This may seem like an elementary question/application, but I'll bring it up nonetheless in hopes of getting a thorough understanding...
In our design, there are 80MHz system-synchronous interfaces between two FPGA's. There is a common clock source on the board with matched trace lengths to each of the FPGA's. The clocks then go into DCM's and the DCM 1x output clock is used to clock the IOB registers and also used as internal feedback to the DCM. Can we [almost] guarantee that the clocks coming out of the DCM's on the separate FPGA's are near phase-aligned, assuming matched trace lengths coming in? These are V4-LX160 parts. I was looking over the V4 user guide and couldn't find a fitting clocking application example. It seems it can never be fully guaranteed, since the DCM's deskew compensation on each of the FPGA's will certainly differ, not to mention small process variations. Since we have a 12.5ns period, I think we should have room in our timing budget to absorb these small phase differences. I will ensure that all the inputs and outputs utilize the IOB registers.
If anyone could reassure me that this design is relatively common and safe, and provide me with some information regarding the DCM output clock relationships on the separate devices, I will feel much better. I've definitely worked with these types of designs in the past, but never fully understood why things just work.