[sldev] Cache politics: performance vs obfuscation

[Laurent Laborde]

2008/6/9 Jason Giglio <gigstaggart at gmail.com>:
>
> Who cares?  A small utility anyone can write in a few hours can do this
> today.
>

*NOT* anyone, just "any coder willing to spend some time on it".
Writing a software to decode the current cache isn't that much complex
than decoding a XOR'd texture cache.

A lot of software protection can be unlocked replacing a JNZ with a
JNC instruction, extremly simple, but most people simply can't do
that.

But the question is : does the current obfuscation have any impact on
cache performance ?
As far as i remember, we agreed that the current cache have a very
little impact on performance and most of the cpu time is spent in
decoding j2c.

The current cache is working, have a very little perfomance impact,
and provide enough protection against script-kiddies.
The probleme is in the jpeg2000 decoding performance.


-- 
F4FQM
Kerunix Flan
Laurent Laborde