Antonio Huerta wrote :
I did reading about DX11. It appears nVidia implemented their own
hardware solution (CUDO and PhysX),
Well, CUDA and PhysX aren't directly related to DX11, the former makes
the GPU accessible for general purpose computing tasks (if it's
parallelizable), the latter is for physics calculations (as the name
suggests). DX11 will have "compute shaders" however - the equivalent to
Nvidia's CUDA.
and AMD implemented their own
solution (hardware support of Havok engine).
I'm not up to date on this, but last I checked when they announced
support for Havok, they actually meant to optimize it for their CPUs
rather than run it on the video hardware. They've got, after all, all
of those quad core (and maybe sexa-core - do they plan desktop variants
of "Istanbul"?) CPUs to sell, which aren't properly supported by games
to date.
But what's likely to
happen, both will also support OpenCL, which is a software interface
to the graphical effects.
OpenCL is another equivalent to Nvidia's proprietary CUDA (meaning
allowing for general purpose computing on the GPU, not necessarily
graphics) - this time a cross-platform multi-vendor open standard
effort which is headed by Apple, IIRC.
<snip>