[mythtv] Slow keyboard annoyance
ijr at case.edu
Sun Feb 26 23:14:42 UTC 2006
On Sunday 26 February 2006 17:20, Lorenz Hahn wrote:
> Isaac Richards wrote:
> > On Sunday 26 February 2006 11:30, Lorenz Hahn wrote:
> >>- A "fast" card is a workaround to the "bounce effect"
> >>- The lack of a fast card is not the cause, otherwise the "bounce
> >> effect" would not be able to disappear.
> >>- The sum of "high load" and something like a race condition causes
> >> the odd behaviour.
> >>- If a 1,8 GHz CPU has high load because one is navigating through a
> >> menu. There may be the possibility of a optimization.
> >>- It took 97 seconds and 100% CPU to iterate over 158 Recordings in "All
> >> Programs". If it's a graphic-card issue, why is the CPU at 100%?
> >> Shouldn't this be waiting for some DMA or other IO without CPU-usage?
> > Because the X server (not myth) is falling back on completely
> > unaccelerated code to do the drawing. None of the graphics drawing code
> > is in myth.
> supposing there is no way to keep the X server from falling back to
> unaccelerated code: why are the mythtv menues the only affected ones?
> Even if the drawing code runs incredible long. Why does it affect the
> processing of keystrokes? Let's say it's the X server. Again, I can't
> reproduce this "bounce effect" in any other application.
There's a big difference in the amount of work necessary to draw lines, filled
rectangles, and some tiny little images with no actual transparency (ie, a
normal user interface on a computer) vs lots of big images with real
transparency being blended around (ie, what mythfrontend does).
I think that's fills my quota of support for 8 year old, obsolete hardware on
the wrong mailing list for the week.
More information about the mythtv-dev