RE: [Algorithms] lights/shadows rendering architecture

[Tom F. <tom...@ee...>]

There is only 1 GPU, but they have very long pipelines, and they also do =
a
lot of things in parallel. So the front of the pipeline can be trying to =
use
the texture before the end of it has finished, which it can't do, so it =
has
to stall.

TomF.

> -----Original Message-----
> From: gda...@li...=20
> [mailto:gda...@li...] On=20
> Behalf Of Joris Mans
> Sent: 21 September 2005 16:21
> To: gda...@li...
> Subject: RE: [Algorithms] lights/shadows rendering architecture
>=20
>=20
> > That means you'll have a pipeline doing:
> >=20
> > -Render shadowbuffer A
> > *stall*
> > -Use shadowbuffer A
> > -Render shadowbuffer B
> > *stall*
> > -Use shadowbuffer B
> > -Render shadowbuffer C
> > *stall*
> > -Use shadowbuffer C
> > etc.
> >=20
>=20
> Why would the system stall between rendering to shadowbuffer=20
> A and using
> it? Its not like the GPU renders a shadowbuffer which the CPU=20
> has to use
> afterwards. I presume it all gets batched up in the command=20
> queue of the
> GPU, and since there is only 1 GPU the shadowbuffer will be rendered
> before the commands using it will be issued, so no stall, right?
>=20
>=20
> Joris