Forum Replies Created
-
AuthorPosts
-
fra81Moderator
Thank you for the info. Actually they exclude that your are experiencing the same problem we had found (and that will be fixed anyway in the next release). In fact from our observations the problem we found has a very limited impact.
Unfortunately we are not able to reproduce any further problem on any of the machines we tested on. Is it possible that the regression you observed is triggered by a system update? Could you maybe try to uninstall the present version and try version 5.0.47 again in order to see if you get the same results?
Another possible explanation is if you had enabled the H.264 codec with the previous version by installing the NoMachine AVC Pack or by compiling the encoding library by hand. Is this a possibility?
Please also take a ‘sample’ of the process that is consuming the CPU. You can find the option to take a sample in the Activity Monitor utility.
January 28, 2016 at 10:37 in reply to: NoMachine hangs when entering small dialogs in Java application #9826fra81ModeratorShould I assume that you are running the Java application in a custom session in ‘floating window mode’? In that case the rendering is delegated to Xquartz, and so it actually depends on the Xquartz implementation. In virtual desktops instead the rendering is handled internally by NoMachine and so it is platform independent. Work on removing such dependency from Xquartz in floating window sessions is something we have already in our development roadmap.
fra81ModeratorHi,
you might be missing to set ‘EnableDisplayServerVideoCodec 1’. Also restarting the server (‘/usr/NX/bin/nxserver –restart’) could be necessary. However you can use the graphical interface to do those settings:
Click on the NoMachine tray icon -> Show the server status -> Server preferences -> Performance tab
fra81ModeratorDragging the app to the trash (and emptying the trash) is enough for a complete uninstallation. You mean that you reinstalled by using directly a 5.0.58 package?
Can you please send the output of this command:
$ ls /Applications/NoMachine.app/Contents/Frameworks/lib/
fra81ModeratorWe managed to reproduce the issue in our labs and we are working on it. It seems it is triggered by the software update process. Please check if uninstalling the software and proceeding with a fresh install solves the problem.
fra81ModeratorYour server should have dumped a core file (if enabled to do so) in the user’s home directory. Please locate it and get a backtrace from it as per these instructions: https://www.nomachine.com/AR09L00810.
Also it would be useful to know the screen data acqusition method the node is currently using. To do so please gather the logs by following the instructions in https://www.nomachine.com/DT07M00098.
You can send them to forum[at]nomachine[dot]com.
fra81ModeratorAt the moment there is no configuration in NoMachine for selecting a specific GPU. Doesn’t Nvidia settings software allow to assign a GPU to applications in your system?
Can you tell us something more. Are you using this system as a client or a server? Are you concerned about accelerated encoding/decoding?
fra81ModeratorHi all,
we would need the diagnostic reports requested by Britgirl to confirm it is the same problem described in the Trouble Report. If you are willing to test the possible fix for this issue, please contact us to forum[at]nomachine[dot]com.
fra81ModeratorHi mopac!
Do you have NoMachine version 5 installed? Connecting to headless machine should work out of the box thanks to this implementation:
https://www.nomachine.com/FR10L02842
If you have this problem with version 5, please gather server side logs as exaplained in https://www.nomachine.com/DT07M00098 and send to forum[at]nomachine[dot]com.
Greetings
fra81ModeratorIt is not possible at the moment, but it is something we are working on. It will be possible to encode a single monitor or even a portion of the remote screen selected by the user.
fra81ModeratorHello,
thank you for reporting and for the very accurate analysis!
We have reproduced this issue in our labs and opened a Trouble Report to track it:
fra81ModeratorThank you for letting us know!
The patch will be released with the next update.
You can track the status of this Trouble Report here:
fra81ModeratorI can’t find any documentation for how to enable this feature.
No configuration is needed. NoMachine is designed to select automatically the encoding method that will provide the best performance among the available ones. The hardware encoder will be chosen as the preferred one when supported.
Does it work with AVC?
It will work with any software or hardware decoder you may have on client side, NoMachine AVC pack included.
Does performance update rate improve if I use AVC and a server side graphics card?
Using the hardware encoder on server side can improve performance and most importantly will offload the CPU. AVC on client side should not be needed if you are running a Windows or Mac client. NoMachine will use anyway the hardware decoder that is provided on your client computer.
Lastly, will this work with Kepler based GPUs on AWS EC2?
Unfortunately only Maxwell based (or next ones) GPUs are supported. Hardware encoder is not used in Kepler or older GPUs due to their insufficient feature level.
fra81ModeratorHi Ventec,
of course even on a slow network, a 15 seconds delay would have been very difficult to explain, so something must be very wrong.
We didn’t receive your email. Did you send it already?
fra81ModeratorWould you exclude a bandwidth problem exacerbated by the more graphics-intensive desktop environment of CentOS 7? In this scenario a test with the lightweight desktop (Xfce) would help to clarify.
Another possibility is a problem with video drivers, like in the other cases you mentioned, even though the CPU is not overloaded in your case. This could be verified by running a debug library. If you are willing to test this library, please contact us to forum[at]nomachine[dot]com. We will send you the library and instructions.
-
AuthorPosts