Re: TEA (was Re: filesystem encryption)

Peter Gutmann (pgut001@cs.auckland.ac.nz)
Thu, 25 Jun 1998 12:54:42 (NZST)

• Next message: Gary Howland: "Blowfish II name"
• Previous message: burt rosenberg: "Re: timing attacks."
• In reply to: Mike Rosing: "Re: timing attacks."
• Next in thread: Pat Ekman: "Re: filesystem encryption"

>>SAFER
>
>I need a [...] *lightning* fast crypt,

Those two bits quoted above... they don't fit together. Safer is a nice design
and works well in a smart card (the application it was designed for), but it
runs like a heavily sedated sloth. Even without trying to implement it, you
should be able to see that an algorithm which uses exclusively byte-oriented
ops is never going to be anything close to "lightning fast" on any modern
processor. If you want some timing comparisons, take a look at the code from
cryptlib, http://www.cs.auckland.ac.nz/~pgut001/cryptlib/, which includes Safer
alongside various alternatives like Blowfish, CAST, RC5, and Skipjack (the last
one isn't in the currently available release yet, my keyboard has only just
stopped smoking :-).

Peter.

• Next message: Gary Howland: "Blowfish II name"
• Previous message: burt rosenberg: "Re: timing attacks."
• In reply to: Mike Rosing: "Re: timing attacks."
• Next in thread: Pat Ekman: "Re: filesystem encryption"