The material references the Intel® Stratix® 10 device architecture as well as aspects of the Intel® Quartus® Prime software and third-party tools that you might use in your design. The guidelines presented in this document can improve productivity and avoid common design pitfalls.
|Stages of the Design Flow||Description|
|System Specification||Planning design specifications, IP selection|
|Device Selection||Device information, determining device variant and density, package offerings, migration, speed grade|
|Early System and Board Planning||Early power estimation, thermal management option, planning for configuration scheme, planning for on-chip debugging|
|Pin Connection Considerations for Board Design||Power-up, power pins, PLL connections, decoupling capacitors, configuration pins, signal integrity, board-level verification|
|I/O and Clock Planning||Pin assignments, early pin planning, I/O features and connections, memory interfaces, clock and PLL selection, simultaneous switching noise (SSN)|
|Design Entry||Coding styles and design recommendations, Platform Designer, planning for hierarchical or team-based design|
|Design Implementation, Analysis, Optimization, and Verification||Synthesis tool, device utilization, messages, timing constraints and analysis, area and timing optimization, compilation time, verification, power analysis and optimization|
In systems that contain an Intel® Stratix® 10 device, the FPGA typically plays a large role in the overall system and affects the rest of the system design. It is important to start the design process by creating detailed design specifications for the system and the FPGA, and determining the FPGA input and output interfaces to the rest of the system.
|1||Create detailed design specifications and a test plan if appropriate.|
|2||Plan clock resources and I/O interfaces early with a block diagram.|
Create detailed design specifications that define the system before you create your logic design or complete your system design, by performing the following:
- Specify the I/O interfaces for the FPGA
- Identify the different clock domains
- Include a block diagram of basic design functions
- Include intellectual property (IP) blocks
Note: Taking the time to create these specifications improves design efficiency, but this stage is often skipped by FPGA designers.
- Create a functional verification/test plan
- Consider a common design directory structure
Create a functional verification plan to ensure the team knows how to verify the system. Creating a test plan at this stage can also help you design for testability and design for manufacturability. For example, do you want to perform built-in-self test (BIST) functions to drive interfaces? If so, you could use a UART interface with a Nios® processor inside the FPGA device. You might require the ability to validate all the design interfaces.
If your design includes multiple designers, it is useful to consider a common design directory structure. This eases the design integration stages.
|1||Select IP that affects system design, especially I/O interfaces.|
|2||If you plan to evaluate Intel® FPGA IP, ensure that your board design supports JTAG connections.|
Intel and its third-party IP partners offer a large selection of off-the-shelf IP cores optimized for Intel devices. You can easily implement these parameterized blocks of IP in your design, reducing your system implementation and verification time, and allowing you to concentrate on adding proprietary value.
IP selection often affects system design, especially if the FPGA interfaces with other devices in the system. Consider which I/O interfaces or other blocks in your system design can be implemented using IP cores, and plan to incorporate these cores in your FPGA design.
The Intel® FPGA IP Evaluation Mode feature available for many IP cores allows you to program the FPGA to verify your design in hardware before you purchase the IP license. The evaluation supports an untethered mode, in which the design runs for a limited time, or a tethered mode. The tethered mode requires an Intel serial JTAG cable connected between the JTAG port on your board and a host computer running the Intel® Quartus® Prime Programmer for the duration of the hardware evaluation period.
|1||Take advantage of Platform Designer for system and processor designs.|
Platform Designer is a system integration tool included as part of the Intel® Quartus® Prime software. Platform Designer captures system-level hardware designs at a high level of abstraction and automates the task of defining and integrating customized Hardware Description Language (HDL) components. These components include IP cores, verification IP, and other design modules. Platform Designer facilitates design reuse by packaging and integrating your custom components with Intel and third-party IP components. Platform Designer automatically creates interconnect logic from the high-level connectivity you specify, thereby eliminating the error-prone and time-consuming task of writing HDL to specify system-level connections.
Platform Designer is more powerful if you design your custom components using standard interfaces. By using standard interfaces, your components inter-operate with the components in the Platform Designer Library. In addition, you can take advantage of bus functional models (BFMs), monitors, and other verification IP to verify your design.
Device selection is the first step in the Intel® Stratix® 10 design process—choosing the device family variant, device density, features, package, and speed grade that best suit your design requirements.
|1||Consider the available device variants.|
|2||Select a device based on transceivers, I/O pin count, LVDS channels, package offering, logic/memory/multiplier density, PLLs, clock routing, and speed grade.|
The Intel® Stratix® 10 device family consists of several device variants optimized to meet different application requirements.
|1||Verify the number of PLLs and clock resources.|
Verify that your chosen device density package combination includes enough PLLs and clock routing resources for your design.
|1||Estimate the required logic, memory, and multiplier density.|
|2||Reserve device resources for future development and debugging.|
Intel® Stratix® 10 devices offer a range of densities that provide different amounts of device logic resources, including memory, multipliers, and adaptive logic module (ALM) logic cells. Determining the required logic density can be a challenging part of the design planning process. Devices with more logic resources can implement larger and potentially more complex designs, but generally have a higher cost. Smaller devices have lower static power utilization. Intel® Stratix® 10 devices support vertical migration, which provides flexibility.
You may observe 10% – 15% increase in resource utilization when you migrate your design from existing devices to Intel® Stratix® 10 devices. Review the resource utilization to find out which device density fits the design. Consider that the coding style, device architecture, and optimization options used in the Intel® Quartus® Prime software can significantly affect a design’s resource utilization and timing performance.
Select a device that meets your design requirements with some safety margin in case you want to add more logic later in the design cycle, upgrade, or expand your design. You might also want additional space in the device to ease design floorplan creation for an incremental or team-based design. Consider reserving resources for debugging.
|1||Estimate the number of I/O pins that you require.|
|2||Consider the I/O pins you need to reserve for debugging.|
|3||Verify that the number of LVDS channels are enough.|
Determine the required number of I/O pins for your application, considering the design’s interface requirements with other system blocks.
Larger densities and package pin counts offer more full-duplex LVDS channels for different signaling; ensure that your device density-package combination includes enough LVDS channels. Other factors can also affect the number of I/O pins required for a design, including simultaneous switching noise (SSN) concerns, pin placement guidelines, pins used as dedicated inputs, I/O standard availability for each I/O bank, differences between I/O standards and speed for row and column I/O banks, and package migration options.
You can compile any existing designs in the Intel® Quartus® Prime software to determine how many I/O pins are used. Also consider reserving I/O pins for debugging.
|1||Determine the device speed grade that you require.|
The device speed grade affects the device timing performance and timing closure, as well as power utilization. One way to determine which speed grade your design requires is to consider the supported clock rates for specific I/O interfaces.
You can use the fastest speed grade during prototyping to reduce compilation time (because less time is spent optimizing the design to meet timing requirements), and then move to a slower speed grade for production to reduce cost if the design meets its timing requirements.
|1||Consider vertical device migration availability and requirements.|
|2||Refer to the Intel® Stratix® 10 External Memory Interfaces IP User Guide for the external memory interface (EMIF) pin pairing.|
Intel® Stratix® 10 devices support vertical migration within the same package, which enables you to migrate to different density devices whose dedicated input pins, configuration pins, and power pins are the same for a given package. This feature allows future upgrades or changes to your design without any changes to the board layout, because you can replace the FPGA on the board with a different density Intel® Stratix® 10 device.
Determine whether you want the option of migrating your design to another device density. Choose your device density and package to accommodate any possible future device migration to allow flexibility when the design nears completion. You should specify any potential migration options in the Intel® Quartus® Prime software at the beginning of your design cycle or as soon as the device migration selection is possible in the Intel® Quartus® Prime software. Selecting a migration device can impact the design’s pin placement, because the Fitter ensures your design is compatible with the selected devices. It is possible to add migration devices later in the design cycle, but it requires extra effort to check pin assignments, and can require design or board layout changes to fit into the new target device. It is easier to consider these issues early in the design cycle than at the end, when the design is near completion and ready for migration.
The Intel® Quartus® Prime Pin Planner highlights pins that change function in the migration device when compared to the currently selected device.
System information related to the FPGA should be planned early in the design process, before designers have completed the design in the Intel® Quartus® Prime software. Early planning allows the FPGA team to provide early information to PCB board and system designers.
|1||Estimate power consumption with the Early Power Estimator (EPE) spreadsheet to plan the cooling solution and power supplies before the logic design is complete.|
FPGA power consumption is an important design consideration and must be estimated accurately to develop an appropriate power budget to design the power supplies, voltage regulators, decouplers, heat sink, and cooling system. Power estimation and analysis have two significant planning requirements:
- Thermal planning—The cooling solution must sufficiently dissipate the heat generated by the device. In particular, the computed junction temperature must fall within normal device specifications.
- Power supply planning—The power supplies must provide adequate current to support device operation.
Power consumption in FPGA devices is dependent on the logic design. This dependence can make power estimation challenging during the early board specification and layout stages. The Intel EPE tool allows you to estimate power utilization before the design is complete by processing information about the device and the device resources that will be used in the design, as well as the operating frequency, toggle rates, and environmental considerations. You can use the tool to obtain thermal design parameters, with which you can perform detailed thermal simulation and cooling solution design.
If you do not have an existing design, estimate the number of device resources used in your design and enter it manually. The EPE tool accuracy depends on your inputs and your estimation of the device resources. If this information changes (during or after your design is complete), your power estimation results are less accurate. If you have an existing design or a partially-completed compiled design, use the Generate Early Power Estimator File command in the Intel® Quartus® Prime software to provide input to the spreadsheet.
The EPE spreadsheet includes the Import Data macro, which parses the information in the Intel® Quartus® Prime-generated power estimation file (.csv), or alternatively from an older version of the EPE, and transfers it into the EPE tool. If you do not want to use the macro, you can transfer the data into the EPE tool manually. You should enter additional resources to be used in the final design manually if the existing Intel® Quartus® Prime project represents only a portion of your full design. You can edit the inputs to the EPE tool and add additional device resources or adjust the parameters after importing the power estimation file information.
When the design is complete, the Power Analyzer tool in the Intel® Quartus® Prime software provides more accurate estimation of power, ensuring that thermal and supply budgets are not violated. For the most accurate power estimation, use gate-level simulation results with an output file (.vcd) from a third-party simulation tool.
|1||Obtain thermal design power and thermal parameters from EPE.|
|2||Perform thermal simulation to determine a proper cooling solution.|
Intel® Stratix® 10 is a multi-chip module, depending on package configuration and design information, power distribution on all dies can be quite different. This feature makes the Intel® Stratix® 10 thermal characteristics design dependent. EPE takes into consideration of your design input, and generates unique thermal parameters for your design in the Thermal Page. You can get power consumption for each die, thermal resistance for all dies (ψJC), cooling solution requirement (ψCA), and maximum allowed package case temperature (Tcase).
The thermal analysis of the Intel® Stratix® 10 device requires you use a Compact Thermal Model (contact your local Intel® representatives to obtain the model) and perform simulation in a Computational Fluid Dynamics (CFD) tool. The result of the CFD analysis gives the Tcase which should be lower than the required value in the EPE Thermal Page. With the simulated Tcase, ψJC, and total package power, you can obtain the actual junction temperature Tj, which needs to stay below your requirement, for example, 95˚C. You can adjust your cooling solution (heatsink design, airflow, and so on) to optimize your thermal design.
|1||Set up the temperature sensing diode (TSD) in your design to measure the device junction temperature for thermal management.|
|2||Include offset values form EPE calculation to TSD reading.|
Intel® Stratix® 10 devices have internal and external temperature sensing capabilities. You can measure the junction temperature, TJ, with the following methods:
- Using internal TSD by instantiating the Intel® FPGA S10 Temperature Sensor IP core.
- Using external thermal diode designed to interface with third-party sensor chip. Ensure the third-party sensor chip matches the external TSD specifications as documented in the Intel® Stratix® 10 Device Datasheet.
Monitoring the actual junction temperature is crucial for thermal management. Intel® Stratix® 10 devices include TSD on each die with embedded analog-to-digital converter (ADC) circuitry. You can access the digital temperature readout through the Intel® FPGA S10 Temperature Sensor IP core.
The Intel® Stratix® 10 TSD can self-monitor the device junction temperature and can be used with external circuitry for activities such as controlling air flow to the FPGA. This requires including the TSD circuitry by instantiating a Stratix 10 ADC IP core.
The flexibility in Intel® Stratix® 10 design can lead to non-uniform power distribution on transceiver dies. Therefore, the hotspot on the transceiver dies may not necessarily be at the TSD location. This results in a temperature difference between TSD readout and real junction temperature. EPE calculates this difference and reports an offset value for each TSD. You need to add the offset values to your TSD reading to get the actual junction temperature.
|1||Determine if you need to use the voltage sensor.|
Intel® Stratix® 10 devices have an on-chip voltage sensor. The sensor provides a 8-bit digital representation of the analog signal being observed. This feature can be used for live monitoring of critical on-chip power supplies and external analog voltage.
|1||Consider whether you require multiple configuration schemes.|
Intel® Stratix® 10 devices are based on SRAM cells. You must download configuration data to the Intel® Stratix® 10 device each time the device powers up, because SRAM is volatile. Consider whether you require multiple configuration schemes, such as one for debugging or testing and another for the production environment.
Choosing the device configuration method early allows system and board designers to determine what companion devices, if any, are required for the system. Your board layout also depends on the configuration method you plan to use for the programmable device, because different schemes require different connections.
In addition, Intel® Stratix® 10 devices offer advanced configuration features, depending on your configuration scheme. Intel® Stratix® 10 devices also include optional configuration pins and a reconfiguration option that you should choose early in the design process (and set up in the Intel® Quartus® Prime software), so you have all the information required for your board and system design.
|1||Select a configuration scheme to plan companion devices and board connections.|
Intel® Stratix® 10 devices offer several configuration schemes.
You can enable any specific configuration scheme by driving the Intel® Stratix® 10 device MSEL pins to specific values on the board.
All configuration schemes use a configuration device, a download cable, or an external controller (for example, MAX® ( MAX® II, MAX® V, Intel® MAX® 10) devices or a microprocessor).
|1||If you want to use the AS configuration mode with large device densities, confirm there is a configuration device available that is large enough for your target FPGA density.|
Intel low-voltage quad-serial configuration (EPCQ-L) devices are used in the AS configuration scheme.
Serial configuration devices can be programmed using a Intel® FPGA Download Cable II or Intel® FPGA Ethernet Cable II download cable with the Intel® Quartus® Prime software through the active serial interface.
Alternatively, you can use supported third-party programmers such as BP Microsystems and System General, or a microprocessor with the SRunner software driver. SRunner is a software driver developed for embedded serial configuration device programming that designers can customize to fit in different embedded systems.
Serial configuration devices do not directly support the JTAG interface; however, you can program the device with JTAG download cables using the Intel® Stratix® 10 FPGA as a bridge between the JTAG interface and the configuration device, allowing both devices to use the same JTAG interface.
Programming the EPCQ-L from JTAG using the Intel® Stratix® 10 FPGA as a bridge is slower than using the standard AS interface.
|1||Use download cables for device configuration.|
The Intel® Quartus® Prime programmer supports configuration of the Intel® Stratix® 10 devices directly using JTAG interfaces with Intel programming download cables. You can download design changes directly to the device with Intel download cables, making prototyping easy and enabling you to make multiple design iterations in quick succession. You can use the same download cable to program configuration devices on the board and use JTAG debugging tools such as the Signal Tap Embedded Logic Analyzer.
|1||Ensure your configuration scheme and board support the required features: design security, remote upgrades, single event upset (SEU) mitigation.|
This section describes Intel® Stratix® 10 configuration features and how they affect your design process.
Configuration Bitstream Compression
Configuration bitstream compression is always enabled in Intel® Stratix® 10 configuration. The Intel® Quartus® Prime software generates configuration files with compressed configuration data. This compressed file reduces the storage requirements in the configuration device or flash memory, and decreases the time required to transmit the configuration bitstream to the Intel® Stratix® 10 device.
Due to compressed configuration bitstream, passive configuration schemes for example Avalon® -ST ×8, ×16, and ×32 might require the external configuration host to pause sending configuration data by deasserting the AVST_READY signal.
Design Security Using Configuration Bitstream Encryption
The design security feature ensures that Intel® Stratix® 10 designs are protected from copying, reverse engineering, and tampering. Intel® Stratix® 10 devices have the ability to decrypt configuration bitstreams using the AES algorithm, an industry standard encryption algorithm that is FIPS-197 certified. Intel® Stratix® 10 devices have a design security feature which utilizes a 256-bit security key.
The design security feature is available in the all configuration schemes supported in the Intel® Stratix® 10 devices.
Dedicated circuitry is built into Intel® Stratix® 10 devices for error detection and correction. When enabled, this feature checks for SEUs continuously and automatically. This allows you to confirm that the configuration data stored in an Intel® Stratix® 10 device is correct and alerts the system to a configuration error.
When using the SEU mitigation features, an SDM pin is used to implement the SEU_ERROR function. This pin flags errors for your system to take appropriate actions. Prior to compiling your design, enable the SEU_ERROR function and select an unused SDM pin to implement the SEU_ERROR function in the Intel® Quartus® Prime software.
|1||Consider the Intel® Quartus® Prime configuration options when you plan your board and system design.|
There are several configuration options that you can set in the Intel® Quartus® Prime software before compilation to generate configuration or programming files. Your board and system design are affected by these settings and pins, so consider them in the planning stages. Set the options on the General category of the Device and Pin Options dialog box.
|1||Plan the board design to support optional configuration pins as required.|
You can enable the following optional configuration pins:
- OSC_CLK_1—Must be connected to a 25 MHz, 100 MHz, or 125 MHz source if used.
Intel® Stratix® 10 devices use OSC_CLK_1 pin as the reference clock for transceiver calibration. You must provide a stable and free running clock input at this pin.
|1||Plan the dual purpose pins that can function as configuration pins and user I/O pins.|
The below configuration pins used for the Avalon® -ST ×16 and ×32 configuration schemes can optionally be used as user I/O pins after configuration has completed. Enable the pins to function as dual purpose pins in the Intel® Quartus® Prime software prior to compilation, if desired.
- AVST_DATA[31:16]—for Avalon® -ST ×32 configuration scheme
On-chip debugging is an optional step in the design flow, and different debugging tools work better for different systems and different designers. Evaluate on-chip debugging options early in your design process to ensure that your system board, Intel® Quartus® Prime project, and design are able to support the appropriate options. Planning can reduce time spent debugging, and eliminates design changes later to accommodate your preferred debugging methodologies. Adding debug pins might not be enough, because of internal signal accessibility and I/O pin accessibility on the device. First, select your preferred debugging tools.
|1||Take advantage of on-chip debugging features to analyze internal signals and perform advanced debugging techniques.|
The Intel® Quartus® Prime portfolio of verification tools includes the following in-system debugging features:
- Signal Probe incremental routing—Quickly routes internal signals to I/O pins without affecting the routing of the original design. Starting with a fully routed design, you can select and route signals for debugging to either previously reserved or currently unused I/O pins.
- Signal Tap Embedded Logic Analyzer—Probes the state of internal and I/O signals without the use of external equipment or extra I/O pins, while the design is running at full speed in an FPGA device. Defining custom trigger-condition logic provides greater accuracy and improves the ability to isolate problems. It does not require external probes or changes to the design files to capture the state of the internal nodes or I/O pins in the design; all captured signal data is stored in the device memory until you are ready to read and analyze the data. The Signal Tap Embedded Logic Analyzer works best for synchronous interfaces. For debugging asynchronous interfaces, consider using Signal Probe or an external logic analyzer to view the signals more accurately. Signal Tap may affect routing of the original design.
- Logic Analyzer Interface—Enables you to connect and transmit internal FPGA signals to an external logic analyzer for analysis, allowing you to take advantage of advanced features in your external logic analyzer or mixed signal oscilloscope. You can use this feature to connect a large set of internal device signals to a small number of output pins for debugging purposes and it can multiplex signals with design I/O pins if required.
- In-System Memory Content Editor—Provides read and write access to in-system FPGA memories and constants through the JTAG interface, so you can test changes to memory content and constant values in the FPGA while the device is functioning in the system.
- In-System Sources and Probes—Sets up customized register chains to drive or sample the instrumented nodes in your logic design, providing an easy way to input simple virtual stimuli and capture the current value of instrumented nodes.
- Virtual JTAG IP core—Enables you to build your own system-level debugging infrastructure, including both processor-based debugging solutions and debugging tools in the software for system-level debugging. You can instantiate the SLD_VIRTUAL_JTAG IP core directly in your HDL code to provide one or more transparent communication channels to access parts of your FPGA design using the JTAG interface of the device.
- EMIF Debug tool kit—A .tcl-based graphical user interface communicating via a JTAG connection to enable external memory interface on the circuit board to retrieve calibration status and debug information. The Driver Margining feature of the tool kit allows you to measure margins on your memory interface using a driver with arbitrary traffic patterns.
- Transceiver Toolkit—Uses System Console technology to help FPGA and board designers validate transceiver link signal integrity real time in a system and improve board bring-up time. Test for bit-error rate (BER) while simultaneously running multiple links at your target data rate to validate your board design with Transceiver Toolkit. Tune transceiver analog settings for optimal link performance while using different test metrics to quantify results. Simultaneously test multiple devices across one or more boards using link tests in the Transceiver Toolkit GUI.
|1||Select on-chip debugging schemes early to plan memory and logic requirements, I/O pin connections, and board connections.|
|2||If you want to use Signal Probe incremental routing, the Signal Tap Embedded Logic Analyzer, Logic Analyzer Interface, In-System Memory Content Editor, In-System Sources and Probes, or Virtual JTAG IP core, plan your system and board with JTAG connections that are available for debugging.|
|3||Plan for the small amount of additional logic resources used to implement the JTAG hub logic for JTAG debugging features.|
|4||For debugging with the Signal Tap Embedded Logic Analyzer, reserve device memory resources to capture data during system operation.|
|5||Reserve I/O pins for debugging with Signal Probe or the Logic Analyzer Interface so you do not have to change the design or board to accommodate debugging signals later.|
|6||Ensure the board supports a debugging mode where debugging signals do not affect system operation.|
|7||Incorporate a pin header or mictor connector as required for an external logic analyzer or mixed signal oscilloscope.|
|8||To use debug tools incrementally and reduce compilation time, ensure incremental compilation is on so you do not have to recompile the design to modify the debug tool.|
|9||To use the Virtual JTAG IP core for custom debugging applications, instantiate it in the HDL code as part of the design process.|
|10||To use the In-System Sources and Probes feature, instantiate the IP core in the HDL code.|
|11||To use the In-System Memory Content Editor for RAM or ROM blocks or the LPM_CONSTANT IP core, turn on the Allow In-System Memory Content Editor to capture and update content independently of the system clock option for the memory block in the IP catalog.|
If you intend to use any of the on-chip debugging tools, plan for the tool(s) when developing the system board, Intel® Quartus® Prime project, and design.
When designing the interfaces to the Intel® Stratix® 10 device, various factors can affect the PCB design.
|1||Design board for power-up: All Intel® Stratix® 10 GPIO pins are tri-stated until the device is configured and configuration pins drive out. The transceiver pins are at high impedance before the device periphery could get programmed. Once the periphery is programmed, the termination and Vcm are set immediately after transceiver calibration is complete.|
|2||Design voltage power supply ramps to be monotonic.|
|3||Set POR time to ensure power supplies are stable.|
|4||Design power sequencing and voltage regulators for best device reliability. Connect the GND between boards before connecting the power supplies.|
The minimum current requirement for the power-on-reset (POR) supplies must be available during device power-up.
The Intel® Stratix® 10 device has Power-On Reset Circuitry, which keeps the device in a reset state until the power supply outputs are within the recommended operating range. The device must reach the recommended operating range within the maximum power supply ramp time. If the ramp time is not met, the device I/O pins and programming registers remain tri-stated and device configuration fails. For the Intel® Stratix® 10 device to exit POR, you must power the VCCBAT power supply even if you do not use the volatile key.
In Intel® Stratix® 10 devices, a pin-selectable option (MSEL) allows you to select between a typical POR time setting of 4 ms or 100 ms. In both cases, you can extend the POR time by using an external component to assert the nSTATUS pin low. Extend POR time if the board cannot meet the maximum power ramp time specifications to ensure the device configures properly and enters user mode.
Intel® Stratix® 10 devices have power-up sequencing and power-down sequencing requirements. You should consider the power-up timing and power-down timing for each rail in order to meet the power sequencing requirements.
Intel uses GND as a reference for I/O buffer designs. Connecting the GND between boards before connecting the power supplies prevents the GND on your board from being pulled up inadvertently by a path to power through other components on your board. A pulled-up GND could otherwise cause an out-of-specification I/O voltage or current condition with the Intel device.
|1||Connect all power pins correctly as specified in the Intel® Stratix® 10 GX and SX Device Family Pin Connection Guidelines.|
|2||Connect VCCIO pins and VREF pins to support each bank’s I/O standards.|
|3||Explore unique requirements for FPGA power pins or other power pins on your board, and determine which devices on your board can share a power rail.|
|4||Follow the suggested power supply sharing and isolation guidance, and the specific guidelines for each pin in the Intel® Stratix® 10 GX and SX Device Family Pin Connection Guidelines.|
|5||Refer to AN 692: Power Sequencing Considerations for Intel® Arria® 10 and Intel® Stratix® 10 Devices to understand the power sequencing design requirements.|
Intel® Stratix® 10 devices require various voltage supplies depending on your design requirements.
Intel® Stratix® 10 devices support a wide range of industry I/O standards. The device output pins do not meet the I/O standard specifications if the VCCIO level is out of the recommended operating range for the I/O standard.
Voltage reference (VREF) pins serve as voltage references for certain I/O standards. The VREF pin is used mainly for a voltage bias and does not source or sink much current. The voltage can be created with a regulator or a resistor divider network.
The VREFP_ADC pin is not a power supply pin. It provides the reference voltage for the ADC for the voltage sensor. For better voltage sensor performance, connect the VREFP_ADC pin to an external reference 1.25 V source. Connecting the VREFP_ADC pin to GND actives an on-chip reference source.
|1||Use the PDN tool to plan your power distribution netlist and decoupling capacitors.|
Board decoupling is important for improving overall power supply integrity while ensuring the rated device performance.
Intel® Stratix® 10 devices include on-die decoupling capacitors to provide high-frequency decoupling. These low-inductance capacitors suppress power noise for excellent power integrity performance, and reduce the number of external PCB decoupling capacitors, saving board space, reducing cost, and greatly simplifying PCB design.
Intel has created an easy-to-use power distribution network (PDN) design tool that optimizes the board-level PDN graphically. The purpose of the board-level PDN is to distribute power and return currents from the voltage regulating module (VRM) to the FPGA power supplies. By using the PDN tool, you can quickly arrive at an optimized PDN decoupling solution for your specific design.
For each power supply, PDN designers must choose a network of bulk and decoupling capacitors. While SPICE simulation could be used to simulate the circuit, the PDN design tool provides a fast, accurate, and interactive way to determine the right number of decoupling capacitors for optimal cost and performance trade-offs.
|1||Connect all PLL power pins to reduce noise even if the design does not use all the PLLs.|
|2||Power supply nets should be provided by an isolated power plane, a power plane cut out, or thick trace of at least 20 mils.|
Plan your board design when you design a power system for PLL usage and to minimize jitter, because PLLs contain analog components embedded in a digital device.
|1||Review the transceiver board design guidelines when designing your board.|
|1||Check that all configuration pin connections and pull-up/pull-down resistors are set correctly for your configuration schemes.|
Depending on your configuration scheme, different pull-up/pull-down resistor or signal integrity requirements might apply. Some configuration pins also have specific requirements if unused. It is very important to connect the configuration pins correctly. The following guidelines address the common issues.
|1||Ensure VCCIO_SDM and VCCIO of the configuration pins match the voltage level of the external devices used for configuration.|
Configuration pins from the Intel® Stratix® 10 device connect to external devices, for example the EPCQ-L configuration device, Avalon® -ST host, SDMMC, or NAND flash memories. The voltage level of the configuration pins need to match the voltage level of the devices connected to them. The JTAG and SDM I/Os used as configuration pins are powered by the VCCIO_SDM supply. For Avalon® -ST ×32 and ×16 configuration schemes, the AVST_CLK, AVST_VALID, and AVST_DATA pins are powered by the VCCIO of the I/O bank in which the pins reside in. When using the Avalon® -ST ×32 or ×16 configuration scheme, VCCIO of the I/O bank which the AVST_CLK, AVST_VALID, and AVST_DATA pins are located in, must match the VCCIO_SDM level.
|1||Design configuration clock traces to be noise-free.|
Board trace for clocks used in configuration, for example TCK, AS_CLK, AVSTx8_CLK, AVST_CLK, SDMMC_CFG_CCLK, and OSC_CLK_1 clock input, should produce clean signals with no overshoot, undershoot, or ringing. When designing the board, lay out the configuration clock traces with the same techniques used to lay out a clock line. Any overshoot, undershoot, ringing, or other noise on the clock signal can cause configuration failure. Make sure to have clock routing as stripline. Keep the clock routing away from any high-speed signals to isolate the clock signals from other signals.
|1||Connect JTAG pins to a stable voltage level if not in use.|
Because JTAG configuration takes precedence over all other configuration methods, the JTAG pins should not be left floating or toggling during configuration if you do not use the JTAG interface. If you are using the JTAG interface, adhere to the following guidelines.
|1||Connect JTAG pins correctly to the download cable header. Ensure the pin order is not reversed.|
|2||To disable the JTAG state machine during power-up, pull the TCK pin low through a resistor to ensure that an unexpected rising edge does not occur on the TCK pin.|
|3||Pull the TMS and TDI pins high through a resistor.|
A device operating in JTAG mode uses four required pins—TDI, TDO, TMS, and TCK. The TCK pin has an internal weak pull-down resistor, while the TDI and TMS pins have weak internal pull-up resistors.
If you have more than one device in the chain, connect the TDO pin of a device to the TDI pin of the next device in the chain.
Noise on the JTAG pins during configuration, user mode, or power-up can cause the device to go into an undefined state or mode.
|1||Ensure the download cable and JTAG pin voltages are compatible because the download cable interfaces with the JTAG pins of your device.|
The operating voltage supplied to the Intel download cable by the target board through the 10-pin header determines the operating voltage level of the download cable.
JTAG pins in the Intel® Stratix® 10 device are powered up by VCCIO_SDM. In a JTAG chain containing devices with different VCCIO levels, ensure that the VIL max, VIH min, and the maximum VI specifications of the device JTAG input pins are not violated. Level shifter might be required between devices to meet the voltage specifications of the devices input pin.
|1||Buffer JTAG signals per the recommendations, especially for connectors or if the cable drives more than three devices.|
|2||If your device is in a configuration chain, ensure all devices in the chain are connected properly.|
You might have to add buffers to a JTAG chain, depending on the JTAG signal integrity, especially the TCK signal, because it is the JTAG clock and the fastest switching JTAG signal. Intel recommends buffering the signals at the connector because cables and board connectors tend to make bad transmission lines and introduce noise to the signals. After this initial buffer at the connector, add buffers as the chain gets longer or whenever the signals cross a board connector.
If a cable drives three or more devices, buffer the JTAG signal at the cable connector to prevent signal deterioration. This also depends on the board layout, loads, connectors, jumpers, and switches on the board. Anything added to the board that affects the inductance or capacitance of the JTAG signals increases the likelihood that a buffer should be added to the chain.
Each buffer should drive no more than eight loads for the TCK and TMS signals, which drive in parallel. If jumpers or switches are added to the path, decrease the number of loads.
|1||Connect the SDM pins with MSEL function to select the configuration scheme; do not leave them floating. Do not hardwire the pins to VCCIO_SDM or GND if they have other configuration functions based on the configuration scheme selected.|
Select the configuration scheme by pulling the SDM pins with MSEL function high or low with external resistors. JTAG configuration is always available, regardless of the MSEL settings. The SDM pins with MSEL function are powered by the VCCIO_SDM power supply, and they have internal weak pull-up resistors.
During POR and reconfiguration, the SDM pins with MSEL function must be at LVTTL VIL and VIH levels to be considered as logic low and logic high, respectively. The SDM pins used for MSEL function also have other configuration functions, depending on the configuration schemes used. Do not hardwire the SDM pins with MSEL function to VCCIO_SDM or GND without pull-up or pull-down resistors, if they are needed for the configuration scheme used.
|1||Use the SDM pins which have multiple configuration functions if power management function is required.|
Most of the SDM pins have multiple configuration functions, depending on the configuration schemes used. Some SDM pins also have power management functions. If power management function is required, choose the SDM pins which do not need to be used for configuration to implement the power management function.
Connect the SDM pins on your board to the external configuration host or configuration device based on the configuration scheme to be used. If more than one configuration scheme are used, ensure there is no contention between configuration host or configuration devices connected to the SDM pins.
|1||Set the settings for the FPGA I/O pins correctly and plan for the functionality during board design.|
The Intel® Quartus® Prime software provides options for the FPGA I/O pins that you should consider during board design. Ensure that these options are set correctly when the Intel® Quartus® Prime project is created, and plan for the functionality during board design.
|1||Specify the reserved state for unused I/O pins.|
|2||Carefully check the pin connections in the Intel® Quartus® Prime software-generated .pin file. Do not connect RESERVED pins.|
You can specify the state of unused pins in the Intel® Quartus® Prime software to allow flexibility in the board design by choosing one of the five allowable states for Reserve all unused pins on the Unused Pins category in the Device and Pin Options dialog box:
- As inputs tri-stated
- As output driving ground
- As outputs driving an unspecified signal
- As input tri-stated with bus-hold circuitry
- As input tri-stated with weak pull-up
The common setting is to set unused pins As inputs tri-stated with weak pull-up. To improve signal integrity, set the unused pins to As output driving ground. Doing this reduces inductance by creating a shorter return path and reduces noise on the neighboring I/Os. This approach should not be used if this results in many via paths causing congestion for signals under the device.
To reduce power dissipation, set clock pins and other unused I/O pins As inputs tri-stated, and tie them to ground.
Signal integrity considerations include detailed board design guidelines, as well as a few guidelines related to VREF pins, SSN, and I/O termination.
|1||Refer to the Board Design Resource Center.|
If your design has high-speed signals, especially with Intel® Stratix® 10 GX/SX device high-speed transceivers, the board design has a major impact on the signal integrity in the system.
|1||Design VREF pins to be noise-free.|
Voltage deviation on a VREF pin can affect the threshold sensitivity for inputs.
|1||Break out large bus signals on board layers close to the device to reduce cross talk.|
|2||Route traces orthogonally if two signal layers are next to each other, if possible. Use a separation of two to three times the trace width.|
SSN is a concern when too many pins (in close proximity) change voltage levels at the same time. Noise generated by SSN can reduce the noise margin and cause incorrect switching. Although SSN is dominant on the device package, plan the board layout according to the board layout recommendations in the PCB guidelines can help with noise reduction.
|1||Check I/O termination and impedance matching for chosen I/O standards, especially for voltage-referenced standards.|
Voltage-referenced I/O standards require both an VREF and a termination voltage (VTT). The reference voltage of the receiving device tracks the termination voltage of the transmitting device. Each voltage-referenced I/O standard requires a unique termination setup.
Although single-ended, non-voltage-referenced I/O standards do not require termination, impedance matching is necessary to reduce reflections and improve signal integrity.
Intel® Stratix® 10 on-chip series and parallel termination provides the convenience of no external components. Alternatively, you can use external pull-up resistors to terminate the voltage-referenced I/O standards such as SSTL and HSTL.
Differential I/O standards typically require a termination resistor between the two signals at the receiver. The termination resistor must match the differential load impedance of the signal line. Intel® Stratix® 10 devices provide an optional on-chip differential resistor when using LVDS.
|1||Perform board-level simulation using IBIS models (when available).|
|2||Configure board trace models for Intel® Quartus® Prime advanced I/O timing analysis.|
To ensure that the I/O signaling meets receiver threshold levels on your board setup, perform full board routing simulation with third-party board-level simulation tools using an IBIS model.
When this feature is available in the Intel® Quartus® Prime software, select IBIS under Board-level signal integrity analysis on the Board-Level page in EDA Tool Settings of the Settings dialog box.
When you include an FPGA device with high-speed interfaces in a board design, knowing the signal integrity and board routing propagation delay is vital for proper system operation. You should analyze board level timing as part of the I/O and board planning, especially for high-speed designs.
You can configure board trace models of selected I/O standards and generate “board-aware” signal integrity reports with the Intel® Quartus® Prime software. When Enable Advanced I/O Timing is turned on (Timing Analyzer page in the Settings dialog box), the Timing Analyzer uses simulation results for the I/O buffer, package, and the board trace model to generate more accurate I/O delays and extra reports to give insight into signal behavior at the system level. You can use these advanced timing reports as a guide to make changes to the I/O assignments and board design to improve timing and signal integrity.
Planning and allocating I/O and clock resources is an important task with the high pin counts and advanced clock management features in Intel® Stratix® 10 devices. Various considerations are important to effectively plan the available I/O resources to maximize utilization and prevent issues related to signal integrity. Good clock management systems are also crucial to the performance of an FPGA design.
The I/O and clock connections of your FPGA affect the rest of your system and board design, so it is important to plan these connections early in your design cycle.
|1||Use the Intel® Quartus® Prime Pin Planner to make pin assignments.|
|2||Use Intel® Quartus® Prime Fitter messages and reports for sign-off of pin assignments.|
|3||Verify that the Intel® Quartus® Prime pin assignments match those in the schematic and board layout tools.|
|4||Plan interfaces and device periphery using the Interface Planner. After design synthesis, use the Interface Planner to rapidly define a legal device floorplan. Planning using the Interface Planner involves initialization of the Interface Planner, reconciliation of project assignments, placement of periphery elements and clocks, and export of plan constraints to your Intel® Quartus® Prime project.|
With the Intel® Quartus® Prime Pin Planner GUI, you can identify I/O banks, VREF groups, and differential pin pairings to help you through the I/O planning process. Right-click in the Pin Planner spreadsheet interface and click the Pin Finder to search for specific pins. If migration devices are selected, the Pin Migration view highlights pins that change function in the migration device when compared to the currently selected device.
You have the option of importing a Microsoft Excel spreadsheet into the Intel® Quartus® Prime software to start the I/O planning process if you normally use a spreadsheet in your design flow. You can also export a spreadsheet compatible (.csv) file containing your I/O assignments when all pins are assigned.
When you compile your design in the Intel® Quartus® Prime software, I/O Assignment Analysis in the Fitter validates that the assignments meet all the device requirements and generates messages if there are any problems.
Intel® Quartus® Prime designers can then pass the pin location information to PCB designers. Pin assignments between the Intel® Quartus® Prime software and your schematic and board layout tools must match to ensure the design works correctly on the board where it is placed, especially if changes to the pin-out must be made. The Pin Planner is integrated with certain PCB design EDA tools and can read pin location changes from these tools to check the suggested changes. When you compile your design, the Intel® Quartus® Prime software generates the .pin file. You can use this file to verify that each pin is correctly connected in the board schematics.
|1||Use the Create Top-Level Design File command with I/O Assignment Analysis to check the I/O assignments before the design is complete.|
In many design environments, FPGA designers want to plan top-level FPGA I/O pins early so that board designers can start developing the PCB design and layout. The FPGA device’s I/O capabilities and board layout guidelines influence pin locations and other types of assignments. In cases where the board design team specifies an FPGA pin-out, it is crucial that you verify pin locations in the FPGA place-and-route software as soon as possible to avoid board design changes.
Starting FPGA pin planning early improves the confidence in early board layouts, reduces the chance of error, and improves the design’s overall time to market. You can create a preliminary pin-out for an Intel FPGA using the Intel® Quartus® Prime Pin Planner before the source code is designed.
Early in the design process, the system architect typically has information about the standard I/O interfaces (such as memory and bus interfaces), IP cores to be used in the design, and any other I/O-related assignments defined by system requirements.
The Pin Planner Create/Import IP Core feature interfaces with the IP catalog, and enables you to create or import custom IP cores that use I/O interfaces. Enter PLL and LVDS blocks, including options such as dynamic phase alignment (DPA), because options affect the pin placement rules. When you have entered as much I/O-related information as possible, generate a top-level design netlist file using the Create Top-Level Design File command in the Pin Planner. You can use the I/O analysis results to change pin assignments or IP parameters and repeat the checking process until the I/O interface meets your design requirements and passes the pin checks in the Intel® Quartus® Prime software.
When planning is complete, the preliminary pin location information can be passed to PCB designers. When the design is complete, use the reports and messages generated by the Intel® Quartus® Prime Fitter for the final sign-off of the pin assignments.
Intel® Stratix® 10 I/O pins are designed for ease of use and rapid system integration, while simultaneously providing high bandwidth. Independent modular I/O banks with a common bank structure for vertical migration lend efficiency and flexibility to the high speed I/O.
The following guidelines provide information pertaining to I/O features and pin connections.
|1||Plan the I/O signaling type based on the system requirements.|
|2||Allow the software to assign locations for the negative pin in differential pin pairs.|
Intel® Stratix® 10 devices support a wide range of industry I/O standards, including single-ended, voltage-referenced single-ended, and differential I/O standards. Follow these general guidelines when you select a signaling type.
Single-ended I/O signaling provides a simple rail-to-rail interface. Its speed is limited by the large voltage swing and noise. Single-ended I/Os do not require termination, unless reflection in the system causes undesirable effects.
Voltage-referenced signaling reduces the effects of simultaneous switching outputs (SSO) from pins changing voltage levels at the same time (for example, external memory interface data and address buses). Voltage-referenced signaling also provides an improved logic transition rate with a reduced voltage swing, and minimizes noise caused by reflection with a termination requirement. However, additional termination components are required for the reference voltage source (VTT).
Differential signaling eliminates the interface performance barrier of single-ended and voltage-referenced signaling, with superior speed using an additional inverted closely-coupled data pair. Differential signaling also avoids the requirement for a clean reference voltage. This is possible because of a lower swing voltage and noise immunity with a common mode noise rejection capability. Considerations for this implementation include the requirements for a dedicated PLL to generate a sampling clock, and matched trace lengths to eliminate the phase difference between an inverted and non-inverted pair.
Intel® Stratix® 10 I/O pins are organized in pairs to support differential standards. Each I/O pin pair can support differential input or output operations, with the exception of certain clock pins that support differential input operations only. In your design source code, define just one pin to represent a differential pair, and make a pin assignment for this positive end of the pair. When you specify a differential I/O standard, the Intel® Quartus® Prime software automatically places the corresponding negative pin.
|1||Select a suitable signaling type and I/O standard for each I/O pin. The I/O banks are located in I/O columns. Each I/O bank contains its own PLL, DPA, and SERDES circuitries.|
|2||Ensure that the appropriate I/O standard support is supported in the targeted I/O bank.|
|3||Place I/O pins that share voltage levels in the same I/O bank.|
|4||Verify that all output signals in each I/O bank are intended to drive out at the bank’s VCCIO voltage level.|
|5||Verify that all voltage-referenced signals in each I/O bank are intended to use the bank’s VREF voltage level.|
|6||Check the I/O bank support for LVDS and transceiver features.|
Intel® Stratix® 10 I/O pins are arranged in groups called modular I/O banks. Be sure to use the correct dedicated pin inputs for signals such as clocks and global control signals.
The board must supply each bank with one VCCIO voltage level for every VCCIO pin in a bank. Each I/O bank is powered by the VCCIO pins of that particular bank, and is independent of the VCCIO pins of other I/O banks. A single I/O bank supports output signals that are driving at the same voltage as the VCCIO. An I/O bank can simultaneously support any number of input signals with different I/O standards.
To accommodate voltage-referenced I/O standards, each I/O bank supports multiple VREF pins feeding a common VREF bus. Set the VREF pins to the correct voltage for the I/O standards in the bank. Each I/O bank can only have a single VCCIO voltage level and a single VREF voltage level at a given time. If the VREF pins are not used as voltage references, they cannot be used as generic I/O pins and should be tied to VCCIO of that same bank or GND.
An I/O bank including single-ended or differential standards can support voltage-referenced standards as long as all voltage-referenced standards use the same VREF setting. Voltage-referenced bi-directional and output signals must drive out at the I/O bank’s VCCIO voltage level.
Different I/O banks include different support for LVDS signaling, and the Intel® Stratix® 10 transceiver banks include additional support. There are two types of I/O banks, LVDS and 3 V.
The LVDS I/O bank supports differential and single-ended I/O standards up to 1.8 V. The LVDS I/O pins form pairs of true differential LVDS channels. Each pair supports a parallel input/output termination between the two pins. You can use each LVDS channel as transmitter or receiver.
The 3 V I/O bank supports only single-ended I/O standards up to 3 V. Each adjacent I/O pair also supports Differential SSTL and Differential HSTL I/O standards. The single-ended output of the 3 V I/O has the same set of features as the single-ended output of the DDR I/O IP, except the programmable pre-emphasis feature.
|1||Use the Stratix 10 External Memory Interfaces IP core for each memory interface, and follow connection guidelines/restrictions in the appropriate documentation.|
|2||For a given bank, most memory pins are tied to a dedicated location. Refer to the Intel® Stratix® 10 GX and SX Device Family Pin Connection Guidelines for pin assignments.|
Intel® Stratix® 10 devices provide an efficient architecture to quickly and easily fit wide external memory interfaces with their small modular I/O banks. The Intel® Stratix® 10 FPGA can support DDR external memory on any I/O banks on all sides of the device that do not support transceivers.
The self-calibrating Stratix 10 External Memory Interfaces IP core is optimized to take advantage of the Intel® Stratix® 10 I/O structure. The Stratix 10 External Memory Interfaces IP core allows you to set external memory interface features and helps set up the physical interface (PHY) best suited for your system. When you use the Intel memory controller Intel® FPGA IP functions, the Stratix 10 External Memory Interfaces IP core is instantiated automatically. If you design multiple memory interfaces into the device using Intel FPGA IP core, generate a unique interface for each instance to ensure good results instead of designing it once and instantiating it multiple times.
The data strobe DQS and data DQ pin locations are fixed in Intel® Stratix® 10 devices. Before you design your device pin-out, refer to the memory interface guidelines for details and important restrictions related to the connections for these and other memory-related signals.
You can implement a protocol that is not supported by Stratix 10 External Memory Interfaces IP core by using the Stratix® 10 Intel® FPGA PHYLite for Parallel Interfaces IP core.
Address and command pins within the address/command bank must follow a fixed pin-out scheme, as defined in the <variation_name>_readme.txt file generated with your IP core. The pin-out scheme varies according to the topology of the memory interface. The pin-out scheme is a hardware requirement that you must follow. Some schemes require three lanes to implement address and command pins, while others require four lanes.
|1||Make dual-purpose pin settings and check for any restrictions when using these pins as regular I/O.|
Intel® Stratix® 10 devices allow I/O flexibility with dual-purpose configuration pins. You can use dual-purpose configuration pins as general I/Os after device configuration is complete. Select the desired setting for each of the dual-purpose pins on the Dual-Purpose Pins category of the Device and Pin Options dialog box. Depending on the configuration scheme, these pins can be reserved as regular I/O pins, as inputs that are tri-stated, as outputs that drive ground, or as outputs that drive an unspecified signal.
You can also use dedicated clock inputs, which drive the programmable clock routing networks, as general-purpose input pins if they are not used as clock pins. When you use the clock inputs as general inputs, I/O registers use ALM-based registers because the clock input pins do not include dedicated I/O registers.
The device-wide reset and clear pins are available as design I/Os if they are not enabled.
|1||Check available device I/O features that can help I/O interfaces: current strength, slew rate, I/O delays, open-drain, bus hold, programmable pull-up resistors, PCI™ clamping diodes, programmable pre-emphasis, and VOD.|
|2||Consider on-chip termination (OCT) features to save board space.|
|3||Verify that the required termination scheme is supported for all pin locations.|
|4||Choose the appropriate mode of DPA, non-DPA, or soft-CDR for high-speed LVDS interfaces.|
The Intel® Stratix® 10 bi-directional I/O element (IOE) features support rapid system integration while simultaneously providing the high bandwidth required to maximize internal logic capabilities and system-level performance. Advanced features for device interfaces assist in high-speed data transfer into and out of the device and reduce the complexity and cost of the PCB.
Intel recommends performing an IBIS or SPICE simulations to optimize your design settings.
|1||Use the correct dedicated clock pins and routing signals for clock and global control signals.|
|2||Use the device PLLs for clock management.|
|3||Analyze input and output routing connections for each PLL and clock pin. Ensure PLL inputs come from the dedicated clock pins or from another PLL.|
The first stage in planning your clocking scheme is to determine your system clock requirements. Understand your device’s available clock resources and correspondingly plan the design clocking scheme. Consider your requirements for timing performance, and how much logic is driven by a particular clock.
Intel® Stratix® 10 devices provide dedicated low-skew and high fan-out routing networks.
The dedicated clock pins drive the clock network directly, ensuring lower skew than other I/O pins. Use the dedicated routing network to have a predictable delay with less skew for high fan-out signals. You can also use the clock pins and clock network to drive control signals like asynchronous reset.
Connect clock inputs to specific PLLs to drive specific low-skew routing networks. Analyze the global resource availability for each PLL and the PLL availability for each clock input pin.
Intel® Stratix® 10 devices contain dedicated resources for distributing signals throughout the fabric with balanced delay. These resources are typically used for clock signals. You can also use these resources for other signals with low-skew requirements. In Intel® Stratix® 10 devices, these resources are implemented as a programmable clock routing, which allows for the implementation of low-skew clock networks of variable size.
If your system requires more clock or control signals than are available in the target device, consider cases where the dedicated clock resource could be spared, particularly low fan-out and low-frequency signals where clock delay and clock skew do not have a significant impact on the design performance. Use the Global Signal assignment in the Intel® Quartus® Prime Assignment Editor to select the type of global routing, or set the assignment to Off to specify that the signal should not use any global routing resources.
|1||Enable PLL features and check settings in the parameter editor.|
Based on your system requirements, define the required clock frequencies for your FPGA design, and the input frequencies available to the FPGA. Use these specifications to determine your PLL scheme. Use the Intel® Quartus® Prime parameter editor to enter your settings in Intel® FPGA IOPLL IP core, and check the results to verify whether particular features and input/output frequencies can be implemented in a particular PLL.
Intel® Stratix® 10 devices contain fractional PLLs in addition to I/O PLLs. You can configure fractional PLLs as integers or as enhanced fractional PLLs.
You can use I/O PLLs and fractional PLLs to reduce the number of oscillators required on the board, as well as to reduce the clock pins used in the FPGA by synthesizing multiple clock frequencies from a single reference clock source. In addition, you can use fractional PLLs for transmit clocking for transceivers.
Intel® Stratix® 10 device PLLs are feature rich, and support advanced capabilities such as clock feedback modes, switchover, and dynamic phase shifting.
|1||Ensure you select the correct PLL feedback compensation mode.|
Intel® Stratix® 10 PLLs support six different clock feedback modes.
|1||Check that the PLL offers the required number of clock outputs and use dedicated clock output pins.|
You can connect clock outputs to dedicated clock output pins or dedicated clock networks. There is no dedicated clock out pin for fractional PLL. I/O PLL can connect to a clock network or a dedicated clock pin.
|1||Use the clock control block for clock selection and power-down.|
Intel® Stratix® 10 devices uses these clock control features: clock gating and clock divider. The clock from the I/O PLL output can be gated dynamically. These clock signals along with other clock sources go to the periphery distributed clock multiplexer (DCM). In the periphery DCM, the clock signal can either pass straight through, be gated by the root clock gate, or be divided by the clock divider.
|1||Reduce the number of pins that switch the voltage level at exactly the same time whenever possible.|
|2||Use differential I/O standards and lower-voltage standards for high-switching I/Os.|
|3||Use lower drive strengths for high-switching I/Os. The default drive strength setting might be higher than your design requires.|
|4||Reduce the number of simultaneously switching output pins within each bank. Spread output pins across multiple banks if possible.|
|5||Spread switching I/Os evenly throughout the bank to reduce the number of aggressors in a given area to reduce SSN (when bank usage is substantially below 100%).|
|6||Separate simultaneously switching pins from input pins that are susceptible to SSN.|
|7||Place important clock and asynchronous control signals near ground signals and away from large switching buses.|
|8||Avoid using I/O pins one or two pins away from PLL power supply pins for high-switching or high-drive strength pins.|
|9||Use staggered output delays to shift the output signals through time, or use adjustable slew rate settings.|
SSN is a concern when too many I/Os (in close proximity) change voltage levels at the same time. Plan the I/O and clock connections according to the recommendations.
In complex FPGA design development, design practices, coding styles, and IP cores use have an enormous impact on your device’s timing performance, logic utilization, compilation time, and system reliability. In addition, while planning and creating the design, plan for a hierarchical or team-based design to improve design productivity.
|1||Use synchronous design practices. Pay attention to clock signals.|
In a synchronous design, a clock signal triggers all events. When all of the registers’ timing requirements are met, a synchronous design behaves in a predictable and reliable manner for all process, voltage, and temperature (PVT) conditions. You can easily target synchronous designs to different device families or speed grades.
Problems with asynchronous design techniques include reliance on propagation delays in a device, incomplete timing analysis, and possible glitches. Pay particular attention to your clock signals, because they have a large effect on your design’s timing accuracy, performance, and reliability. Problems with clock signals can cause functional and timing problems in your design. Use dedicated clock pins and clock routing for best results. For clock inversion, multiplication, and division, use the device PLLs. For clock multiplexing and gating, use the dedicated clock control block or PLL clock switchover feature instead of combinational logic. If you must use internally generated clock signals, register the output of any combinational logic used as a clock signal to reduce glitches. For example, if you divide a clock using combinational logic, clock the final stage with the clock signal that was used to clock the divider circuit.
|1||Use IP cores with the parameter editor.|
Intel provides parameterizable IP cores that are optimized for Intel device architectures. You can save design time by using IP cores instead of coding your own logic. Additionally, the Intel-provided IP cores can offer more efficient logic synthesis and device implementation. You can scale the IP core’s size and set various options with parameters. IP cores include the library of parameterized modules (LPM) and Intel device-specific IP cores. You can also take advantage of Intel and third-party IP cores and reference designs to save design time. The Intel® Quartus® Prime IP catalog provides a user interface to customize IP cores. You should build or change IP core parameters using the parameter editor to ensure you set all ports and parameters correctly.
|1||Consider the reconfiguration feature for your board development.|
Intel® Stratix® 10 devices allow you to easily modify your transceivers and FPGA-core while other portions of your design are still running by using dynamic reconfiguration and partial reconfiguration, respectively.
Intel® Stratix® 10 devices allow you to dynamically reconfigure different portions of the transceivers for different protocols, data rates, and PMA settings without powering down any part of the device or interrupting adjacent transceiver channels. This feature will be available in a future release of the Intel® Quartus® Prime software.
If you are interested in using partial reconfiguration, contact your local Intel representatives for support.
|1||Follow recommended coding styles, especially for inferring device dedicated logic such as memory and DSP blocks.|
HDL coding styles can have a significant effect on the quality of results for programmable logic designs. Use Intel’s recommended coding styles to achieve optimal synthesis results. When designing memory and digital system processing (DSP) functions, understand the device architecture so you can take advantage of the dedicated logic block sizes and configurations.
After you create your design source code and apply constraints including the device selection and timing requirements, your synthesis tool processes the code and maps it to elements of the device architecture. The Intel® Quartus® Prime Fitter then performs placement and routing to implement the design elements in specific device resources. If required, you can use the Intel® Quartus® Prime software to optimize the design’s resource utilization and achieve timing closure, preserve the performance of unchanged design blocks, and reduce compilation time for future iterations. You can also verify the design functionality with simulation. This section provides guidelines for these stages of the compilation flow.
|1||Specify your synthesis tool and use the correct supported version.|
The Intel® Quartus® Prime software includes advanced and easy-to-use integrated synthesis that fully supports Verilog HDL and VHDL, as well as the Intel hardware description language (AHDL) and schematic design entry. You can also use industry-leading third-party EDA synthesis tools to synthesize your Verilog HDL or VHDL design, and then compile the resulting output netlist file in the Intel® Quartus® Prime software. Specify a third-party synthesis tool in the New Project Wizard or the EDA Tools Settings page of the Settings dialog box to use the correct Library Mapping File (.lmf) for your synthesis netlist.
Intel recommends using the most recent version of third-party synthesis tools, because tool vendors are continuously adding new features, fixing tool issues, and enhancing performance for Intel devices.
Different synthesis tools can give different results. If you want to select the best-performing tool for your application, you can experiment by synthesizing typical designs for your application and coding style and comparing the results. Be sure to perform placement and routing in the Intel® Quartus® Prime software to get accurate timing analysis and logic utilization results.
Your synthesis tool might offer the capability to create a Intel® Quartus® Prime project and pass constraints such as the EDA tool setting, device selection, and timing requirements that you specified in your synthesis project. You can use this capability to save time when setting up your Intel® Quartus® Prime project for placement and routing.
|1||Review resource utilization reports after compilation.|
After compilation in the Intel® Quartus® Prime software, review the device resource utilization information to determine whether the future addition of extra logic or other design changes introduce fitting difficulties. If your compilation results in a no-fit error, resource utilization information is important so you can analyze the fitting problems in your design.
To determine resource usage, refer to the Flow Summary section of the Compilation Report for a percentage representing the total logic utilization, which includes an estimation of resources that cannot be used due to existing connections or logic use.
For Intel® Stratix® 10 devices, low logic utilization does not have the lowest ALM utilization possible. In addition, a design that is reported as close to 100% full might still have space for extra logic. The Fitter uses ALUTs in different ALMs, even when the logic can be placed within one ALM, so that it can achieve the best timing and routability results. Logic might be spread throughout the device when achieving these results. As the device fills up, the Fitter automatically searches for logic that can be placed together in one ALM.
More detailed resource information is available by viewing the reports under Fitter > Place section of the Compilation Report. The Fitter Resource Usage Summary report breaks down the logic utilization information and indicates the number of fully and partially used ALMs, and provides other resource information including the number of bits in each type of memory block. There are also reports that describe some of the optimizations that occurred during compilation. For example, if you use the Intel® Quartus® Prime integrated synthesis, the reports under Analysis & Synthesis > Partition <partition_name> > Optimization Results provide information, including registers that were removed during synthesis. Use this report to estimate device resource utilization for a partial design to ensure that registers were not removed due to missing connections with other parts of the design.
|1||Review all Intel® Quartus® Prime messages, especially warning or error messages.|
Each stage of the compilation flow generates messages, including informational notes, warnings, and critical warnings. Review these messages to check for any design problems. Ensure that you understand the significance of any warning messages, and make changes to the design or settings if required. In the Intel® Quartus® Prime user interface, you can use the Message window tabs to look at only certain types of messages, and you can suppress messages if you have determined that they do not require any action from you.
|1||Ensure timing constraints are complete and accurate, including all clock signals and I/O delays.|
|2||Review the Timing Analyzer reports after compilation to ensure there are no timing violations.|
|3||Ensure that the input I/O times are not violated when data is provided to the Intel® Stratix® 10 device.|
In an FPGA design flow, accurate timing constraints allow timing-driven synthesis software and place-and-route software to obtain optimal results. Timing constraints are critical to ensure designs meet their timing requirements, which represent actual design requirements that must be met for the device to operate correctly. The Intel® Quartus® Prime software optimizes and analyzes your design using different timing models for each device speed grade, so you must perform timing analysis for the correct speed grade. The final programmed device might not operate as expected if the timing paths are not fully constrained, analyzed, and verified to meet requirements.
The Intel® Quartus® Prime software includes the Intel® Quartus® Prime Timing Analyzer, a powerful ASIC-style timing analysis tool that validates the timing performance of all logic in your design. It supports the industry standard Synopsys® Design Constraints (SDC) format timing constraints, and has an easy-to-use GUI with interactive timing reports. It is ideal for constraining high-speed source-synchronous interfaces and clock multiplexing design structures.
A comprehensive static timing analysis includes analysis of register to register, I/O, and asynchronous reset paths. It is important to specify the frequencies and relationships for all clocks in your design. Use input and output delay constraints to specify external device or board timing parameters. Specify accurate timing requirements for external interfacing components to reflect the exact system intent.
The Timing Analyzer performs static timing analysis on the entire system, using data required times, data arrival times, and clock arrival times to verify circuit performance and detect possible timing violations. It determines the timing relationships that must be met for the design to correctly function.
You can use the report_datasheet command to generate a datasheet report that summarizes the I/O timing characteristics of the entire design.
|1||Turn on Optimize multi-corner timing on the Fitter Settings page in the Settings dialog box.|
|2||Use create_clock and create_generated_clock to specify the frequencies and relationships for all clocks in your design.|
|3||Use set_input_delay and set_output_delay to specify the external device or board timing parameters.|
|4||Use derive_clock_uncertainty to automatically apply inter-clock, intra-clock, and I/O interface uncertainties.|
|5||Use check_timing to generate a report on any problem with the design or applied constraints, including missing constraints.|
|6||Use set_false_path or set_clock_groups for asynchronous paths.|
These assignments and settings are important for large designs such as those in Intel® Stratix® 10 devices.
When you turn on the Optimize multi-corner timing option, the design is optimized to meet its timing requirements at all timing process corners and operating conditions. Therefore, turning on this option helps create a design implementation that is more robust across PVT variations.
In your Timing Analyzer .sdc constraints file, apply the recommended constraints to your design.
|1||Run Fitter (Plan) if you want timing estimates before running a full compilation.|
|2||Use Intel® Quartus® Prime optimization features to achieve timing closure or improve the resource utilization.|
|3||Use the Timing Optimization Advisors to suggest optimization settings.|
This section highlights some of the features offered in the Intel® Quartus® Prime software to help optimize area (or resource utilization) and timing performance. If the timing analysis reports that your design requirements were not met, you must make changes to your design or settings and recompile the design to achieve timing closure. If your compilation results in no-fit messages, you must make changes to get successful placement and routing.
You can run Fitter (Plan) to estimate your design’s timing results before the software performs full placement and routing. Click Processing > Start > Start Fitter (Plan) to generate initial compilation results after you have run analysis and synthesis.
Physical synthesis optimizations make placement-specific changes to the netlist that improve results for a specific Intel device. You can optimize for performance and fitting in the Physical Synthesis Optimizations page of the Settings dialog box. The options in the Physical Synthesis Optimizations page typically increase compilation time significantly but can provide significant improvements to the quality of results with push-button optimizations. If you turn on these options, ensure that they do improve the results for your design. If you do not require these options to meet your design timing requirements, turn off the options to reduce the compilation time.
The Design Space Explorer II (DSE II) is a utility that automates the process to find optimal project settings for resource, performance, or power optimization goals. DSE II attempts multiple seeds to identify one that meets your requirements. The Exploration Panel > Exploration mode allows you a predefine exploration space to target design performance, area of improvements, or power reduction with multiple compilations.
The Optimization Advisors provide guidance in making settings that optimize your design. On the Tools menu, click Advisor > Timing Optimization Advisor. Evaluate the options and choose settings that suit your requirements.
|1||Use incremental compilation to preserve performance for unchanged blocks in your design and to reduce compilation times.|
|2||Ensure parallel compilation is enabled if you have multiple processors available for compilation.|
|3||Use the Compilation Time Advisor to suggest settings that reduce compilation time.|
Use the incremental compilation feature to preserve logic in unchanged parts of your design, preserve timing performance, and reach timing closure more efficiently. You can speed up design iteration time by an average of 60% when making changes to the design with the incremental compilation feature.
The Intel® Quartus® Prime software can run some algorithms in parallel to take advantage of multiple processors and reduce compilation time when more than one processor is available to compile the design. Set the Parallel compilation option on the Compilation Process Settings page of the Settings dialog box, or change the default setting in the Options dialog box in the Processing page from the Tools menu.
The Compilation Time Advisor provides guidance in making settings that reduce your design compilation time. On the Tools menu, point to Advisors and click Compilation Time Advisor. Using some of these techniques to reduce compilation time can reduce the overall quality of results.
|1||Use Intel® Hyperflex™ feature to optimize your design and achieve enhanced performance.|
Intel’s Intel® Hyperflex™ core architecture adds registers to both the interconnect routing and the inputs of all major functional blocks in the FPGA. These added registers, called Hyper-Registers, are different from conventional registers. Conventional registers are present only in the adaptive logic modules (ALMs). Hyper-Registers can help to achieve significant core performance improvement.
To achieve this enhanced performance, you must optimize your designs using the following steps:
|1||Specify your simulation tool, and use the correct supported version and simulation models.|
The Intel® Quartus® Prime software supports both RTL and gate level functional simulations. Perform functional simulation at the beginning of your design flow to check the design functionality or logical behavior of each design block. You do not have to fully compile your design; you can generate a functional simulation netlist that does not contain timing information.
Intel provides the ModelSim® - Intel® FPGA Starter Edition and offers the higher performance ModelSim® - Intel® FPGA Edition, which enable you to take advantage of advanced testbench capabilities and other features. In addition, the Intel® Quartus® Prime EDA Netlist Writer can generate timing netlist files to support other third-party simulation tools such as Synopsys® VCS, Cadence NC-Sim, and Aldec Active-HDL. Specify your simulation tool in the EDA Tools Settings page of the Settings dialog box to generate the appropriate output simulation netlist.
If you use a third-party simulation tool, use the software version that is supported with your Intel® Quartus® Prime software version. The Intel® Quartus® Prime Software Release Notes list the version of each simulation tool that is officially supported with that particular version of the Intel® Quartus® Prime software. Use the model libraries provided with your Intel® Quartus® Prime software version, because libraries can change between versions, which might cause a mismatch with your simulation netlist. To create a testbench, on the Processing menu, point to Start and click Start Test Bench Template Writer.
|1||After compilation, analyze power consumption and heat dissipation in the Power Analyzer.|
|2||Provide accurate signal activities, preferably with a gate-level simulation .vcd, to get accurate power analysis results.|
|3||Specify the correct operating conditions for power analysis.|
Before design completion, estimate power consumption using the EPE spreadsheet. After compiling your design, analyze the power consumption and heat dissipation with the Intel® Quartus® Prime Power Analyzer to ensure the design has not violated power supply and thermal budgets.
You must compile a design (to provide information about design resources, placement and routing, and I/O standards) and provide signal activity data (toggle rates and static probabilities) to use the Power Analyzer. You can derive signal activity data from simulation results or a user-defined default toggle rate and vectorless estimation. The signal activities used for analysis must be representative of the actual operating behavior. For the most accurate power estimation, use gate-level simulation results with a .vcd output file from a third-party simulation tool. The simulation activity should include typical input vectors over a realistic time period and not the corner cases often used during functional verification. Use the recommended simulator settings (such as glitch filtering) to ensure good results.
You must also specify operating conditions, including the core voltage, device power characteristics, ambient and junction temperature, cooling solution, and the board thermal model. Select the appropriate settings on the Operating Settings and Conditions page in the Settings dialog box.
To calculate the dynamic, static, and I/O thermal power consumption, on the Processing menu, click Power Analyzer Tool. The tool also provides a summary of the signal activities used for analysis and a confidence metric that reflects the overall quality of the data sources for signal activities.
The report is a power estimate based on the data provided, and is not a power specification. Always refer to the datasheet for the power specification of your device.
Intel® Stratix® 10 devices utilize advanced process and circuit techniques, along with major circuit and architecture innovations, to minimize power and deliver high performance.
To reduce dynamic power consumption in Intel® Stratix® 10 devices, you can use various design and software techniques to optimize your design.
Power optimization in the Intel® Quartus® Prime software depends on accurate power analysis results. Use the guidelines in the previous section to ensure the software optimizes the power utilization correctly for the design’s operating behavior and conditions.
|1||Use recommended design techniques and Intel® Quartus® Prime options to optimize your design for power consumption, if required.|
|2||Use the Power Optimization Advisor to suggest optimization settings.|
|1||Consider using a faster speed grade device.|
If your design includes many critical timing paths that require the high-performance mode, you might be able to reduce power consumption by using a faster speed grade device if available.
|1||Optimize the clock power management.|
Clocks represent a significant portion of dynamic power consumption, because of their high switching activity and long paths. The Intel® Quartus® Prime software automatically optimizes clock routing power by enabling only the portions of a clock network that are required to feed downstream registers. You can also use clock control features to dynamically enable or disable the clock network. When a clock network is powered down, all the logic fed by that clock network does not toggle, thereby reducing the overall power consumption of the device.
To reduce LAB-wide clock power consumption without disabling the entire clock tree, use the LAB-wide clock enable signal to gate the LAB-wide clock. The Intel® Quartus® Prime software automatically promotes register-level clock enable signals to the LAB level.
|1||Reduce the number of memory clocking events.|
Reduce the number of memory clocking events to reduce memory power consumption. You can use clock gating or the clock enable signals in the memory ports.
|1||Review the I/O power guidelines.|
The dynamic power consumed in the I/O buffer is proportional to the total load capacitance; therefore, lower capacitance reduces power consumption.
Non-terminated I/O standards such as LVTTL and LVCMOS have a rail to-rail output swing equal to the VCCIO supply voltage. Because dynamic power is proportional to the square of the voltage, use lower voltage I/O standards to reduce dynamic power. These I/O standards consume little static power.
Because dynamic power is also proportional to the output transition frequency, use resistively-terminated I/O standards such as SSTL for high-frequency applications. The output load voltage swings by an amount smaller than the VCCIO around a bias point; therefore, dynamic power is lower than for non-terminated I/O under similar conditions.
Resistively-terminated I/O standards dissipate significant static power because current is constantly driven into the termination network. Use the lowest drive strength that meets your speed and waveform requirements to minimize static power when using resistively terminated I/O standards.
The power used by external devices is not included in the EPE calculations, so be sure to include it separately in your system power calculations.
|1||Review recommended design techniques and Intel® Quartus® Prime options to optimize power consumption.|
The Intel® Quartus® Prime software offers power-optimized synthesis and fitting to reduce core dynamic power.
Optimizing your design for area also saves power because fewer logic blocks are used; therefore, there is typically less switching activity. Improving your design source code to optimize for performance can also reduce power usage. You can use the DSE and Power Optimization Advisor to provide additional suggestions to reduce power.
|1||Use the Power Optimization Advisor to suggest optimization settings.|
The Intel® Quartus® Prime software includes the Power Optimization Advisor, which provides specific power optimization advice and recommendations based on the current design project settings and assignments. On the Tools menu, point to Advisors and click Power Optimization Advisor. After making any of the recommended changes, recompile your design and run the Power Analyzer to check the change in your power results.
|October 2016||2016.10.31||Initial release.|