External Memory Interfaces Intel® Agilex™ FPGA IP User Guide

Download PDF
ID 683216
Date 3/28/2022
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents
Document Table of Contents x
1. About the External Memory Interfaces Intel® Agilex™ FPGA IP 2. Intel® Agilex™ FPGA EMIF IP – Introduction 3. Intel® Agilex™ FPGA EMIF IP – Product Architecture 4. Intel® Agilex™ FPGA EMIF IP – End-User Signals 5. Intel® Agilex™ FPGA EMIF IP – Simulating Memory IP 6. Intel® Agilex™ FPGA EMIF IP – DDR4 Support 7. Intel® Agilex™ FPGA EMIF IP – QDR-IV Support 8. Intel® Agilex™ FPGA EMIF IP – Timing Closure 9. Intel® Agilex™ FPGA EMIF IP – I/O Timing Closure 10. Intel® Agilex™ FPGA EMIF IP – Controller Optimization 11. Intel® Agilex™ FPGA EMIF IP – Debugging 12. External Memory Interfaces Intel® Agilex™ FPGA IP User Guide Archives 13. Document Revision History for External Memory Interfaces Intel® Agilex™ FPGA IP User Guide
1. About the External Memory Interfaces Intel® Agilex™ FPGA IP x
1.1. Release Information
2. Intel® Agilex™ FPGA EMIF IP – Introduction x
2.1. Intel® Agilex™ EMIF IP Protocol and Feature Support 2.2. Intel® Agilex™ EMIF IP Design Flow 2.3. Intel® Agilex™ EMIF IP Design Checklist
3. Intel® Agilex™ FPGA EMIF IP – Product Architecture x
3.1. Intel® Agilex™ EMIF Architecture: Introduction 3.2. Intel® Agilex™ EMIF Sequencer 3.3. Intel® Agilex™ EMIF Calibration 3.4. Intel® Agilex™ EMIF Controller 3.5. User-requested Reset in Intel® Agilex™ EMIF IP 3.6. Intel® Agilex™ EMIF for Hard Processor Subsystem 3.7. Using a Custom Controller with the Hard PHY
3.1. Intel® Agilex™ EMIF Architecture: Introduction x
3.1.1. Intel® Agilex™ EMIF Architecture: I/O Subsystem 3.1.2. Intel® Agilex™ EMIF Architecture: I/O SSM 3.1.3. Intel® Agilex™ EMIF Architecture: I/O Bank 3.1.4. Intel® Agilex™ EMIF Architecture: I/O Lane 3.1.5. Intel® Agilex™ EMIF Architecture: Input DQS Clock Tree 3.1.6. Intel® Agilex™ EMIF Architecture: PHY Clock Tree 3.1.7. Intel® Agilex™ EMIF Architecture: PLL Reference Clock Networks 3.1.8. Intel® Agilex™ EMIF Architecture: Clock Phase Alignment
3.1.4. Intel® Agilex™ EMIF Architecture: I/O Lane x
3.1.4.1. Considerations for Designing DDR4 x72 Interface Together with an AVST x8/x16/x32 Configuration Scheme
3.3. Intel® Agilex™ EMIF Calibration x
3.3.1. Intel® Agilex™ Calibration Stages 3.3.2. Intel® Agilex™ Calibration Stages Descriptions 3.3.3. Intel® Agilex™ Calibration Flowchart 3.3.4. Intel® Agilex™ Calibration Algorithms
3.3.4. Intel® Agilex™ Calibration Algorithms x
3.3.4.1. Calibration Algorithms for DDR4 3.3.4.2. Calibration Algorithms for QDR-IV 3.3.4.3. Guidelines for Debugging Calibration Issues
3.3.4.1. Calibration Algorithms for DDR4 x
3.3.4.1.1. DDR4 Read Calibration 3.3.4.1.2. DDR4 Write Calibration
3.3.4.2. Calibration Algorithms for QDR-IV x
3.3.4.2.1. QDR-IV Read Calibration 3.3.4.2.2. QDR-IV Write Calibration
3.3.4.3. Guidelines for Debugging Calibration Issues x
3.3.4.3.1. Debugging Calibration Failure Using Information from the Calibration report 3.3.4.3.2. Debugging Address and Command Leveling Calibration Failure 3.3.4.3.3. Debugging Address and Command Deskew Failure 3.3.4.3.4. Debugging DQS Enable Failure 3.3.4.3.5. Debugging Read Deskew Calibration Failure 3.3.4.3.6. Debugging VREFIN Calibration Failure 3.3.4.3.7. Debugging LFIFO Calibration Failure 3.3.4.3.8. Debugging Write Leveling Failure 3.3.4.3.9. Debugging Write Deskew Calibration Failure 3.3.4.3.10. Debugging VREFOUT Calibration Failure
3.4. Intel® Agilex™ EMIF Controller x
3.4.1. Hard Memory Controller 3.4.2. Intel® Agilex™ Hard Memory Controller Rate Conversion Feature
3.4.1. Hard Memory Controller x
3.4.1.1. Hard Memory Controller Features 3.4.1.2. Hard Memory Controller Main Control Path 3.4.1.3. Data Buffer Controller
3.6. Intel® Agilex™ EMIF for Hard Processor Subsystem x
3.6.1. Restrictions on I/O Bank Usage for Intel® Agilex™ EMIF IP with HPS
4. Intel® Agilex™ FPGA EMIF IP – End-User Signals x
4.1. Intel® Agilex™ EMIF IP Interface and Signal Descriptions 4.2. Intel® Agilex™ EMIF IP AFI Signals 4.3. Intel® Agilex™ EMIF IP AFI 4.0 Timing Diagrams 4.4. Intel® Agilex™ EMIF IP Memory Mapped Register (MMR) Tables
4.1. Intel® Agilex™ EMIF IP Interface and Signal Descriptions x
4.1.1. Intel Agilex EMIF IP Interfaces for DDR4 4.1.2. Intel Agilex EMIF IP Interfaces for QDR-IV
4.1.1. Intel Agilex EMIF IP Interfaces for DDR4 x
4.1.1.1. local_reset_req for DDR4 4.1.1.2. local_reset_status for DDR4 4.1.1.3. pll_ref_clk for DDR4 4.1.1.4. pll_locked for DDR4 4.1.1.5. ac_parity_err for DDR4 4.1.1.6. oct for DDR4 4.1.1.7. mem for DDR4 4.1.1.8. status for DDR4 4.1.1.9. afi_reset_n for DDR4 4.1.1.10. afi_clk for DDR4 4.1.1.11. afi_half_clk for DDR4 4.1.1.12. afi for DDR4 4.1.1.13. emif_usr_reset_n for DDR4 4.1.1.14. emif_usr_clk for DDR4 4.1.1.15. ctrl_amm for DDR4 4.1.1.16. ctrl_amm_aux for DDR4 4.1.1.17. ctrl_auto_precharge for DDR4 4.1.1.18. ctrl_user_priority for DDR4 4.1.1.19. ctrl_ecc_user_interrupt for DDR4 4.1.1.20. ctrl_ecc_readdataerror for DDR4 4.1.1.21. ctrl_ecc_status for DDR4 4.1.1.22. ctrl_mmr_slave for DDR4 4.1.1.23. hps_emif for DDR4 4.1.1.24. emif_calbus for DDR4 4.1.1.25. emif_calbus_clk for DDR4
4.1.2. Intel Agilex EMIF IP Interfaces for QDR-IV x
4.1.2.1. local_reset_req for QDR-IV 4.1.2.2. local_reset_status for QDR-IV 4.1.2.3. pll_ref_clk for QDR-IV 4.1.2.4. pll_locked for QDR-IV 4.1.2.5. oct for QDR-IV 4.1.2.6. mem for QDR-IV 4.1.2.7. status for QDR-IV 4.1.2.8. afi_reset_n for QDR-IV 4.1.2.9. afi_clk for QDR-IV 4.1.2.10. afi_half_clk for QDR-IV 4.1.2.11. afi for QDR-IV 4.1.2.12. emif_usr_reset_n for QDR-IV 4.1.2.13. emif_usr_clk for QDR-IV 4.1.2.14. ctrl_amm for QDR-IV 4.1.2.15. emif_calbus for QDR-IV 4.1.2.16. emif_calbus_clk for QDR-IV
4.2. Intel® Agilex™ EMIF IP AFI Signals x
4.2.1. AFI Clock and Reset Signals 4.2.2. AFI Address and Command Signals 4.2.3. AFI Write Data Signals 4.2.4. AFI Read Data Signals 4.2.5. AFI Calibration Status Signals 4.2.6. AFI Tracking Management Signals 4.2.7. AFI Shadow Register Management Signals
4.3. Intel® Agilex™ EMIF IP AFI 4.0 Timing Diagrams x
4.3.1. AFI Address and Command Timing Diagrams 4.3.2. AFI Write Sequence Timing Diagrams 4.3.3. AFI Read Sequence Timing Diagrams 4.3.4. AFI Calibration Status Timing Diagram
4.4. Intel® Agilex™ EMIF IP Memory Mapped Register (MMR) Tables x
4.4.1. ctrlcfg0 4.4.2. ctrlcfg1 4.4.3. dramtiming0 4.4.4. sbcfg1 4.4.5. caltiming0 4.4.6. caltiming1 4.4.7. caltiming2 4.4.8. caltiming3 4.4.9. caltiming4 4.4.10. caltiming9 4.4.11. dramaddrw 4.4.12. sideband0 4.4.13. sideband1 4.4.14. sideband4 4.4.15. sideband6 4.4.16. sideband7 4.4.17. sideband9 4.4.18. sideband11 4.4.19. sideband12 4.4.20. sideband13 4.4.21. sideband14 4.4.22. dramsts 4.4.23. niosreserve0 4.4.24. niosreserve1 4.4.25. sideband16 4.4.26. ecc3: ECC Error and Interrupt Configuration 4.4.27. ecc4: Status and Error Information 4.4.28. ecc5: Address of Most Recent SBE/DBE 4.4.29. ecc6: Address of Most Recent Correction Command Dropped 4.4.30. ecc7: Extension for Address of Most Recent SBE/DBE 4.4.31. ecc8: Extension for Address of Most Recent Correction Command Dropped
5. Intel® Agilex™ FPGA EMIF IP – Simulating Memory IP x
5.1. Simulation Options 5.2. Simulation Walkthrough
5.2. Simulation Walkthrough x
5.2.1. Calibration Modes 5.2.2. Simulation Scripts 5.2.3. Functional Simulation with Verilog HDL 5.2.4. Functional Simulation with VHDL 5.2.5. Simulating the Design Example
6. Intel® Agilex™ FPGA EMIF IP – DDR4 Support x
6.1. Intel® Agilex™ FPGA EMIF IP Parameter Descriptions 6.2. Intel® Agilex™ External Memory Interfaces Intel® Calibration IP Parameters 6.3. Register Map IP-XACT Support for Intel® Agilex™ EMIF DDR4 IP 6.4. Intel® Agilex™ FPGA EMIF IP Pin and Resource Planning 6.5. DDR4 Board Design Guidelines
6.1. Intel® Agilex™ FPGA EMIF IP Parameter Descriptions x
6.1.1. Intel Agilex EMIF IP DDR4 Parameters: General 6.1.2. Intel Agilex EMIF IP DDR4 Parameters: Memory 6.1.3. Intel Agilex EMIF IP DDR4 Parameters: Mem I/O 6.1.4. Intel Agilex EMIF IP DDR4 Parameters: FPGA I/O 6.1.5. Intel Agilex EMIF IP DDR4 Parameters: Mem Timing 6.1.6. Intel Agilex EMIF IP DDR4 Parameters: Controller 6.1.7. Intel Agilex EMIF IP DDR4 Parameters: Diagnostics 6.1.8. Intel Agilex EMIF IP DDR4 Parameters: Example Designs
6.4. Intel® Agilex™ FPGA EMIF IP Pin and Resource Planning x
6.4.1. Intel® Agilex™ FPGA EMIF IP Interface Pins 6.4.2. Intel® Agilex™ FPGA EMIF IP Resources 6.4.3. Pin Guidelines for Intel® Agilex™ FPGA EMIF IP
6.4.1. Intel® Agilex™ FPGA EMIF IP Interface Pins x
6.4.1.1. Estimating Pin Requirements 6.4.1.2. DIMM Options 6.4.1.3. Maximum Number of Interfaces
6.4.2. Intel® Agilex™ FPGA EMIF IP Resources x
6.4.2.1. OCT 6.4.2.2. PLL
6.4.3. Pin Guidelines for Intel® Agilex™ FPGA EMIF IP x
6.4.3.1. Intel® Agilex™ FPGA EMIF IP Banks 6.4.3.2. General Guidelines 6.4.3.3. x4 DIMM Implementation 6.4.3.4. Specific Pin Connection Requirements 6.4.3.5. Command and Address Signals 6.4.3.6. Clock Signals 6.4.3.7. Data, Data Strobes, DM/DBI, and Optional ECC Signals
6.5. DDR4 Board Design Guidelines x
6.5.1. Terminations for DDR4 with Intel® Agilex™ Devices 6.5.2. Clamshell Topology 6.5.3. General Layout Routing Guidelines 6.5.4. Reference Stackup 6.5.5. Intel® Agilex™ EMIF-Specific Routing Guidelines for Various DDR4 Topologies 6.5.6. DDR4 Routing Guidelines: Discrete (Component) Topologies 6.5.7. Intel® Agilex™ EMIF Pin Swapping Guidelines
6.5.1. Terminations for DDR4 with Intel® Agilex™ Devices x
6.5.1.1. Dynamic On-Chip Termination (OCT) 6.5.1.2. Dynamic On-Die Termination (ODT) in DDR4 6.5.1.3. Choosing Terminations on Intel® Agilex™ FPGA Devices 6.5.1.4. On-Chip Termination Recommendations for Intel® Agilex™ FPGA Devices
6.5.5. Intel® Agilex™ EMIF-Specific Routing Guidelines for Various DDR4 Topologies x
6.5.5.1. One DIMM per Channel (1DPC) for UDIMM, RDIMM, LRDIMM, and SODIMM DDR4 Topologies 6.5.5.2. Two DIMMs per Channel (2DPC) for UDIMM, RDIMM, and LRDIMM DDR4 Topologies 6.5.5.3. Two DIMMs per Channel (2DPC) for SODIMM Topology 6.5.5.4. Skew Matching Guidelines for DIMM Configurations 6.5.5.5. Power Delivery Recommendations for the Memory / DIMM Side
6.5.6. DDR4 Routing Guidelines: Discrete (Component) Topologies x
6.5.6.1. Single Rank x 8 Discrete (Component) Topology 6.5.6.2. Single Rank x 16 Discrete (Component) Topology 6.5.6.3. ADDR/CMD Reference Voltage/RESET Signal Routing Guidelines for Single Rank x 8 and R Rank x 16 Discrete (Component) Topologies 6.5.6.4. Skew Matching Guidelines for DDR4 Discrete Configurations 6.5.6.5. Power Delivery Recommendations for DDR4 Discrete Configurations
7. Intel® Agilex™ FPGA EMIF IP – QDR-IV Support x
7.1. Intel® Agilex™ FPGA EMIF IP Parameter Descriptions 7.2. Intel® Agilex™ External Memory Interfaces Intel® Calibration IP Parameters 7.3. Intel® Agilex™ FPGA EMIF IP Pin and Resource Planning 7.4. QDR-IV Board Design Guidelines
7.1. Intel® Agilex™ FPGA EMIF IP Parameter Descriptions x
7.1.1. Intel Agilex EMIF IP QDR-IV Parameters: General 7.1.2. Intel Agilex EMIF IP QDR-IV Parameters: Memory 7.1.3. Intel Agilex EMIF IP QDR-IV Parameters: FPGA I/O 7.1.4. Intel Agilex EMIF IP QDR-IV Parameters: Mem Timing 7.1.5. Intel Agilex EMIF IP QDR-IV Parameters: Controller 7.1.6. Intel Agilex EMIF IP QDR-IV Parameters: Diagnostics 7.1.7. Intel Agilex EMIF IP QDR-IV Parameters: Example Designs
7.3. Intel® Agilex™ FPGA EMIF IP Pin and Resource Planning x
7.3.1. Intel® Agilex™ FPGA EMIF IP Interface Pins 7.3.2. Intel® Agilex™ FPGA EMIF IP Resources 7.3.3. Pin Guidelines for Intel® Agilex™ FPGA EMIF IP
7.3.1. Intel® Agilex™ FPGA EMIF IP Interface Pins x
7.3.1.1. Estimating Pin Requirements 7.3.1.2. Maximum Number of Interfaces
7.3.2. Intel® Agilex™ FPGA EMIF IP Resources x
7.3.2.1. OCT 7.3.2.2. PLL
7.3.3. Pin Guidelines for Intel® Agilex™ FPGA EMIF IP x
7.3.3.1. Intel® Agilex™ FPGA EMIF IP Banks 7.3.3.2. General Guidelines 7.3.3.3. QDR IV SRAM Commands and Addresses, AP, and AINV Signals 7.3.3.4. QDR IV SRAM Clock Signals 7.3.3.5. QDR IV SRAM Data, DINV, and QVLD Signals 7.3.3.6. Specific Pin Connection Requirements 7.3.3.7. Resource Sharing Guidelines (Multiple Interfaces)
7.4. QDR-IV Board Design Guidelines x
7.4.1. General Layout Routing Guidelines 7.4.2. Reference Stackup 7.4.3. Routing Guidelines for QDR-IV Topology 7.4.4. Intel® Agilex™ EMIF Design Guideline Summary
7.4.3. Routing Guidelines for QDR-IV Topology x
7.4.3.1. QDR-IV Single Device Memory Topology 7.4.3.2. Skew Matching Guidelines for QDR-IV Configurations 7.4.3.3. Power Delivery Recommendation for QDR-IV Configurations
8. Intel® Agilex™ FPGA EMIF IP – Timing Closure x
8.1. Timing Closure 8.2. Optimizing Timing
8.1. Timing Closure x
8.1.1. Timing Analysis
8.1.1. Timing Analysis x
8.1.1.1. PHY or Core
9. Intel® Agilex™ FPGA EMIF IP – I/O Timing Closure x
9.1. I/O Timing Closure Overview 9.2. Collateral Generated with Your EMIF IP 9.3. SPICE Decks 9.4. File Organization 9.5. Top-level Parameterization File 9.6. IP-Supplied Parameters that You Might Need to Override 9.7. Understanding the *_ip_parameters.dat File and Making a Mask Polygon 9.8. Multi-Rank Topology 9.9. Pin Parasitics 9.10. Mask Evaluation
9.3. SPICE Decks x
9.3.1. Address/Command Simulation Deck FPGA PCB — 12-Line Channel Model Multi-Rank and Connector Model DIMM/Fly-By + VTT Model 9.3.2. FPGA Write Operation Simulation Deck 9.3.3. FPGA Read Operation Simulation Deck
10. Intel® Agilex™ FPGA EMIF IP – Controller Optimization x
10.1. Interface Standard 10.2. Bank Management Efficiency 10.3. Data Transfer 10.4. Improving Controller Efficiency
10.4. Improving Controller Efficiency x
10.4.1. Auto-Precharge Commands 10.4.2. Additive Latency 10.4.3. Bank Interleaving 10.4.4. Additive Latency and Bank Interleaving 10.4.5. User-Controlled Refresh 10.4.6. Frequency of Operation 10.4.7. Series of Reads or Writes 10.4.8. Data Reordering 10.4.9. Starvation Control 10.4.10. Command Reordering 10.4.11. Bandwidth 10.4.12. Enable Command Priority Control 10.4.13. Controller Pre-pay and Post-pay Refresh (DDR4 Only)
10.4.1. Auto-Precharge Commands x
10.4.1.1. Using Auto-precharge to Achieve Highest Memory Bandwidth for DDR4 Interfaces
11. Intel® Agilex™ FPGA EMIF IP – Debugging x
11.1. Interface Configuration Performance Issues 11.2. Functional Issue Evaluation 11.3. Timing Issue Characteristics 11.4. Verifying Memory IP Using the Signal Tap Logic Analyzer 11.5. Hardware Debugging Guidelines 11.6. Categorizing Hardware Issues 11.7. Debugging with the External Memory Interface Debug Toolkit 11.8. Using the Default Traffic Generator 11.9. Using the Configurable Traffic Generator (TG2) 11.10. EMIF On-Chip Debug Port 11.11. Efficiency Monitor
11.1. Interface Configuration Performance Issues x
11.1.1. Interface Configuration Bottleneck and Efficiency Issues
11.2. Functional Issue Evaluation x
11.2.1. Intel® IP Memory Model 11.2.2. Vendor Memory Model 11.2.3. Transcript Window Messages 11.2.4. Modifying the Example Driver to Replicate the Failure
11.3. Timing Issue Characteristics x
11.3.1. Evaluating FPGA Timing Issues 11.3.2. Evaluating External Memory Interface Timing Issues
11.4. Verifying Memory IP Using the Signal Tap Logic Analyzer x
11.4.1. Signals to Monitor with the Signal Tap Logic Analyzer
11.5. Hardware Debugging Guidelines x
11.5.1. Create a Simplified Design that Demonstrates the Same Issue 11.5.2. Measure Power Distribution Network 11.5.3. Measure Signal Integrity and Setup and Hold Margin 11.5.4. Vary Voltage 11.5.5. Operate at a Lower Speed 11.5.6. Determine Whether the Issue Exists in Previous Versions of Software 11.5.7. Determine Whether the Issue Exists in the Current Version of Software 11.5.8. Try A Different PCB 11.5.9. Try Other Configurations 11.5.10. Debugging Checklist
11.6. Categorizing Hardware Issues x
11.6.1. Signal Integrity Issues 11.6.2. Hardware and Calibration Issues
11.6.1. Signal Integrity Issues x
11.6.1.1. Characteristics of Signal Integrity Issues 11.6.1.2. Evaluating Signal Integrity Issues
11.6.1.2. Evaluating Signal Integrity Issues x
11.6.1.2.1. Skew 11.6.1.2.2. Crosstalk 11.6.1.2.3. Power System 11.6.1.2.4. Clock Signals 11.6.1.2.5. Address and Command Signals 11.6.1.2.6. Read Data Valid Window and Eye Diagram 11.6.1.2.7. Write Data Valid Window and Eye Diagram 11.6.1.2.8. OCT and ODT Usage
11.6.2. Hardware and Calibration Issues x
11.6.2.1. Postamble Timing Issues and Margin 11.6.2.2. Intermittent Issue Evaluation 11.6.2.3. Memory Timing Parameter Evaluation 11.6.2.4. Verify that the Board Has the Correct Memory Component or DIMM Installed
11.7. Debugging with the External Memory Interface Debug Toolkit x
11.7.1. Prerequisites for Using the EMIF Debug Toolkit 11.7.2. Configuring a Design to Use the Toolkit 11.7.3. Launching the EMIF Debug Toolkit 11.7.4. Using the EMIF Debug Toolkit 11.7.5. Exporting Tables 11.7.6. Viewing Reports Graphically in the Eye Viewer
11.7.2. Configuring a Design to Use the Toolkit x
11.7.2.1. Generating a Design Example with the Debug Toolkit 11.7.2.2. Creating a Design Example with Multiple External Memory Interfaces 11.7.2.3. Enabling the EMIF Toolkit in an Existing Design
11.7.4. Using the EMIF Debug Toolkit x
11.7.4.1. Memory Configuration Tab 11.7.4.2. Calibration Tab 11.7.4.3. Guidelines for Debugging Calibration Issues 11.7.4.4. Calibration Report Tab 11.7.4.5. Calibrate Termination Tab 11.7.4.6. Vref Margining Tab 11.7.4.7. Driver Margining Tab 11.7.4.8. ISSPs Tab 11.7.4.9. Pin Delay Settings Tab
11.7.4.2. Calibration Tab x
11.7.4.2.1. Rerunning Calibration 11.7.4.2.2. Rerunning the Traffic Generator
11.7.4.3. Guidelines for Debugging Calibration Issues x
11.7.4.3.1. Debugging Calibration Failure Using Information from the Calibration report 11.7.4.3.2. Debugging Address and Command Leveling Calibration Failure 11.7.4.3.3. Debugging Address and Command Deskew Failure 11.7.4.3.4. Debugging DQS Enable Failure 11.7.4.3.5. Debugging Read Deskew Calibration Failure 11.7.4.3.6. Debugging VREFIN Calibration Failure 11.7.4.3.7. Debugging LFIFO Calibration Failure 11.7.4.3.8. Debugging Write Leveling Failure 11.7.4.3.9. Debugging Write Deskew Calibration Failure 11.7.4.3.10. Debugging VREFOUT Calibration Failure
11.8. Using the Default Traffic Generator x
11.8.1. Reading the Default Traffic Generator Status 11.8.2. Running Infinite Traffic using the Default Traffic Generator 11.8.3. Changing the Reset Trigger of the Default Traffic Generator 11.8.4. Observing Generated Traffic with Signal Tap
11.9. Using the Configurable Traffic Generator (TG2) x
11.9.1. Enabling the Traffic Generator in a Design Example 11.9.2. Traffic Generator Block Description 11.9.3. Default Traffic Pattern 11.9.4. Configuration and Status Registers 11.9.5. User Pattern 11.9.6. Traffic Generator Status 11.9.7. Starting Traffic with the Traffic Generator 11.9.8. Traffic Generator Configuration User Interface
11.9.5. User Pattern x
11.9.5.1. Test Duration / Instruction pattern 11.9.5.2. Address Pattern 11.9.5.3. Data Pattern and Byte Enable
11.9.5.2. Address Pattern x
11.9.5.2.1. Address Generator Modes 11.9.5.2.2. Address Generator MSB Indices 11.9.5.2.3. Address Generator Effective Width 11.9.5.2.4. Address Generator Relative Frequencies 11.9.5.2.5. Address Pattern Examples - Basic Mode 11.9.5.2.6. Address Pattern Examples - Advanced Mode
11.9.8. Traffic Generator Configuration User Interface x
11.9.8.1. Connecting the Traffic Generator 11.9.8.2. Configuring the Traffic Generator 11.9.8.3. Traffic Generator Preset Selection 11.9.8.4. Traffic Generator Status Report 11.9.8.5. Examples of Configuring the TG2 Traffic Generator
11.10. EMIF On-Chip Debug Port x
11.10.1. Enabling the On-Chip Debug Port 11.10.2. I/O SSM calbus Bridge Data Structures and Usage 11.10.3. I/O SSM User-RAM Data Structures and Usage 11.10.4. Debug Interface Structure 11.10.5. Example: Reading Calibration Results and Margins with the On-Chip Debug Port
11.10.3. I/O SSM User-RAM Data Structures and Usage x
11.10.3.1. Parameter Tables 11.10.3.2. Parameter Table Enums
11.10.3.1. Parameter Tables x
11.10.3.1.1. Global Parameter Table 11.10.3.1.2. Per-interface Parameter Table Structure 11.10.3.1.3. Parameter Table Arrays
11.10.4. Debug Interface Structure x
11.10.4.1. Debug Data Structures 11.10.4.2. Debug Enums
11.11. Efficiency Monitor x
11.11.1. Enabling the Efficiency Monitor in a Design Example 11.11.2. Efficiency Monitor Block Descriptions 11.11.3. Control and Status Registers 11.11.4. Opening the Efficiency Monitor Toolkit
1. About the External Memory Interfaces Intel® Agilex™ FPGA IP
2. Intel® Agilex™ FPGA EMIF IP – Introduction
3. Intel® Agilex™ FPGA EMIF IP – Product Architecture
4. Intel® Agilex™ FPGA EMIF IP – End-User Signals
5. Intel® Agilex™ FPGA EMIF IP – Simulating Memory IP
6. Intel® Agilex™ FPGA EMIF IP – DDR4 Support
7. Intel® Agilex™ FPGA EMIF IP – QDR-IV Support
8. Intel® Agilex™ FPGA EMIF IP – Timing Closure
9. Intel® Agilex™ FPGA EMIF IP – I/O Timing Closure
10. Intel® Agilex™ FPGA EMIF IP – Controller Optimization
11. Intel® Agilex™ FPGA EMIF IP – Debugging
12. External Memory Interfaces Intel® Agilex™ FPGA IP User Guide Archives
13. Document Revision History for External Memory Interfaces Intel® Agilex™ FPGA IP User Guide

9.3.1. Address/Command Simulation Deck

The address/command simulation deck is configured to mimic Lane 0 of the External Memory Pin Interface table.
Note: All pin indices are 0-based throughout this section.

The configuration has the following characteristics:

  • Two pins within the 12-pin lane are driven with a differential clock pattern to map to the CK0 and CK0_N clock pairs. These pins are driven with a repeating clock pattern, phase-adjusted such that the rising/falling edge of CK0/CK0_N are centered within the valid data window.
    • For DDR4, CK0/CK0_N map to pins 8/9 within the lane.
    • For QDR4, CK0/CK0_N map to pins 4/5 within the lane.
  • The symbol rate of the address/command pins matches the characteristics of the memory technology.
    • For DDR4, the address/command channel is single-data-rate.
    • For QDR4, the address/command channel is double-data-rate.
  • One pin is designated as a victim pin, driven with a PRBS-10 pattern (repeated three times).
    • For DDR4, pin 5 is designated as the victim pin.
    • For QDR4, pin 8 is designated as the victim pin.
  • The remaining pins are driven with the same aggressor PRBS-15 data pattern.

The structure of the address/command simulation deck is illustrated below.

Figure 123. Address/Command Simulation Deck Structure

The simulation deck assumes a 12-line channel model; the purpose of each block is described in the sections below.

FPGA

At the far left of the above diagram, 12 FPGA IBIS models are instantiated and configured to match the IP-specified memory-clock and address/command electrical settings. The pattern generators are embedded within this sub-circuit and automatically configured via the IP-generated parameter file.

PCB — 12-Line Channel Model

This block connects to a 12-line (24-port plus ground) channel model that matches the signals implementing Lane 0 of the address/command bus for that interface. By default, this subcircuit instantiates 12 independent ideal 50-ohm transmission lines, which can be replaced by specifying the following options in the IP-generated parameter file:

Parameter Name Default Value Description
USE_AC_PCB_EXTRACTION False Specifies if a 24-port Touchstone extraction is to be used in the Address/Command simulation. If set to true, then the AC_PCB_EXTRACTION_FILE parameter must specify a valid Touchstone file location. If set to false, ideal transmission line models are used.
AC_PCB_EXTRACTION_FILE <empty> Specifies the file name for the 24-port Touchstone extraction file (.s24p) that represents the address/command channel. The first 12 pins map to Lane 0, Pins 0 through 11 connected to the balls of the FPGA. The next 12 pins map to the far-end of the channel, which is either the memory balls (for direct point-to-point connections) or to the bifurcation point for a multi-rank topology.
Note: Refer to the ac_pcb_wrapper.sp file for details on how the PCB extraction file is integrated into the SPICE simulation deck.

Multi-Rank and Connector Model

Use this block in cases where you want to do the following:

  • Model the multi-rank topology of your interface to explore the effect of parameters such as DIMM spacing, channel loading, or component stacking.
  • Model the effect of the DIMM connector by inserting a vendor-supplied simulation model.

This block is situated between the end of the address/command PCB extraction model and the start of the fly-by topology for each rank. If you don't specify an extraction model, then the default model of the block is as follows:

  • For single rank DDR4 systems, the default model for this block is a direct pass-through.
  • For dual-rank DDR4 systems, the default model for this block is to connect directly to rank 0, and insert 12 independent 50-ohm, 50ps transmission lines between rank 0 and rank 1 to model a slight delay difference between the two ranks.
  • For quad-rank DDR4 systems, the default model for this block is to connect directly to ranks 0 and 1, and insert 12 independent 50-ohm, 50ps transmission lines between ranks 0/1 and ranks 2/3.

You can override this default behavior by specifying an S-Parameter model of your own, using the following options. (Note that the number of ports of the model should align with the IP-calculated value of the number of A/C ranks in the system, MEMAC_RANKS. A value of 1, 2 or 4 correspond to single, dual- or quad-rank systems and require 24, 36 or 60-port models, respectively.)

Parameter Name Default Value Description
USE_AC_MULTIRANK_CONNECTOR_EXTRACTION False Specifies if a N-port Touchstone extraction is to be used in the address/command simulation to model the multi-rank bifurcation topology and/or the DIMM connector. If set to true, then the AC_MULTIRANK_CONNECTOR_EXTRACTION_FILE parameter must specify a valid Touchstone file location. If set to false, ideal transmission line models are used.
AC_MULTIRANK_CONNECTOR_EXTRACTION_FILE <empty> Specifies the file name for the N-port Touchstone extraction file (.sNp) that represents the address/command channel. The first 12 pins map to Lane 0, Pins 0 through 11 connected to the endpoint of the 12-line A/C channel model. The next 12 pins connect to the start of the A/C Fly-by topology for Rank 0. If MEMAC_RANKS is greater than 1, the next 12 pins connect to the address/command fly-by topology for Rank 1, and Rank 2 and Rank 3 for quad-rank systems.
Note: Refer to the ac_mr_conn_wrapper.sp file for details on how the multi-rank/connector model integrates into the SPICE simulation deck.

DIMM/Fly-By + VTT Model

This block models the fly-by topology of the address/command bus for a multi-component memory interface. You can use this file to adjust factors that can affect the signal integrity of the address/command interface, including the following:

  • Component loading
  • Component spacing
  • Trace geometry/impedance between components
  • Effect of using DDP components vs planar components
  • Effect of placement/value of termination resistor network
  • Effect of trace spacing on signal integrity
  • Effect of vias, back-drilling, and/or placement of components on both sides of board

By default, this block generates MEMAC_COMP_PER_RANKS components in the fly-by chain. The first component is connected directly to the first 12 ports of the subcircuit. Subsequent components are connected to the previous component using 12 independent 50-ohm, 50ps ideal transmission lines. The final component in the flyby chain has each of its signals terminated to the VTT source via 50-ohm resistors. Note that the differential clock is AC-coupled to the VTT source through a 0.01uF capacitor.

You can override this default behavior by specifying an S-Parameter model of your own, using the following options. (Note that the number of ports of the model should align with the IP-calculated value of the number of A/C components in the system, MEMAC_COMP_PER_RANKS. The minimum number of ports for a value of 1 is 24 ports, increasing by 12 ports for every additional component in the flyby chain.

Parameter Name Default Value Description
USE_AC_FLYBY_EXTRACTION False Specifies if an N-port Touchstone extraction is to be used in the address/command simulation to model the fly-by topology. If set to true, then the AC_FLYBY_EXTRACTION_FILE parameter must specify a valid Touchstone file location. If set to false, ideal transmission line models are used.
AC_FLYBY_EXTRACTION_FILE <empty> Specifies the file name for the N-port Touchstone extraction file (.sNp) that represents the address/command channel. The first 12 pins map to Lane 0, Pins 0 through 11 connected to the endpoint of the 12-line A/C multirank/connector model. The next 12 pins connect to the first component of the address/command Fly-by topology. If MEMAC_COMP_PER_RANKS is greater than 1, the next 12 pins connect to each additional component in the chain up to a maximum of 9 components (a 12-port model).
Note: Refer to the ac_dimm_flyby_wrapper.sp file for details on how the multi-rank/connector model integrates into the SPICE simulation deck.
Level Two Title