Return to the Products Page
  homesearchagentssupportask xilinxmap

 
Foundation Express System Implementation Flow (text version) 
 
Foundation Series Page Implementation Flow Overiew 
Foundation Express Tips 
Device Architecture Support 
Implementation Flow Diagram 
Foundation Express Implementation Flow Diagram
 
Foundation Express Implementation Flow Overview: 

1.  Create a Project 

    Click the  New Project  button to create a project.   All   HDL files processed by Foundation Express must be done through a project.
2. Add HDL files to a project and analyze HDL files 
    After creating a project, HDL files for the design can be  added to the project by clicking the  Add Source  button.  After  adding the HDL files  for  the design,   Foundation  Express   will automatically analyze the HDL files.
3. Implement the design 
    Select the top-level module/entity in the “Design Sources” window, then  click   the implement  button and specify the target package, part, and speed grade.
4.  Enter Constraints 
    In the “Chips” window, select the implementation.   Right-click  on the  selected  implementation  and  select   Edit Constraints.  The constraints  window  will appear for that implementation.  Edit the necessary constraints, and close the constraints window.
5. Optimize the Design 
    With the desired implementation selected  in  the  “Chips” window, click the optimization button. 
6. Write Out Detailed Design Report 
    Information  on  the  design  hierarchy,  inferred  operators,  cell count, timing constraints and clock speed estimates is available by clicking on the Project Report button.
7. Write Out an XNF File and Place and Route With F1.4 
    After optimization is complete, click on the Export Netlist button.  Open the Foundation Series Design Manager (Start -> Programs -> Xilinx Foundation Series -> Design Manager) and identify the XNF netlist as the Input Design source.
 

Foundation Express Tips:  
 
1. Control Finite State Machine (FSM) Encoding 

    The  default  FSM  encoding  style  is  set  in  the  Synthesis -> Options -> Project window.   One-Hot encoding is recommended for FPGAs, Binary encoding is recommended for CPLDs. 
2. Perform Static Timing Analysis and Debugging 
    Static  Timing Analysis  may be performed on optimized designs.   Select the design in the  “Chips”  window,  right-click  and select View results.   You  can  view  estimated  performance and other critical path information,  I/O  setup and clock-to-output timing as well as detailed area reports
3. Achieve optimum Synthesis performance 
    Optimum  synthesis  results  can  be  driven  using  the   Create Implementation   dialog  box.    Selecting  the   speed option will synthesize a design to optimize for highest speed. Selecting the area option will give minimal area.
 

Device Architecture Support:  
 

FPGA XC4000E/L
XC4000X (EX/XL/XV)
Spartan
XC5200
XC3X00A 
CPLD  XC9500
 
 

© 1998 Xilinx, Inc. All rights reserved
Trademarks and Patents