Do you need all the text to match, meaning are you only concerned about the text in a PDF or does the formatting matter?

I'm asking because they're different. I could reformat a PDF slightly, not changing any of the information (pictures/words) and a binary comparison would be different.

SQL Server does have the ability to read the words from PDFs using the iFilter interface and full text search. However, I'm not sure this is what you want, or if this would work. I haven't tried an exact comparison of large text in an iFTS situation. That's interesting. I'll see if I can find a way to try this.

Note: Performance would likely be poor, so if you were comparing more than two items, meaning a search, this would probably not work well.