[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: MAC speeds
On 9 Mar 2000, D. J. Bernstein wrote:
> Tero Kivinen reports a time of 154000 cycles (308 microseconds) for
> HMAC-MD5 on 2048 bytes on a 500MHz Alpha.
>
> That's slower than today's fast ciphers, notably the AES candidates,
> which are below 40 cycles/byte. Some people seem to be making decisions
> on this basis.
Just a quick note that AES candidates run _much_ faster on a Pentium II
but also on Alpha. For example, my own implementations of RC6, Rijndael,
Twofish and MARS on P2 run with speed 13.9, 14.8, 17.9, 19.1 cycles per
byte, respectively. On Alpha 21264, the best implementations obtain in
average 20-30 cycles per byte - but I also guess they are much less
optimised than mine.
For comparison, Bosselaers' SHA-1 implementation (on Pentium, not on
Pentium II) takes 13.1 cycles per byte. (and RIPEMD-160 --- 15.8 --- is
slower than RC6 and Rijndael).
(As I hope up-to-date) speed Information about AES candidates can be
obtained from http://home.cyber.ee/helger/aes. If anybody knows personally
something that is missing from this table there, please let me know!
Helger Lipmaa
http://home.cyber.ee/helger
Follow-Ups:
References: