Someone has pointed out to me that the implementation in the Virtex2PRO will only go through the alignment once, i.e. ignoring any subsequent ALIGN characters in the data stream.
The spec is a little ambiguous and I am unsure if this does actually happen. Quote: "if a comma is detected and the data is aligned, no further alteration takes place" which sounds like what i was told, but it then goes on to say:
"The transceiver continuously monitors the data for the presence of the 10-bit chars. Upon each occurrence of the 10-bit char, the data is checked for word alignment"
Does anyone have any practical experience of the alignment (comma detection enabled)? If so once the data is aligned are all subsequent align chars in the stream ignored? or does the det/Align re-align the data every time?