Will you be using BOINC?

No. Although i am not familiar with boinc i already have made plans for my own infrastructure, which will be developed in the next weeks. If you want to take part in the planning and/or implementation, write to the mailing list.

I have multiple cards should i generate multiple tables?

This depends on your hardware. For every table you need to generate about 500 million chains. With a gtx260 and 162 chains per second that takes 35 days and is totally reasonable. A 9600M-GT does 20 chains per second, which would take 9 month, which is quite a long time. So it is best to calculate the same table with all your hardware in this case. If you had 2 gtx260 you could finish 2 independent tables in a month. But since you do not loose anything when generating the first table with 2 cards and the second table with 2 cards afterwards, it is up to you. just keep in mind that each table needs about 500 million chains, so divide the work based on the hardware capability (or just start them up and monitor the generated files till they have the expected size and then stop the generators manually).

Who will come and use the data that i am generating?

Eventually your tables should be distributed by you as a torrent and you will keep a background process running to accept lookup requests from the network to search in your finished table. If you do not want to participate in the lookup process with your table then you can quit after generating the tables and distributing them. The ideal situation in the lookup phase would be to have sufficient nodes online 24/7. As this will probably not be possible, we try to have enough nodes, so that maybe an online time of 8 hours on average per user gives close to 100% table coverage 24/7. The code for table lookup and automatic torrent file creation is in development and is expected to be available at the time the tables are there. Since you would not want $some_guy on the internet use your resources, there will be some system to earn points doing lookups in your data and spend points looking up chains in other peoples data. You may or may not want to take this in consideration when distributing your data. There should be a system to trade a table for a table, since redundancy a necessary feature and some users may want to get something back for their month of generating instead of giving the table out for free. This is a topic worth discussing on the mailing list.

Are there differences between the ports

There are minor differences which are not part of the interface described in the documentation. All file formats and network message formats in particular are the same. The only difference that is remotely of interest is the location of the named pipe in the filesystem, which has to be \\.\pipe\ on windows.