Use Reddit? Join us on our very own subreddit: https://www.reddit.com/r/Linuxstuff/
You can sign up to get a daily email of our articles, see the Mailing List page!
Support us on Patreon to keep GamingOnLinux alive. This ensures we have no timed articles and no paywalls. Just good, fresh content! Alternatively, you can support us on Paypal and Liberapay!
  Go to:
Unreal Engine 4 Linux demos download links are dead
bintsmok commented on 6 February 2018 at 3:40 am UTC

dubigrasuGo to ...Matinee Fight Scene Demo/MatineeFightScene/Saved/Config/LinuxNoEditor and add these two lines to the Engine.ini file:

[/Script/Engine.Engine]
bSmoothFrameRate=false

Actually, that file is blank. But I added what you said.

It now exceeded 62 FPS. Using an RX 480 8GB, I get a max of about 70 FPS and an average of about 60 FPS when I used these arguments:

-resx=1920
-resy=1080
-novsync
-fullscreen
-notexturestreaming

Can you try with your hardware?

By the way, for reference, here are the results from Phoronix when running the Matinee Fight Scene demo:

GTX 780 Ti 3GB = Avg @ 40.34 FPS / Max @ 68.86 FPS
GTX 970 4GB = Avg @ 36.58 FPS / Max @ 70.88 FPS
GTX 980 4GB = Avg @ 38.59 FPS / Max @ 67.48 FPS


Thanks a lot.

dubigrasu commented on 6 February 2018 at 9:35 am UTC

bintsmokCan you try with your hardware?

By the way, for reference, here are the results from Phoronix when running the Matinee Fight Scene demo:

GTX 780 Ti 3GB = Avg @ 40.34 FPS / Max @ 68.86 FPS
GTX 970 4GB = Avg @ 36.58 FPS / Max @ 70.88 FPS
GTX 980 4GB = Avg @ 38.59 FPS / Max @ 67.48 FPS
Thanks a lot.

Here are the results with Nvidia GTX 780 and AMD HD 7970 (I just swapped the card and rebooted in Ubuntu):
https://openbenchmarking.org/result/1802063-AL-1802069TY16

Impressive results from AMD, considering that the HD 7970 is a weaker card compared with GTX 780.

bintsmok commented on 7 February 2018 at 5:57 pm UTC

dubigrasuI think Alt-Enter is for fullscreen mode or use the -fullscreen argument when launched in terminal.
The demo is not a benchmark in itself, you need to launch it through a shell which will load libframetime, let it run for about 70 sec or so, and then kill it. Then you can collect the data.
The most convenient solution is to run it through phoronix suite, which will take care of everything.
Yes, you can also use a custom shell, but with phoronix suite is a more complete/fast solution (graphs and everything).

Just found a post on reddit that showed a benchmark using modern hardware. The UE4 Matinee Fight Scene demo can log frame times without the need for libframetime or Phoronix Test Suite. From the frame time data, the frame rate can be derived. The results are much more detailed.

dubigrasu commented on 7 February 2018 at 7:18 pm UTC

bintsmok
dubigrasuI think Alt-Enter is for fullscreen mode or use the -fullscreen argument when launched in terminal.
The demo is not a benchmark in itself, you need to launch it through a shell which will load libframetime, let it run for about 70 sec or so, and then kill it. Then you can collect the data.
The most convenient solution is to run it through phoronix suite, which will take care of everything.
Yes, you can also use a custom shell, but with phoronix suite is a more complete/fast solution (graphs and everything).

Just found a post on reddit that showed a benchmark using modern hardware. The UE4 Matinee Fight Scene demo can log frame times without the need for libframetime or Phoronix Test Suite. From the frame time data, the frame rate can be derived. The results are much more detailed.

Well, for a manual run you can use that, but when you want to obtain meaningful results you'd need to do multiple runs until you reach a certain threshold in variations, which often needs automated runs in the exact same conditions/duration/sequence. With libframetime/voglperf you can automate that and obtain the same detailed results (percentiles etc).
You could probably(?) find a way to automate the benchmark with their own FPS charts tool, but I haven't looked much into that.

In this particularly case (Matinee demo) is not much of a problem because it runs in a loop, you can start the bench manually and let it loop as many times you want so you can get a good average. Not so the case with other demos/games etc, many don't even have the console option.

For my benchmarks (when I used to do that) I often did up to 12 runs to get a real average. Imagine doing this manually Phoronix does it clever, you set (from its config) a certain threshold and it keeps running until that value is reached.

In short, the reason I mentioned libframetime and voglperf is because they are universal tools, regardless of the engine or game. I should've mentioned the FPS chart tool, but it slipped my mind since I'm not using it.
(btw, try stat UnitGraph in console for a real time cool performance graph, or stat unit for more detailed perf info)

Kuduzkehpan commented on 7 February 2018 at 7:55 pm UTC

I noticed that my computer has lower fps in games with unity 7 desktop
but has more fps with gnome-shell.
btw i read about mutter doens't run apps at fullscreens but runs at borderless windowed.
still very good results for ATI AMD.

  Go to:

Due to spam you need to Register and Login to comment.


Or login with...

Livestreams & Videos
Community Livestreams
  • RPGoodness: „The Dwarves“
  • Date:
See more!
Popular this week
View by Category
Contact
Latest Comments
Latest Forum Posts