FPS test 1224

FPS test 1224

Postby Rudi De Vos » 2019-03-10 11:11

I'm current trying to get most FPS when using uVnc over LAN.
Need HD and possible also 4K test result on win 10

winvnc:
PowerMode is activated when you set Max Cpu: 100
vncviewer:
[v] preemtive Updates
[v] u2
jepg (6 or 7)
The viewer info show the current Ffp
Youtube testvideo (4K Video Ultra HD 60fps) https://www.youtube.com/watch?v=R3AKlscrjmQ&t=193s

ps: There is a big difference bewteen 2 monitors on the same card and each on his own.
On the same card are handled serialized, 2 cards are handled simultaneous.

Thanks for the feedbacks.

Download: ( 32 /64)
This is the current 1224
-server:
with activated powermode when max cpu is set to 100
-viewer:
FPS added in the viewer info window
-ddengine:
change capture for one monitor per card ( using more it's the same code as before)
flahsing cursor when hovering over video fix

https://www.uvnc.eu/download/1224/FPS_test.zip
Rudi De Vos
Admin & Developer
Admin & Developer
 
Posts: 5956
Joined: 2004-04-23 10:21

Re: FPS test 1224

Postby Rudi De Vos » 2019-03-10 11:12

Result:
Cpu: i7
Desktop: (1920x1200)
LAN: ( 1Gb 10GB loopback)
video card (s): (Nvidia GTW 1050, Intel HD 630)
multi monitors: 2, one on each card
FPS: 37 (loopback)
Rudi De Vos
Admin & Developer
Admin & Developer
 
Posts: 5956
Joined: 2004-04-23 10:21

Re: FPS test 1224

Postby Andrew Henry » 2019-03-11 16:42

Cpu: i7-6820
Desktop: (1920x1080) Intel 530
LAN: 1Gb
video card (s): (AMD FirePro W5170M, Intel HD 530)
multi monitors: 3, two on AMD, one on Intel
Encoder: Ultra2
Speed: 20000 kbits/s
FPS: 22

Same setup with desktop: (1600x1200) AMD FirePro
FPS: 22, no significant difference

Do the server settings make any real difference? As long as it is set to show video on the full resolution that VNC viewer shows maximized at the desktop resolution?
Andrew Henry
 
Posts: 5
Joined: 2013-04-17 21:29

Re: FPS test 1224

Postby Rudi De Vos » 2019-03-11 20:37

Retested an the only difference is the powermode activated by settings the server to cpu = 100
Internal, when 100 is detected other parameters are used to incease the FPS.

Yes, the server option make the difference.
Rudi De Vos
Admin & Developer
Admin & Developer
 
Posts: 5956
Joined: 2004-04-23 10:21

Re: FPS test 1224

Postby WKjun » 2019-04-05 10:15

Hello!

First of all I want to thank you for you efforts - UltraVNC is the perfect tool for me to stream music videos and some movie clips in realtime as smooth as possible from one PC to another over 1Gbit LAN.

Therefore my goal is a steady fps of 25-30, depending on the material being PAL or NTSC. But it is not the fps alone that counts! Some settings seem to me counter-productive although they should speed things up further more.

The best settings I could find are:

Server:
Max CPU: 100
Turbo: ON

Viewer:
u2, Full Colors
[v] Use CopyRect encoding
[ ] Use Cache Encoding
[v] Zip/Tight Compression: 9
[v] Jpeg (Tight) - Quality: 9
[ ] Preemtive Updates

  • CopyRect, as description tells, is good for moving parts of the picture. So I think it helps smooth it out.
  • Reducing Zip or Jpeg to lower (or extremely low) numbers does not speed up things any more - no bottleneck there.
  • Preemtive Updates is one of that options in my case which seem to introduce a bit more stuttering, for example a car drives by.

That said, I should tell you that I still use Win7 & Mirror Driver because I couldn't get the same performance on Win10 and its Desktop Duplication API. Firstly I thought the API may be not included in Home Edition because it lacks remote desktop functions. But on Win10 Pro it was the same. And for practical reasons my desktop resolution is 1600x1200, not higher.

On Win7 I am 99% satisfied with the results so far. Especially powermode helped! :thumbs:
WKjun
 
Posts: 4
Joined: 2019-03-03 10:45
Location: Austria

Re: FPS test 1224

Postby WKjun » 2019-04-13 00:02

I just want to say that the latest version excels on Windows 10! Full-HD, completely fluid, and that between two not-too-fast AMD FX-9590 PCs.

To get the most out of it, I have tried a lot of combinations of settings, but basicly all your recommendations were perfect. Interesting that different behavior on Windows 7 though.

Desktop Duplication API is - of corse - of paramount importance; now I have proven to me that it works on Win10 Home as well as Pro.

Important: Power plan options in Windows have to be set much higher than 5% (Minimum processor state) for the FX-9590. 50% helps, 75% even more and on 100% it is perfectly fluid. Either this is common for (slower) CPUs or the FX series in particular, because of the unoptimized Windows Scheduler for that type.

u2, Full Colors
[ ] Use CopyRect encoding
[ ] Use Cache Encoding
[v] Zip/Tight Compression: 6
[v] Jpeg (Tight) - Quality: 9
[v] Preemtive Updates

After setting CPU to 100% (on both ends) and checking Preemtive Updates, it is a dream come true, even on max. Jpeg Quality 9!

One question: Shouldn't Zip/Tight Compression be off or as low as possible for this low-latency situation? When comparing screenshots I cannot tell any difference between 1 and 9, while unchecked there is a slight difference, although nearly undeterminable.

Thanks and best regards!
WKjun
 
Posts: 4
Joined: 2019-03-03 10:45
Location: Austria

Re: FPS test 1224

Postby Rudi De Vos » 2019-04-13 20:53

zlib/tight/zrle encoding use zlib/jpeg
u2 is a pure jpeg decoder
Rudi De Vos
Admin & Developer
Admin & Developer
 
Posts: 5956
Joined: 2004-04-23 10:21


Return to 1.2.2.x

Who is online

Users browsing this forum: No registered users and 1 guest