GLX available?

[NoiseEHC]

Oh, I did it in May:
http://lists.laptop.org/pipermail/devel/2007-May/005227.html

Got no response but now I know that only Jordan Crouse is reading this 
list from AMD and he is concerned with the X driver and not the Geode 
core so no luck.
Probably you (the OLPC team) should push AMD to document the 
FPU/3DNow/MMX pipeline since my measurements show that the databook is 
about a hypothetical processor which does not exists in silicon (of 
course I can be wrong). Without this I can guarantee you that no 
optimization effort will succeed. See:
http://gcc.gnu.org/ml/gcc-patches/2006-08/msg00452.html

Also it would be nice if they would write something about the L1/L2 
cache and the branch predictor. For example the last thing which I tried 
to measure is the L1 miss -> L2 hit penalty but could not do it. It 
would be much easier if they would document it (write down a number, 
that is it)...

Ivan Krstić wrote:
> On Jan 7, 2008, at 7:11 AM, NoiseEHC wrote:
>> The databook of the LX is so low quality that hand optimization requires
>> testing everything
>
> Please feel free to ask questions about the Geode on this list where 
> you find the published documentation inadequate. We have some great 
> AMD folks here, with a long history of being helpful and friendly to 
> us above and beyond the call of duty. They will assist you if at all 
> possible.
>
> -- 
> Ivan Krstić <krstic at solarsail.hcs.harvard.edu> | http://radian.org
>
>
>