Re: Generate a one-time pad from say a 256bit key?


David Wagner wrote:
Bill Unruh wrote:
As you point out, AES, the suggested alternative, is slow.

I'm not sure what your definition of "slow" is, but I wouldn't
characterize AES as "slow". It seems to be good enough for most
purposes.

More importantly it's "fast enough" for most purposes.

The C reference code gets ~16 cycles per byte on an AMD64. At 2.6Ghz
(my sweet sweet workstations clockrate) that's 154MiB/sec. My network
is 100Mbit and my RAID-5 can only sustain ~30MiB/sec. So even if AES
is "slow" it's not the bottleneck in ANYTHING I'd be doing here other
than artificial in cache benchmarks.

If you were maitaining a multi-gigabit switch or something you'd be
using hardware AES anyways. In that case, multi-gigabit AES already
exists.

So Unruh clearly needs to be beaten with a sufficiently large foam clue
bat.

Tom

.

Relevant Pages

  • Re: The importance of IVs
    ... AES candidates had to have 128 bit blocks. ... Are there specific purposes where AES is better than Blowfish? ...
    (sci.crypt)
  • Re: Job Finished - Adacrypt
    ... They serve different purposes. ... If RSA were faster than AES, ... In other words, do private-key ciphers have desirable properties, other than speed, that public-key ciphers lack? ...
    (sci.crypt)
  • Re: Generate a one-time pad from say a 256bit key?
    ... purposes. ... (my sweet sweet workstations clockrate) ... So even if AES ... What is the rate you can get for RC4? ...
    (sci.crypt)
  • Re: Generate a one-time pad from say a 256bit key?
    ... (my sweet sweet workstations clockrate) ... Therefore, clearly, we should scrap AES and stick ... RC4 is better as it's a stupid cipher anyways]. ...
    (sci.crypt)