[Question] Why is the entry binary itself zapping all that CPU?
IngwiePhoenix opened this issue · 6 comments
For context, a snippet from htop:
PID PPID USER VIRT RES SHR S CPU% MEM% Command
9075 8976 nagios 2283M 46720 6064 S 95.9 0.6 /usr/bin/neko serve -d --static /var/www
(ignore "nagios" as the user - htop is being stupid. It matched the docker UID to a local UID...)
I just would like to know why it's zapping away sll of that CPU runtime? I mean, technically, it should only need to watch over a few processes - xorg, firefox-esr, ... - but yet it's needing all this CPU. Also - nobody is connected.
This is just a question of curiosity :)
2 words: Software Encoding.
It has to encode the stream entirely in software, which takes up a chunk of cpu.
Ohh, I see. Thanks!
What would the efford be to add hardware accellerated rendering? It wouldn't make a difference on my server since it's an iGPU anyway, but I'd love to know ^^
We've been exploring the possibilities,
but haven't booked any results yet.
We know it's possible, the road to get there is just unclear
Ahh, fair enough.
Good luck on the journey of implementing it then! ^_^
@IngwiePhoenix This is actually a bug, because of outdated docker image, see #115 (comment).
Heh, seems I missed the bit that said "Also - nobody is connected." at that time.