Xorg Glamo

Thomas White taw at bitwiz.org.uk
Thu Nov 19 13:38:12 CET 2009

On Thu, 19 Nov 2009 12:09:32 +0100
David Garabana Barro <david at garabana.com> wrote:

> I know the bandwith handicap, and I know it's not possible to improve
> that, but having free CPU time while Glamo is drawing should allow to
> calculate things (as next frames) instead of waiting glamo to finish.
> Shouldn't it?

That's right.  All I was saying is that the improvement only applies
for accelerated operations, and that (at the moment) we don't ask it to
do very many of those.  At least, not operations that are large enough
to be worth accelerating.

> Should FIFO patch have some impact on "normal" (Xorg) use?

A limited impact (because of the above), but so far (for me) it
certainly doesn't seem to hurt.  The situation is slightly odd: the
FIFO patch makes the waitqueue patch have less impact (because it's
less useful to be able to wait when the accelerated operations are much
faster), and the waitqueue patch also makes the FIFO patch have less
impact (because we don't mind waiting as long if we can do it without
blocking).  But on the other hand, there's only one Xorg process doing
all the requests.


Thomas White <taw at bitwiz.org.uk>

More information about the community mailing list