Stratix® 10 Embedded Memory User Guide

Download PDF
ID 683423
Date 3/29/2024
Public
Document Table of Contents
Document Table of Contents x
1. Stratix® 10 Embedded Memory Overview 2. Stratix® 10 Embedded Memory Architecture and Features 3. Stratix® 10 Embedded Memory Design Considerations 4. Stratix® 10 Embedded Memory IP References 5. Intel Stratix 10 Embedded Memory Design Example 6. Stratix® 10 Embedded Memory User Guide Archives 7. Document Revision History for the Stratix® 10 Embedded Memory User Guide
1. Stratix® 10 Embedded Memory Overview x
1.1. Stratix® 10 Embedded Memory Features
2. Stratix® 10 Embedded Memory Architecture and Features x
2.1. Byte Enable in Stratix® 10 Embedded Memory Blocks 2.2. Address Clock Enable Support 2.3. Asynchronous Clear and Synchronous Clear 2.4. Memory Blocks Error Correction Code (ECC) Support 2.5. Force-to-Zero 2.6. Coherent Read Memory 2.7. Freeze Logic 2.8. True Dual Port Dual Clock Emulator 2.9. 'X' Propagation Support in Simulation 2.10. Stratix® 10 Supported Embedded Memory IPs 2.11. Stratix® 10 Embedded Memory Clocking Modes 2.12. Stratix® 10 Embedded Memory Configurations 2.13. Initial Value of Read and Write Address Registers
2.1. Byte Enable in Stratix® 10 Embedded Memory Blocks x
2.1.1. Byte Enable Controls 2.1.2. Data Byte Output 2.1.3. Byte Enable Behavior
2.4. Memory Blocks Error Correction Code (ECC) Support x
2.4.1. Parity Bit 2.4.2. ECC Parity Flip 2.4.3. ECC Read-During-Write Behavior 2.4.4. Error Correction Code Truth Table
2.6. Coherent Read Memory x
2.6.1. Forwarding Logic
2.8. True Dual Port Dual Clock Emulator x
2.8.1. Reducing the DCFIFO Depth
2.11. Stratix® 10 Embedded Memory Clocking Modes x
2.11.1. Single Clock Mode 2.11.2. Read/Write Clock Mode 2.11.3. Input/Output Clock Mode 2.11.4. Asynchronous/Synchronous Clears in Clocking Modes 2.11.5. Output Read Data in Simultaneous Read/Write 2.11.6. Independent Clock Enables in Clocking Modes
2.12. Stratix® 10 Embedded Memory Configurations x
2.12.1. Mixed-Width Port Configurations
3. Stratix® 10 Embedded Memory Design Considerations x
3.1. Consider the Memory Block Selection 3.2. Consider the Concurrent Read Behavior 3.3. Customize Read-During-Write Behavior 3.4. Consider Power-Up State and Memory Initialization 3.5. Reduce Power Consumption 3.6. Avoid Providing Non-Deterministic Input 3.7. Avoid Changing Clock Signals and Other Control Signals Simultaneously 3.8. Including the Reset Release Intel® FPGA IP in Your Design 3.9. Resource and Timing Optimization Feature in MLAB Blocks 3.10. Consider the Memory Depth Setting 3.11. Consider Registering the Memory Output
3.3. Customize Read-During-Write Behavior x
3.3.1. Same-Port Read-During-Write Mode 3.3.2. Mixed-Port Read-During-Write Mode
4. Stratix® 10 Embedded Memory IP References x
4.1. On Chip Memory RAM and ROM Intel® FPGA IPs 4.2. eSRAM Intel® FPGA IP 4.3. FIFO Intel® FPGA IP 4.4. FIFO2 Intel® FPGA IP 4.5. Shift Register (RAM-based) Intel® FPGA IP
4.1. On Chip Memory RAM and ROM Intel® FPGA IPs x
4.1.1. Release Information for RAM and ROM Intel® FPGA IPs 4.1.2. RAM: 1-PORT Intel® FPGA IP Parameters 4.1.3. RAM: 2-PORT Intel® FPGA IP Parameters 4.1.4. RAM: 4-PORT Intel® FPGA IP Parameters 4.1.5. ROM: 1-PORT Intel® FPGA IP Parameters 4.1.6. ROM: 2-PORT Intel® FPGA IP Parameters 4.1.7. RAM and ROM Interface Signals 4.1.8. Changing Parameter Settings Manually
4.1.8. Changing Parameter Settings Manually x
4.1.8.1. RAM and ROM Parameter Settings
4.2. eSRAM Intel® FPGA IP x
4.2.1. Release Information for eSRAM Intel® FPGA IP 4.2.2. eSRAM System Features 4.2.3. eSRAM Intel® FPGA IP Parameters 4.2.4. eSRAM Intel® FPGA IP Interface Signals 4.2.5. eSRAM Intel® FPGA IP Simulation Walk-Through 4.2.6. eSRAM Timing Diagrams
4.2.2. eSRAM System Features x
4.2.2.1. eSRAM Specifications 4.2.2.2. eSRAM Usage Model
4.3. FIFO Intel® FPGA IP x
4.3.1. Release Information for FIFO Intel® FPGA IP 4.3.2. Configuration Methods 4.3.3. Specifications 4.3.4. FIFO Functional Timing Requirements 4.3.5. SCFIFO ALMOST_EMPTY Functional Timing 4.3.6. FIFO Output Status Flag and Latency 4.3.7. FIFO Metastability Protection and Related Options 4.3.8. FIFO Synchronous Clear and Asynchronous Clear Effect 4.3.9. SCFIFO and DCFIFO Show-Ahead Mode 4.3.10. Different Input and Output Width 4.3.11. DCFIFO Timing Constraint Setting 4.3.12. Coding Example for Manual Instantiation 4.3.13. Design Example 4.3.14. Gray-Code Counter Transfer at the Clock Domain Crossing 4.3.15. Guidelines for Embedded Memory ECC Feature 4.3.16. FIFO Intel® FPGA IP Parameters 4.3.17. Reset Scheme
4.3.3. Specifications x
4.3.3.1. Verilog HDL Prototype 4.3.3.2. VHDL Component Declaration 4.3.3.3. VHDL LIBRARY-USE Declaration 4.3.3.4. FIFO Signals 4.3.3.5. FIFO Parameter Settings
4.3.8. FIFO Synchronous Clear and Asynchronous Clear Effect x
4.3.8.1. Recovery and Removal Timing Violation Warnings when Compiling a DCFIFO
4.3.11. DCFIFO Timing Constraint Setting x
4.3.11.1. Embedded Timing Constraint 4.3.11.2. User Configurable Timing Constraint
4.3.11.2. User Configurable Timing Constraint x
4.3.11.2.1. SDC Commands
4.4. FIFO2 Intel® FPGA IP x
4.4.1. Release Information for FIFO2 Intel® FPGA IP 4.4.2. Configuration Methods 4.4.3. Fmax Target Measuring Methodology 4.4.4. Performance Considerations 4.4.5. FIFO2 Intel® FPGA IP Features 4.4.6. FIFO2 Intel® FPGA IP Parameters 4.4.7. FIFO2 Intel® FPGA IP Interface Signals 4.4.8. Reset and Clock Schemes
4.4.5. FIFO2 Intel® FPGA IP Features x
4.4.5.1. FIFO2 Specifications
4.4.6. FIFO2 Intel® FPGA IP Parameters x
4.4.6.1. FIFO2 Parameter Settings
4.4.7. FIFO2 Intel® FPGA IP Interface Signals x
4.4.7.1. SCFIFO Signals 4.4.7.2. DCFIFO Signals
4.4.8. Reset and Clock Schemes x
4.4.8.1. Clock Domains 4.4.8.2. Reset
4.4.8.2. Reset x
4.4.8.2.1. FIFO2 Intel® FPGA IP Reset Guidelines
4.5. Shift Register (RAM-based) Intel® FPGA IP x
4.5.1. Release Information for Shift Register (RAM-based) Intel® FPGA IP 4.5.2. Shift Register (RAM-based) Intel® FPGA IP Features 4.5.3. Shift Register (RAM-based) Intel® FPGA IP General Description 4.5.4. Shift Register (RAM-based) Intel® FPGA IP Parameter Settings 4.5.5. Shift Register Ports and Parameters Setting
5. Intel Stratix 10 Embedded Memory Design Example x
5.1. FIFO and FIFO2 Design Example
5.1. FIFO and FIFO2 Design Example x
5.1.1. Generating the Design Example 5.1.2. Simulating the Design Example
5.1.2. Simulating the Design Example x
5.1.2.1. Simulation Results Part 1 (0 - 100 ns) Part 2 (100 - 480 ns) Part 3 (480 - 700 ns) Part 4 (700 ns and onwards)
1. Stratix® 10 Embedded Memory Overview
2. Stratix® 10 Embedded Memory Architecture and Features
3. Stratix® 10 Embedded Memory Design Considerations
4. Stratix® 10 Embedded Memory IP References
5. Intel Stratix 10 Embedded Memory Design Example
6. Stratix® 10 Embedded Memory User Guide Archives
7. Document Revision History for the Stratix® 10 Embedded Memory User Guide

5.1.2.1. Simulation Results

The simulation result of the design example shows the comparison of behavior between FIFO and FIFO2 Intel® FPGA IP cores.
Note: The signal names used in this design example for both FIFO and FIFO2 Intel® FPGA IP cores are defined with reference to the FIFO input and output ports.

Part 1 (0 - 100 ns)

This section of the simulation shows the read-after-write operation with an invalid reset condition for FIFO2. FIFO2 requires a reset period of 32 cycles. As a result, some signals cannot return to their known states with only two clock cycles of reset. This causes FIFO2 unable to operate as expected. In contrast, FIFO demonstrates a correct write-then-read operation because FIFO only requires a minimum of one clock cycle of reset.
Figure 56. Part 1—Read-after-write Operation of FIFO

Part 2 (100 - 480 ns)

This section of the simulation shows FIFO2 operation with a valid reset scheme, followed by a write-then-read operation. After 32 cycles of reset, all signals return to known states. An 8 cycles of wait is also required before any operation.
Figure 57. Part 2—Valid Reset Scheme for FIFO2
Figure 58. Write Operation
Notes to Write Operation:
  • usedw signal: After the write request signal (e.g., i_wrreq) is asserted, the usedw signal of FIFO (e.g., o_usedw1) starts counting as data is being stored into FIFO. For FIFO2, it has usedw signal on both write (e.g., o_w_usedw2) and read (o_r_usedw2) sides. The write usedw starts counting after 3 cycles of delay, and it takes another 8 cycles to update the read usedw signal.
  • Full and empty signals: The FIFO full signal (e.g., o_full1) asserts high after the memory of FIFO is full. The FIFO2 full signal (e.g., o_full2) is asserted high on the same cycle as the o_w_usedw2 signal, which indicates that the memory is full. Note that when the three cycles before the full signal is asserted, the ready signal is already asserted low, which is an expected pipeline behavior of FIFO2. The FIFO empty signal (e.g., o_empty1), which has only one clock cycle of latency, asserts low after the first data is being written into memory. In contrast, FIFO2 takes three cycles of delay to indicate that its empty signal (e.g., o_empty2) has turned not-empty, after the o_r_usedw2 signal is risen to 1.
Figure 59. Read Operation
Notes to Read Operation:
  • On the read operation, the FIFO o_usedw1 signal is reflected on same positive clock edge with the read request signal. For FIFO2, two cycles of latency is observed before it is reflected on the o_r_usedw2 signal. It also takes another 9 cycles of propagation from o_r_usedw2 to o_w_usedw2.
Figure 60. FIFO Full Condition
Notes to FIFO Full Condition:
  • Ready signal: After a valid reset scheme, the FIFO2 ready signal (e.g., o_ready2) asserts high, which indicates that it is ready for the write operation.
    Note: FIFO does not have this signal.
  • Valid signal: The FIFO2 valid signal (e.g., o_valid2) is to indicate that the data being read out is valid. In the case of a valid read operation, the o_valid2 signal will assert high after two cycles of delay, with the assertion of the read request (e.g., i_rdreq) signal.
    Note: FIFO does not have this signal.

Part 3 (480 - 700 ns)

This section of the simulation showcases a valid reset condition, followed by a read-during-write operation. After reset, the write request signal (e.g., i_wrreq) asserts high. After a few cycles later, the i_rdreq signal also asserts high. For FIFO, the number of used word maintains while the data is being read out, which follows an expected behavior. Note that in this case, FIFO2 has no data being read out because the o_empty2 signal is not asserted low. Therefore, it is not a valid read operation for FIFO2.

Part 4 (700 ns and onwards)

This section of the simulation observes the behaviour of FIFO and FIFO2 under different scenario. When the FIFO2 o_r_usedw2 signal counts down to 2, the o_empty2 signal asserts high and undergoes an internal check. After two cycles, the o_empty2 signal goes back to a not-empty state after the status is validated. This mechanism is specially designed for FIFO2 according to its multiple pipeline nature.

Level Two Title