It sounds like you saw my correction. I typed 3DES when I meant DES, so I'll reply to your comments DES.
> DES doesn't go back nearly as far as 1972. (nor does DES for that matter)
Below is the official NIST paper describing DES. You'll note that after a four year approval process, DES officially became a government standard in 1977. As described in the paper, IBM was using it by 1974 after it was developed in the years prior.
http://nvl.nist.gov/pub/nistpubs/sp958-lide/250-253.pdf
You could reasonably choose any year between 1972-1977 as the beginning of DES usage, so you're mistaken about "not nearly as far back as 1972, sorry.
> rather a large number of milliseconds
Try cracking a password database sometime. I do this stuff for a living. The larger the database, the faster you'll get working passwords, so we'll give you the benefit of the doubt and use a fairly small database of only 1,000 accounts as an example. We'll also be generous to you and not use a rainbow table. With a small (difficult) database like that, you can expect to get maybe 12 passwords in the first second or two. In the first ten minutes, probably 250 working accounts.
A 100X larger database will yield roughly 100X as many passwords per time - around 1,000 working accounts in the first few seconds, or 2-3ms per account at first.
If we want to go fast, we use a rainbow table. Standard DES password hashing ala crypt() collides at about 1:1000 since it uses only the first eight characters. On modern PCs with GBs of RAM, we can use in-memory tables and crack millions per second. No need for that, though, I don't mind waiting several milliseconds.