[MPlayer-users] Image too small, temporary buffers cannot function

[Stephen Mollett]

Hi,

On 25/11/2019 23:58, Scott W. Larson wrote:
> This is not what I am seeing on my 4 GHz four core CPU. When I use 
> software rendering ("-vo xv -vc ffmpeg12") with a 2160p display, mplayer 
> uses five times more CPU and, much worse, Xorg suddenly becomes nearly 
> CPU bound. As you would expect, the video is jerky especially whenever I 
> do anything else on the system.

It sounds like something else is wrong there. I'm assuming the MPEG-1/2 
stream isn't a 2160p video (can MPEG-1/2 even handle that resolution?) 
so the decoding CPU usage should be negligible on a modern system.

I've got the impression over the years that -vo xv is generally 
considered to be outdated. Have you tried using the VDPAU output device 
with software decoding ("-vo vdpau -vc ffmpeg12")?

This is the configuration I use on my NVIDIA-based systems - I get the 
advantage of the fantastic spatial-temporal deinterlacer without any of 
the disadvantages (mainly quality issues and poor handling of damaged or 
non-compliant streams) of the hardware video decoder. (I used the 
hardware decoder for a while after I first got a capable GPU but soon 
started noticing the problems.)

As far as the CPU's decoding capability is concerned, my 
rather-underpowered media PC has a 1.8GHz dual-core Athlon Neo CPU which 
is powerful enough for 1080p or 720p50 H.264 without hardware 
assistance. Hardware decoding via VDPAU does work on its Radeon HD 3200 
IGP but stability is poor - even seeking can crash it - and although it 
claims to have deinterlacing capability in vdpauinfo, I've never been 
able to get it to work.

Stephen