Which design would be faster ...

Hi

I'm creating an application that have an fixed length alphanumeric key
and each key has (currently) up to 5 fixed length alphanumeric
variables.  I was wondering which of the designs would be faster
considering that I could have about million keys...

The first design is the obvious one. create one table with ID and
"KEYNAME", and a second table which uses the ID of the first table as
foreign key to list values.

The second design is to use one table with the columns: "key",
"value1", "value2", etc...
Since it's highly unlikely that we'll ever change the number of
values, I'm ready to consider this option if it proves to be much
faster then the first design. (although I do prefer the first design
if the speed improvements is minor).

Any ideas?

Thanks in advance
--
Haim