public key size

Hi, everyone! :)

On Thursday, August 07, 2003 02:58, auto97841-revL73yDgGBWk0Htik3J/[email protected] wrote:
> 'lo hez, nop, et al
>
> Two comments:
>
> - First comment, wrt: Why 2048/128 vs 4096/256?
> Simple answer: Computational resources required.  ...

	OK. Here's something I keep noticing. People group 2048/128 and 4096/256 
together. I'm saying, "If we're going to do 2048 bit ElG keys, why not do a 
256 bit AES key anyway?" i.e. 2048/256 . While there _is_ a significant 
difference in computational resources used by a 4096 bit ElG key over a 2048 
bit one, the difference between 256 bit AES and 128 bit AES, isn't even 
noticeable to me. (Please remember everyone that I run a 333MHz CPU. If it's 
not noticeable, that's saying something.)
	Allow me to be more specific: I encrypted a 16 megabyte block of data with a 
256 bit AES key and a 128 bit AES key. On average they were about the same. 
There was what appeared to be a subsecond difference. (Mind you I had to use 
the somewhat slow Python crypto library I've been using for the Python code 
so far. However, the AES backend it written in C, so that helps.)
	Then I tired a 30 megabyte block. That took the 256 bit key 9 seconds on 
average and the 128 bit key 8 seconds on average.

	Now most messages are not going to be 30 megabytes. And a one second 
difference every thirty megabytes won't matter. (Not to mention that a _good_ 
crypto library could probably get better stats than the Python thing I'm 
using.) So the speed difference really isn't relevent. As for the size the 
AES key takes up in the message, it doesn't matter. The ElG key encrypted 
plaintext can be up to 256 bytes. Even with a 256 bit AES key and 32 byte 
pre-IV, we still have 192 bytes left over. So the size of the key shouldn't 
matter since it doesn't change the size of the message anyway.

	So are there any reasons _not_ to use a 256 bit AES key that I haven't 
thought of here?

> -jrandom
>

	-- Hezekiah

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQA/MhDmeHiZTMH32ioRAogZAJ9jLFp/ZiH2h25KIwkLYESgK30NEQCeNpmT
vjakAHpr7XTZBR8asEID0OM=
=i3qQ
-----END PGP SIGNATURE-----