Building IBIS-AMI Models for DDR5 Applications. Todd Westerhoff, SiSoft

Similar documents
A Beginner s Guide to SerDes and AMI Modeling. Todd Westerhoff, SiSoft Corey Mathis, MathWorks

IBIS-AMI: Assumptions, Terminology & Analytical Flows

IBIS-ATM Update: SerDes Modeling and IBIS

Understanding IBIS-AMI Simulations

SerDes Modeling: IBIS-ATM & Model Validation July 2007

Understanding IBIS-AMI Simulations

SerDes Modeling: Demonstrating IBIS-AMI Model Interoperability

UNDERSTANDING IBIS-AMI SIMULATIONS

Two for One: Leveraging SerDes Flows for AMI Model Development

Predicting BER with IBIS-AMI: experiences correlating SerDes simulations and measurement

SerDes Channel Simulation in FPGAs Using IBIS-AMI

Multi-Gigabit Serial Link Analysis using HSPICE and AMI Models

BIRD AMI Model New IBIS Support

Using IBIS-AMI in the Modeling of Advanced SerDes Equalization for Serial Link Simulation

Leveraging IBIS Capabilities for Multi-Gigabit Interfaces. Ken Willis - Cadence Design Systems Asian IBIS Summit, Shanghai, PRC November 13, 2017

Status Report IBIS 4.1 Macro Working Group

Explore your design space including IBIS AMI models with Advanced Channel Simulation

Proposal for modeling advanced SERDES

SPISim StatEye/AMI User s Guide

A New Method For Developing IBIS-AMI Models

Automated AMI Model Generation & Validation

Using DATA Files for IBIS-AMI Models

IBIS-AMI Modeling Using Scripts and Spice Models

Keysight Technologies IBIS-AMI Based Link Analysis of Realistic 56G PAM4 Channels

IBIS-AMI Model Simulations Over Six EDA Platforms

Agilent Technologies EZJIT and EZJIT Plus Jitter Analysis Software for Infiniium Series Oscilloscopes

IBIS-AMI Modeling and Simulation of 56G PAM4 Link Systems

Modeling and Verifying Mixed-Signal Designs with MATLAB and Simulink

Concerns when applying Channel Simulation to DDR4 Interface

ADS USB 3.1 Compliance Test Bench

Comparison of BER Estimation Methods which Account for Crosstalk

PCI Express 3.0 Characterization, Compliance, and Debug for Signal Integrity Engineers

Electrical optimization and simulation of your PCB design

An Innovative Simulation Workflow for Debugging High-Speed Digital Designs using Jitter Separation

10A AMI PARAMETER DEFINITION FILE STRUCTURE

Serial Link Analysis and PLL Model

PCI Express Link Equalization Testing 서동현

HFSS Solver On Demand for Package and PCB Characterization Using Cadence. Greg Pitner

Keysight U7231B, U7231C DDR3 and LPDDR3 Compliance Test Application For Infiniium Series Oscilloscopes DATA SHEET

An Overview of High-Speed Serial Bus Simulation Technologies

HFSS Solver-On-Demand for Package and PCB Characterization Using Cadence Greg Pitner

New AMI API to Resolve Model Parameter Dependencies

SPISim1. SPISim Modeling Suite. IBIS, IBIS-AMI model generation and general modeling

Memory Interface Verification and Debug. Analog Validation Presentation

Measurement and Simulation of a High- Speed Electro/Optical Channel

Application Note 1242

PCI Express 4.0. Electrical compliance test overview

The 3S Proposal: A SPICE Superset Specification for Behavioral Modeling

AN 608: HST Jitter and BER Estimator Tool for Stratix IV GX and GT Devices

Making the Most of your MATLAB Models to Improve Verification

Achieving PCI Express Compliance Faster

Simulation Results for 10 Gb/s Duobinary Signaling

DisplayPort 1.4 Webinar

Using DATA Files for IBIS-AMI Models. Lance Wang DesignCon IBIS Summit Santa Clara, CA, USA Feburary 3 rd, 2017

Connecting MATLAB & Simulink with your SystemVerilog Workflow for Functional Verification

HARI Interface Chip for Serial PMD A comparison of two clocking schemes. Vipul Bhatt, Finisar, 2/2/00 1

HyperLynx DDRx Interface Analysis. Student Workbook

N1014A SFF-8431 (SFP+)

CIRCUIT DESIGN. is published monthly by: UP Media Group Inc Powers Ferry Road, Ste. 600 Atlanta, GA Tel (678) Fax (678)

New Software-Designed Instruments

High-speed I/O test: The ATE paradigm must change

100GEL C2M Channel Analysis Update

Harmony-AMS Analog/Mixed-Signal Simulator

AMI Applications in High-speed Serial Channel Analysis and Measurement Correlation

LVDS applications, testing, and performance evaluation expand.

Software Driven Verification at SoC Level. Perspec System Verifier Overview

Application Note. PCIE-RA Series Final Inch Designs in PCI Express Applications Generation GT/s

FPGA Algorithm Development Using a Graphical Environment

PRISMTECH. Benchmarking OMG DDS for Large-scale Distributed Systems. Powering Netcentricity

Application Note. PCIE-EM Series Final Inch Designs in PCI Express Applications Generation GT/s

DDR Testing:Compliance,Verify and Debug( 一 )

Cadence simulation technology for PCB design

Allegro Sigrity SI Streamlining the creation of high-speed interconnect on digital PCBs and IC packages

VIBbox 64-Channel Sound & Vibration Solution

Keysight Technologies EZJIT Plus Jitter Analysis Software for Infiniium Oscilloscopes. Data Sheet

DT9837. High Performance USB Module for Sound & Vibration Analysis. DT9837 Features

Practical Importance of the FOUNDATION TM Fieldbus Interoperability Test System

Optimization of Modern Memory

IBIS Model Validation Review. Lance Wang IBIS Asian Summit 2012 Nov. 16, 2012 Yokohama, Japan

PCI Express Electrical Basics

Arria V GX Video Development System

Ethernet SFF-8431 SFP+ SFF-8635 QSFP+ Compliance and Debug Testing. Anshuman Bhat Product Manager

Portable Stimulus Working Group

Automated Requirements-Based Testing

How to Solve DDR Parametric and Protocol Measurement Challenges

Serial Digital Audio Routing Switchers?

Signal Integrity Analysis

DDR3 Memory Buffer: Buffer at the Heart of the LRDIMM Architecture. Paul Washkewicz Vice President Marketing, Inphi

APDS Wizard v4 for Nexxim/Designer 4.0 Specially optimized for DDR2/3 SI. Ansoft Korea Team

Multi-Gigahertz Parallel FFTs for FPGA and ASIC Implementation

Characterize and Debug Crosstalk Issues with Keysight Crosstalk Analysis App

Best Practices for Architecting Embedded Applications in LabVIEW Jacques Cilliers Applications Engineering

Experiment 3. Getting Start with Simulink

Features. Applications

Agilent U7231B DDR3 and LPDDR3 Compliance Test Application for Infiniium Series Oscilloscopes. Data Sheet

780A Handheld Test Instrument. 780A HANDHELD TEST INSTRUMENT Now you can test 4K Ultra HD HDMI /60Hz

What s New in HyperLynx 8.0

OrCAD for Education Program

IBISCHK6 V6.1.3 and Executable Model File Checking

Keysight U7233A DDR1 Compliance Test Application with LPDDR and mobile-ddr Support

Transcription:

Building IBIS-AMI Models for DDR5 Applications Todd Westerhoff, SiSoft

SPEAKERS Image Todd Westerhoff VP, Semiconductor Relations, SiSoft twesterh@sisoft.com www.sisoft.com Todd has over 37 years of experience in electronic system modeling and simulation, including 20 years in signal integrity. He is responsible for SiSoft's activities working with semiconductor vendors to develop high-quality simulation models and has been heavily involved with the IBIS- AMI modeling specification since its inception. He has held senior technical and management positions for Cisco and Cadence and worked as an independent signal integrity consultant. 2

IBIS-AMI Models Goals Interoperable: different vendor models work together Portable: one model runs in multiple simulators Flexible: supports Statistical and Time-Domain simulation High Performance: simulates a million bits per CPU minute Accurate: high correlation to simulations / measurement Secure: represents IP behavior without exposing internal details Assumptions Analog I/O operates in linear region High impedance node between analog I/O & equalization circuitry Equalization behavior modeled by code linked into simulator Equalization models use AMI API AMI 3

IBIS-AMI Modeling Challenges AMI Algorithmic models: o Compiled objects linked into EDA simulator at runtime o Interface defined in C language Skills required: o IBIS-AMI specification o C/C++ software programming o Communications analysis o Analog circuit design / modeling o Digital signal processing o Control loop theory You shouldn t have to be a wizard to create an IBIS-AMI model! 4

AMI Model Development Goals Interactive development environment o High level modeling language not C/C++ o Library of configurable building blocks / examples o User extensible; can add new model controls and algorithms o Import test case data from real world (customer) topologies o No need to run compilers from the command line Faithfully model data path and control algorithms Portable, compliant IBIS-AMI models 5

Wouldn t It Be Great If When AMI models weren t working properly in your favorite channel simulator 6

Wouldn t It Be Great If You could push into an AMI model to see how it was constructed 7

Wouldn t It Be Great If and edit MATLAB code for the different model blocks 8

Wouldn t It Be Great If then rerun your simulation using updated models? 9

SiSoft / MathWorks Integration QCD/QSI Simulink/MATLAB Integrated AMI channel simulation and AMI model development 10

AMI Model Development Flow Simulink / MATLAB Generate test case data Interactive design & debug IBIS-AMI model creation Tight analysis loop speeds three critical phases of AMI development o Test case definition & generation Identify cases of interest Done? Regression testing o Interactive algorithmic development o Regression testing / issue identification Quantum-SI IBIS-AMI validation & distribution 11

Quantum-SI: Test Case Definition & Generation Circuit topology Data rate & simulation setup Different test setups Driver strength / terminations TX/RX EQ settings 12

AMI Modeling Development Environment Simulink o General purpose, time-domain modeling / simulation environment o Libraries of sources / sinks / building blocks o Integrated MATLAB support with interactive debugging o C code generation with embedded support for IBIS-AMI API 13

AMI Case 2: TX: Init-only RX: Getwave-only Configure Simulink to act like an AMI simulator h AC (t) Bit stream b(t) o Fixed time step analysis TX AMI_Init TX AMI_Init o Static ( Init ) TX equalization h AC (t) h TE (t) h AC (t) h TE (t) o o Dynamic ( GetWave ) RX equalization Clock recovery performed by RX model Bit stream b(t) g RE [ b(t) h AC (t) h TE (t) ] h AC (t) g RE [ b(t) h AC (t) h TE (t) ] AMI models are generated from TX/RX subsystems in the Simulink model RX AMI_Getwave Post-processing Time-Domain Flow Static TX EQ, Dynamic RX EQ RX AMI_Getwave Post-processing Time-Domain Flow Static TX EQ, Dynamic RX EQ AMI Specification Simulink Setup 14

Simulink AMI Model Development Setup Channel Impulse Response Stimulus TX ( Init ) RX ( GetWave ) Replicates the IBIS-AMI Tx_Init / Rx_Getwave analysis flow 15

Importing Test Case Data QSI Simulink Channel model Data rate Stimulus TX/RX EQ settings Simulation setup 16

AMI Algorithm Development Start / stop / step simulation Waveform & scope displays MATLAB code debugger Examine / modify model variables Stop on breakpoint or user-defined condition Modify algorithms & rerun 17

AMI Code Generation / Compilation C code automatically generated & compiled Create different models for internal / external use Models conform to IBIS-AMI API 18

Comparing Simulation Results Simulink Waveforms Eyes DFE Taps QCD/QSI 19

Automated AMI Regression Testing 1. Setup tests in QSI 2. Simulations run in parallel 3. Metrics & plots presented in SiViewer Setup & run 1,000 s of simulations using QSI s solution space QSI extracts performance metrics and compliance pass/fail criteria Scalable high performance simulation o Compiled models run 5-7x faster o Run simulations in parallel o Speedups >100x are possible 20

Unique DDR5 Modeling Needs IBIS-AMI was originally developed for lossy, differential serial channel applications DDR5 presents new AMI modeling challenges o Single-ended signaling o DRAM DFE training performed by controller o Clock forwarding vs. Rx clock recovery o Representing I/O nonlinearity, power and noise 21

Single-Ended AMI This is a tricky one! o Because AMI simulations use an impulse response for the channel, DC bias information technically gets lost during AMI model processing o One way to address this is to extract the DC bias from the analog channel analysis and add it back after EQ processing This approach allows us to directly leverage existing AMI algorithms, many of which have the expectation of differential signaling baked in to the code. 22

Modeling DRAM DFE Training DDR5 controllers set DRAM DFE taps by training at system startup Modeling training sequences with AMI models is difficult & time consuming Manually optimizing DRAM DFE settings through simulation is difficult & time consuming Approach: let DRAM DFE taps adapt, then lock settings when Ignore_Bits is reached 23

Modeling Clock Forwarding DDR5 transfers use DQS signals to sample the incoming data Traditional AMI models recover the sampling clock from the incoming data stream AMI provides no facility for a separate clock input to the algorithmic model Approach: model the CDR in the algorithmic model as usual, lock adaptation as Ignore_Bits is reached, use Rx jitter parameters to include sampling clock noise 24

AMI Model Development Flow 1. Define test cases: Topologies, data rates, transfers, terminations, TX/RX equalization & simulation settings. 2. Load a test case into AMI development environment. 3. Design & debug TX/RX modeling algorithms. 6. Run regression tests to validate compiled models & identify issues. 4. Create compiled AMI models. 5. Run compiled models with original test cases. 25

Portable, Compliant Models Eye Diagram Output Clock Output Control Inputs Performance Test Simulation Time Reference Time Relative Speed Statistical 1 sec 1 sec 1.000x TimeDomain_008spb 1.46 min/mbit 1.30 min/mbit 0.889x TimeDomain_016spb 1.87 min/mbit 1.46 min/mbit 0.783x TimeDomain_032spb 3.33 min/mbit 2.60 min/mbit 0.781x TimeDomain_064spb 7.97 min/mbit 6.02 min/mbit 0.755x TimeDomain_128spb 27.97 min/mbit 22.77 min/mbit 0.814x Simulation Speed Compliance - Samples/Bit Compliance - Block Size Validation tests show generated AMI models work as well as models created by hand 26

Thank You! --- QUESTIONS? 27