Re: additional core draft nits in need of WG attention.

[Matt Johnston <matt%ucc.asn.au@localhost>]

On Tue, Nov 11, 2003 at 05:30:53PM -0500, Bill Sommerfeld wrote:
> 
> Anyone have data on:
> 
> - what proportion of the code size of a "minimized" implementation is
> the actual algorithm code?  i.e., are we talking 1% or 10% bloat here?

Looking at a x86 dynamically linked binary for Dropbear (SSH2 server), the
size is ~110kB with 3DES and AES compiled in (plus rsa, dss, dh-group1,
sha1).

Removing 3DES drops the size by ~5kB, removing AES drops it by ~15kB,
although the AES implementation is quite large.

So the size difference could be in the order of 5%, at least for this
implementation. I guess the size decrease from removing 3DES isn't actually
that significant.

Matt