfeng Feng Department of Computer Science Warwick Univ., UK
Advertisements

ECE 8443 – Pattern Recognition EE 3512 – Signals: Continuous and Discrete Objectives: Response to a Sinusoidal Input Frequency Analysis of an RC Circuit.
Lecture 7: Basis Functions & Fourier Series
Copyright 2001, Agrawal & BushnellVLSI Test: Lecture 181 Lecture 18 DSP-Based Analog Circuit Testing  Definitions  Unit Test Period (UTP)  Correlation.
Analog-to-Digital Converter (ADC) And
Lecture 17: Analog to Digital Converters Lecturers: Professor John Devlin Mr Robert Ross.
Digital Control Systems INTRODUCTION. Introduction What is a control system? Objective: To make the system OUTPUT and the desired REFERENCE as close as.
Leo Lam © Signals and Systems EE235. Transformers Leo Lam ©
Statistical properties of Random time series (“noise”)
DNC, GEC & Non-linear Interpolation DNC, GEC & Non-linear interpolation A Review of ”A Digitally Enhanced 1.8V 15-bit 40-MSample/s CMOS.
Real time DSP Professors: Eng. Julian S. Bruno Eng. Jerónimo F. Atencio Sr. Lucio Martinez.
Analog-to-Digital Converters Prepared by: Mohammed Al-Ghamdi, Mohammed Al-Alawi,
Introduction to Analog-to-Digital Converters
Name: Dr. Peter Tsang Room: G6505 Ext: 7763
Lecture 9: Fourier Transform Properties and Examples
04/26/2006VLSI Design & Test Seminar Series 1 Phase Delay in MAC-based Analog Functional Testing in Mixed-Signal Systems Jie Qin, Charles Stroud, and Foster.
Leo Lam © Signals and Systems EE235. Leo Lam © Fourier Transform Q: What did the Fourier transform of the arbitrary signal say to.
Department of Electrical & Computer Engineering 1 ES585a - Computer Based Power System Protection Course by Dr.T.S.Sidhu - Fall 2005 Class discussion presentation.
Over-Sampling and Multi-Rate DSP Systems
11 Lecture Slides ME 3222 Kinematics and Control Lab Lab 2 AD DA and Sampling Theory By Dr. Debao Zhou.
Frequency Domain Representation of Sinusoids: Continuous Time Consider a sinusoid in continuous time: Frequency Domain Representation: magnitude phase.
The sampling of continuous-time signals is an important topic It is required by many important technologies such as: Digital Communication Systems ( Wireless.
ECE 8443 – Pattern Recognition EE 3512 – Signals: Continuous and Discrete Objectives: First-Order Second-Order N th -Order Computation of the Output Signal.
Data Converters ELEC 330 Digital Systems Engineering Dr. Ron Hayne
Electronics Involves the use of devices and circuits to control the flow of electric current to achieve some purpose. These circuits contain: Resistors,
DOLPHIN INTEGRATION TAMES-2 workshop 23/05/2004 Corsica1 Behavioural Error Injection, Spectral Analysis and Error Detection for a 4 th order Single-loop.
1 Process-Variation Tolerant Design Techniques for Multiphase Clock Generation Manohar Nagaraju +, Wei Wu*, Cameron Charles # + University of Washington,
By Grégory Brillant Background calibration techniques for multistage pipelined ADCs with digital redundancy.
1HSSPG Georgia Tech High Speed Image Acquisition System for Focal-Plane-Arrays Doctoral Dissertation Presentation by Youngjoong Joo School of Electrical.
Digital Signal Processing
Module 2 SPECTRAL ANALYSIS OF COMMUNICATION SIGNAL.
Han Q Le© ECE 3336 Introduction to Circuits & Electronics Lecture Set #9 Linear Circuit-Fourier-Digital Dr. Han Le ECE Dept.
ECE 8443 – Pattern Recognition EE 3512 – Signals: Continuous and Discrete Objectives: Review Resources: Wiki: Superheterodyne Receivers RE: Superheterodyne.
Lecture 24: CT Fourier Transform
Fourier Series. Introduction Decompose a periodic input signal into primitive periodic components. A periodic sequence T2T3T t f(t)f(t)
The World Leader in High-Performance Signal Processing Solutions Design a Clock Distribution for a WCDMA Transceiver System CSNDSP 2006 Session: B.11 Systems.
Ring Oscillator in Switched Capacitor Feedback
Fourier Analysis of Discrete-Time Systems
Introduction Advantage of DSP: - Better signal quality & repeatable performance - Flexible  Easily modified (Software Base) - Handle more complex processing.
Representing Numerical Data Analog Any signal that varies continuously over time Mechanical Pneumatic Hydraulic Electrical Digital Quantities are represented.
Digital Phase Control System for SSRF LINAC C.X. Yin, D.K. Liu, L.Y. Yu SINAP, China
12/14/2010Sophia University Solid –State Circuits & Devices Laboratory 1 A low-power delta-sigma modulator using dynamic-source-follower integrators Ryoto.
THERMAL NOISE ESTIMATION IN SWITCHED-CAPACITOR CIRCUITS
Figure Analog-to-digital conversion.. Figure The DAC output is a staircase approximation to the original signal. Filtering removes the sharp.
ECE 2799 Electrical and Computer Engineering Design ANALOG to DIGITAL CONVERSION Prof. Bitar Last Update:
Linear Constant-Coefficient Difference Equations
Modelling and Simulation of Passive Optical Devices João Geraldo P. T. dos Reis and Henrique J. A. da Silva Introduction Integrated Optics is a field of.
Fourier Representation of Signals and LTI Systems.
Gustavo Cancelo Analysis of the phase shift error between A and B signals in BPMs BPM project.
1 Fourier Representation of Signals and LTI Systems. CHAPTER 3 School of Computer and Communication Engineering, UniMAP Amir Razif B. Jamil Abdullah EKT.
ΨΗΦΙΑΚΟΣ ΕΛΕΓΧΟΣ (22Δ802) Β΄ ΕΞΑΜΗΝΟ Καθηγητής Πέτρος Π. Γρουμπός  Ώρες Γραφείου: Τετάρτη Πέμπτη Παρασκευή 11:00- 12:00 Γραφείο: 1.
Math for CS Fourier Transforms
Real-time Digital Signal Processing Digital Filters.
Chapter 4 Dynamical Behavior of Processes Homework 6 Construct an s-Function model of the interacting tank-in-series system and compare its simulation.
Linear Constant-Coefficient Difference Equations
Chapter 4 Dynamical Behavior of Processes Homework 6 Construct an s-Function model of the interacting tank-in-series system and compare its simulation.
B.Sc. Thesis by Çağrı Gürleyük
Mixed-Digital/Analog Simulation and Modeling Research
A Readout Electronics System for GEM Detectors
Digital Control Systems
Oversampling A/D Conversion
The sampling of continuous-time signals is an important topic
Quiz: Introduction to SAR ADC Component Selection TIPL 4401 TI Precision Labs – ADCs Hello, and welcome to the TI Precision Lab discussing input offset.
topics Basic Transmission Line Equations
4.1 DFT In practice the Fourier components of data are obtained by digital computation rather than by analog processing. The analog values have to be.
Advanced Digital Signal Processing
Conversation between Analogue and Digital System
數位控制理論簡介.
Quiz: Introduction to SAR ADC Component Selection TIPL 4401 TI Precision Labs – ADCs Hello, and welcome to the TI Precision Lab discussing input offset.
Presentation transcript:

Time-Domain Segmentation based Massively Parallel Simulation Bichen Wu Dept. Micro/nano electronics Tsinghua Univ., Beijing, China Mobile:

Background On the one hand, transient circuit simulation is time consuming Sigma-Delta ADC: 71h SAR ADC: 41h FLASH ADC: 38h PLL: 148h On the other hand, we got strong parallel computing resource Many core, distributed computing How can we increase the circuit simulation efficiency with parallel computing?

Circuit simulation  Transient circuit simulation is to solve the initial condition problem of the following differential equation  Backward Euler was used to convert the above to sequence of discrete equations

Previous parallel circuit simulation  Domain decomposition A New Partitioning Method for Parallel Simulation of VLSI Circuits on Transistor Level ( 1996 )  Parallel numerical integral method WavePipe: Parallel Transient Simulation of Analog and Digital Circuits on Multi-Core Shared-Memory Machines(DAC 2008)  Multi-algorithm parallel Hierarchical Multi-algorithm Parallel Circuit Simulation (TCAD 2011)

Time domain segmentation based parallel simulation  Method description t0t0 t end t0t0 t1t1 t2t2

Time domain segmentation based parallel simulation  In general, solution of equation is dependent on initial condition  Uncertainty brought by time domain segmentation may leads to erroneous result

Short memory assumption  Short memory : the state of the circuit at a certain time point is correlated only to the a limited history of the the input as well as the internal state of the circuit.  Circuits such as ADC , possess short memory property A D Analog input Digital output

Calculating SNDR through transient simulation for ADC A D Analog input Digital output Fourier Transform Power spectrum density

Two examples 5bits FLASH ADC F sam =3.75GHz, T total =2us, T ov =6ns Time for serial simulation : 38h Number of segmentation : 1 , 2 , 4 , 8 , 100 6bits SAR ADC F sam =2GHz, T total =8us, T ov =13.3ns Time for serial simulation : 41h Number of segmentation : 1 , 2 , 4 , 8 , 100

Transient simulation for ADC Power spectrum density of FLASH ADC(left) and SAR ADC ( right ) output, with serial and parallel simulation FLASHSAR

Transient simulation for ADC  Result

Transient simulation for Sigma-Delta ADC  Sigma-Delta modulator Because of integrator, it’s not short memory

Transient simulation for Sigma-Delta ADC  Direct segmentation with the modulator output of modulator brings erroneous result Time domain deviation between serial and parallel result Frequency domain deviation between serial and parallel result

Transient simulation for Sigma-Delta ADC  Uncertainty brought by time domain segmentation

Transient simulation for Sigma-Delta ADC Digital filter realized in matlab  Solution:

Transient simulation for Sigma-Delta ADC  Result showed efficiency enhancement with high accuracy

Transient simulation for PLL  Phase lock loop(PLL) is not short memory circuit  Self-driven system  Integral property f[ ] h(t) + -

Transient simulation for PLL  Phase lock loop(PLL) The expression of the phase growth in the equation above is as follow. Let So we get:

Transient simulation for PLL  Phase lock loop(PLL)

Transient noise analysis  PSD of 1/f noise generated in serial and parallel Parallel: Serial The bandwidth gets wider

Transient noise analysis  1/f noise in TDSM serial parallel

Summary  We proposed TDSM for transient circuit simulation  Testified on FLASH , SAR , Sigma-Delta ADC  Maximum 78X acceleration with 100 cores parallelization  Efficient for phase noise simulation for PLL, testified by behavior model.  Proposed 1/f noise generator that support TDSM, testified on a single transistor  Now submitted to DAC 2013