Thanks for all your reply. We will most likely select the Cyclone IV device for this application. I believe good technical support is also a key factor in selection of devices, as I got no response from Lattice in both local sales and forum.
After I read the datasheet carefully, there is a section in I/O Timing, and it mentioned that there is an Excel-based I/O Timing worksheet. Then I download the worksheet and looked at the devices. The worse case delay is (tsu + tco), which equate to a little higher than 100MHz for a C8 speed grade. So I think it should be okay. We will install the software and further verify this with simulation.
BR - Henry