System Level Perspective - 2020.2 English

UltraFast Design Methodology Guide for Xilinx FPGAs and SoCs

Document ID
UG949
Release Date
2021-02-18
Version
2020.2 English

The I/O paths are modeled like register-to-register paths by the Vivado Design Suite timing engine, except that you must define a constraint to model the part of the path delay located outside the device. When analyzing internal paths, minimum and maximum delays are considered for both setup and hold analysis. This is also true for I/O paths. For this reason, it is important to describe both min and max delay conditions. The I/O timing paths are analyzed as single-cycle paths by default, which means the following:

  • For max delay analysis (setup), the data is captured one clock cycle after the launch edge for single data rate interface, and half clock cycle after the launch edge for a double data rate interface.
  • For min delay analysis (hold), the data is launched and captured by the same clock edge.

If the relationship between the clock and I/O data must be timed differently, like for example in a source synchronous interface, different I/O delays and additional timing exceptions must be specified. This corresponds to an advanced I/O timing constraints scenario.