[GTALUG] Intel GPU support for HDMI 2.0 (UltraHD)

Lennart Sorensen lsorense at csclub.uwaterloo.ca
Tue Feb 11 13:02:03 EST 2020


On Tue, Feb 11, 2020 at 09:39:29AM -0500, D. Hugh Redelmeier via talk wrote:
> OK.  Most but not all TV sets are fine for this.  I would not trust
> the RGBW displays but I haven't tried them.

My TV is RGBW, but it is OLED.  Some RGBW LCD panels from LG in the past
did not actually have RGB subpixels for every pixel which was a problem.
The OLED ones do have RGBW subpixels for every pixel (so 32M subpixels).

I would not use it as my computer display, although I am sure it could
do it just fine.  I do have a mythtv frontend connected to it, and due to
HDMI 1.4 limitations on that machine, I have to force it to 1920x1080 at 60Hz
rather than the 3840x2160 at 30Hz it likes to default to since my content
on mythtv is never more than 1920x1080 but is often 60Hz.

> Also: go for IPS or VBA technology.

For a computer monitor, sure, but for my TV, not way.

> If you are letting your monitor do scaling you are doing it wrong.
> You should let your computer do that.

Well sure but if your HDMI link is too slow, letting the screen do it
is a better option than having terrible frame rate.

> One exception: if a computer only does 1920x1080, you can let a
> TV/monitor double the pixels in each dimension.  This is dumb in the
> long term but sometimes you need to do it for a short time (eg. to
> adjust firmware settings in a server).  (It's really annoying to not
> be able to see text during POST and the subsequent startup.)
> 
> Scaling TV or movies is an interesting problem since it extends into
> the 4th dimension (time).  You really don't want to get into that with
> a monitor.  In fact, you want to turn off any multi-frame processing
> that a TV does because it will add latency to the display.
> 
> True.  Again, this is why my talk was "what I've learned about
> UltraHD".  Actual experience is enlightening.  Reading specs is
> important but not sufficient.  Ergonomics is full of surprises.  One
> of them is: not everyone is the same.  That's why I tried to frame my
> talk as about me :-)

The specs are often lacking in details.

-- 
Len Sorensen


More information about the talk mailing list