1 Feb
2012
1 Feb
'12
6:12 p.m.
One of the things I loved about PowerPC was that the
endianness was switchab$
Please don't use paragraph-length lines.
Compensating,
One of the things I loved about PowerPC was that the
endianness was
switchable at run-time (though I'm given to understand that never
worked all that well) [...]
Not surprising; it was never designed all that well.
I once spent some time reading through the documentation for a
particular PowerPC CPU (the 750, I think it was). Little-endian
support was...incomplete, I would call it. Put the CPU into
little-endian mode and you don't get a little-endian CPU; you get a
big-endian CPU with some swizzling in the memory path, and the
difference shows. It looked to me as though they wanted to be able to
share data structures in memory between code running big-endian and
code running little-endian without having to byte-swap when crossing
the endianness boundary, and considered that more important than making
little-endian mode a self-consistent little-endian machine.
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B