mirror of
https://github.com/games-on-whales/wolf.git
synced 2026-10-15 21:12:08 +00:00
[GH-ISSUE #177] FPS drop and high cpu usage in multi client connection while it just streaming #115
Labels
No labels
🧠 1 - 🤖 999
AMD
bug
bug
cannot reproduce
documentation
duplicate
easy
enhancement
enhancement
enhancement
fixed?
fixed?
good first issue
hard
help wanted
Intel
invalid
LXC
medium
moonlight-protocol
Multi GPU
nixos
not a bug
nvidia
pull-request
question
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
games-on-whales/wolf#115
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @SIRMaxis on GitHub (Feb 24, 2025).
Original GitHub issue: https://github.com/games-on-whales/wolf/issues/177
Hello , and thank you for this awesome project.
i have 2x rtx 4090 , and a amd threadripper 7985wx pro
the story is , when i run heroic , without even start any games in it , my cpu usage around 30% ( single session client )
when i run the game it goes around 40% usage ,
second user connect to heroic via android client of moonlight , and do nothing and just stay inside the heroic app , fps drop a lot in stream and i believe in game for no reason , i have ubuntu 24 gnome as host
is there any reasonable explanation for this behavior? my goal is at least i run 2 game per gpu in 60 fps , and i beleive it more than enough with the resource i have and it must not have any issue in this regards.
///not relevent to the issue///
is there anyway that use sunshine as a streamer to the moonlight with this setup ( create seperate docker for each game/app with input control , but stream via sunshine?)
/// /// /// /// /// /// ///
the log of the docker :
htop output cpu usage in 2 session connection :
when one session running game "silent hill remake" in heroic and other one just inside the heroic app:
both running game on heroic + a tird user just goes inside steam app on moonlight via an i phone:
also wolf dont use my other gpu at all , it use same gpu for all sessions of apps :
@ABeltramo commented on GitHub (Feb 24, 2025):
I think there are a few different things going on at the same time here:
config.toml, see the docsI have tried that route a few years ago, and I was hitting all kind of issues. Input separation is a big one, automatic resolution and IP/port handling was another big headache. Wolf is born exactly to implement this multi-user behaviour. 😉
@SIRMaxis commented on GitHub (Feb 25, 2025):
thank you for your detailed reply ,
wolf is amazing exactly and work like a charm.
for gpu load balancing , i have read the docs for manually give the gpu ,
but there is a issue , how i can seperate one app to for example if i have 4 user , it give gpu 1 to first 2 user , and second gpu to other 2
as i saw in the config file (.toml) i belive , i put it manually , but it do not use that renderer gpu for some reason ,
i wanted to run seperate wolfs , but it simply not running ( like give gpu 1 to wolf1 container , and gpu 2 to wolf2 container )
but it return me lot of errors include ports conflict , so i just gave up :D
also im trying to change the base image of each app from ubuntu to something simpler and lightweight like alpine or arch based images.
currently my base host os is ubuntu 24 which is not that much good for gaming , and possibly losing lot of performance here.
btw,
have you tried other os images for apps , like arch for example? if nvidia drivers works on them maybe it can gain some perfomance for games.
@MekayelAnik commented on GitHub (Apr 10, 2025):
Wolf is an amazing project. Huge shoutout to you guys for making this possible.
As I can see @ABeltramo have mentioned that the implementation Sunshine is very complex in this scenario.
But, I want to note a few things:
Thanks again.
Keep up the Amazing works!
@salty2011 commented on GitHub (Apr 10, 2025):
Hi @MekayelAnik ,
Thank you for the feedback,
To start better assist in regards to the performance are you able to provide us with your GPU details? Also when you were testing sunshine was this on Windows or Linux?
This just help create better context for exploring performance issues. For example nvidia’s nvenc performance is far better than VAAPI on AMD. And these differences can be more pronounced between Windows and Linux
@MekayelAnik commented on GitHub (Apr 12, 2025):
Hello @salty2011
In my system:
Wolf Server GPU: AMD RX550 4GB GDDR5 (Mesa Diver Version 24)
Note:
I am very very hopeful on Wolf. Please keep up the good work. If you need any other feedback, please let me know.
@ABeltramo commented on GitHub (Nov 27, 2025):
Since this was opened we've implemented zero copy pipelines which greatly improved performances and we should've now have better support for multi-GPU setups, just setting
WOLF_RENDER_NODE=/dev/dri/renderD129should properly pick and use the right Nvidia GPU. We've also changed a lot of things adding wolf-ui with support for profiles and lobbies, so I'm going to close this issue but feel free to open a new one with the latestwolf:stableif there are still issues present!