Re: [MiNT] Xaaes is slow

["olivier.landemarre@utbm.fr" <Olivier.Landemarre@utbm.fr>]

Selon Adam K³obukowski <atari@gabo.pl>:

> Xaaes is slow.
>
> Machine: Falcon CT60 running 66MHz with 25MHz bus
>
> Software:
> Kernel from CVS
> Xaaes from CVS
> Toswin from CVS
> Naes 2.1
>
> Kernel compilation time: (time make 040):
>
>  - without aes (plain console): real: 10min, user: 9min, sys: 1min
>  - with Naes: real: 11min, user: 10min, sys: 1min
>  - with Xaaes: real: 51min, user: 47min, sys: 4min
>
> Same kernel in all cases. Same Toswin in Naes and Xaaes. Fresh kernel
> sources in all cases. Same GCC, Make, name_unix_tool_here in all cases.
>
> Someone can tell me why kernel compilation under Xaaes is 5 times slower
> than under Naes?
>
> Something seems to be very wrong here.
>

Hello
In any case it can be an AES fault directly, because AES sleep most of time, it
take perhaps around about 2% of cpu time, thats all and it's true for all
multitask AES (NAES, Magic, XAAES, MyAeS), now there is some diffrence beetween
AESs and some softwares that can explain this sort of problems, it's possible
you have software calling a lot the AES with wrong arguments like doing alway's
a MU_TIMER with 0 as time, some AESs try to answer to this and other prefer wait
a little bit. Try to find the software doing this (look perhaps in ACC)

Olivier