fra81

[fra81]

Yes, indeed.

In general, when streaming a Windows desktop, RDP has access to the internal OS graphics primitives. But this doesn’t occur when streaming the NoMachine session, whose content is a Linux screen that is rendered remotely.

RDP is not able to stream this type of content as efficiently, so it’s not a NoMachine problem. This is a performance issue that must be solved in RDP.

[fra81]

Yes, getting the dongle should be the easiest solution to have GPU acceleration, but probably you are good to go with the llvmpipe solution, so I’d suggest to wait for your user 😉

[fra81]

Hi,

I think this Feature Request describes what you need:

https://www.nomachine.com/FR12K02799

Check the box to be notified when implemented 😉

[fra81]

Hi,

the fact the X server starts up doesn’t mean necessarily that rendering actually happens when no montitor is connected to the GPU. You could try one of those dongles that simulate a monitor. Try to search for “headless hdmi dongle”.

Or you could give one more try with the NoMachine’s virtual display server. In this regard, the following article should be more appropriate to address your issue:

https://www.nomachine.com/AR01Q01012

[fra81]

Hi,

this is not expected in general, but consider that when you disconnect without logging out, desktop environement and applications continue running, doing their stuff and possibly making requests to the X server (which is built in the nxnode process in this case).

If you feel the CPU usage is not justified anyway, you can try to gather more info on what is happening with the following commands:

# strace -f -p <pid>

# top -b -n 1 -H -p <pid>

Where <pid> is the process ID of the nxnode process which is consuming CPU. You can send the outputs of those commands to forum[at]nomachine[dot]com. Please send along also server side logs gathered as explained in https://www.nomachine.com/AR10K00697.

[fra81]

However, for NoMachine got a new error: “NvEncode: ERROR! Error is 15, ‘Invalid version’.”

This will be fixed in the upcoming software update. It is a compatibility problem with most recent drivers (and the Trouble Report you mentions is indeed related). However this error should not prevent sessions from working correctly.

[fra81]

I was unable to get NoMachine to start my desktop environment. I tried adding unix-xsession-default to AvailableSessionTypes in node.cfg and server.cfg, and also setting DefaultDesktopCommand to “/usr/bin/dbus-launch –exit-with-session /usr/bin/mate-session”, but mate-session couldn’t connect to DBus.

It should work out of the box. Please check this article for some common issues with the desktop environment:

https://www.nomachine.com/AR08N00898

[fra81]

It could be a problem with slow access to video memory. We happened to observe such behaviour in the past on headless machines, but it could also depend on the drivers. Please attach the output of glxinfo.

You can also test what would happen if you use the virtual framebuffer provided by NoMachine instead of the “physical” output. In order to try that you have to turn off the graphical environment on the server (‘sudo systemctl stop gdm‘ or ‘sudo systemctl stop lightdm‘ or any other command which is suitable for the display manager in use) and then restart the NoMachine service (‘sudo /usr/NX/bin/nxserver –restart‘). So NoMachine will create a virtual framebuffer which you can connect to and test performance.

[fra81]

Hi,

if you are connecting to the “physical” display of your Ubuntu machine, it is a matter of configuring your X.org server. For example in this link you can find a few hints on how to add virtual monitors: https://askubuntu.com/questions/453109/add-fake-display-when-no-monitor-is-plugged-in. For example I tested myself successfully the configuration with the Intel driver (VirtualHeads option).

If you are using one of the NoMachine products providing virtual desktops, the implementation of this Feature Request should address your case: https://www.nomachine.com/FR12K02799.

[fra81]

Hi!

I start to say that this is strange. Given that server and client should settle, more or less, for a frame rate of 30 fps by default, added lag should be no more than 1000/30, that is 33 ms.
Please check the Display settings panel in the session menu. What value frame rate is set to?
And can you describe exactly how you are measuring such input lag? What procedure, what measurement method?

That said, are you connecting to the physical display of the server? Is it a headless machine? I would also check cpu usage on both client and server side.

[fra81]

Hello,

please gather logs from the server as explained in https://www.nomachine.com/AR10K00697. If possible, also client side logs could be useful. Instructions here: https://www.nomachine.com/DT10O00163#2. You can send everything to forum[at]nomachine[dot]com.

In the meanwhile you could try to uncheck the ‘Use acceleration for display processing’ option in Server preferences -> Performance tab.

[fra81]

Hi,

in order to detect what is the bottleneck, network or hardware resources, for a start I would:

– check CPU usage on both client and server side;

– gather NoMachine session statistics as shown in https://www.nomachine.com/DT10O00157&dn=session%20menu#9.

 

Please also make sure you are using H.264 encoding. More info in this article:

What do I need to do to allow NoMachine software to use H.264?
https://www.nomachine.com/AR10K00706

[fra81]

Hi,

indeed it is a very strange issue!

Please try checking the ‘Disable client side image post-processing’ option in the session menu and send us a screenshot if possible.

[fra81]

Hello,

thanks for the logs. It seems your desktop is now using the Wayland compositor. It’s not clear whether, before using the new NVidia drivers, the old desktop was doing so. We are definitely trying to reproduce the problem in our labs, so that we can work on this Wayland issue. In the meanwhile, can you try to disable Wayland and check again? Just to see if it is Wayland-specific or a more general problem with this NVidia driver.

To disable Wayland:

– set WaylandEnable=false in /etc/gdm3/custom.conf;

– run systemctl restart gdm3

• This reply was modified 6 years, 10 months ago by fra81.

[fra81]

Hi Mick,

unfortunately we weren’t able to reproduce this issue even with Virtuoso. Can you please provide more info:

– What is Cadence Virtuoso version?

– Do you have NoMachine version 6.3.6 on both sides?

– Do you have the possibility to verify if the issue is reproducible when connecting from a different OS than Mac?

– Does the issue occur also in a NoMachine virtual desktop?

– Does the issue occur when opening a specific dialog or with any dialog?

– If the issue occurs with a specific dialog or a specific sequence of actions, can you provide step-by-step instructions?

– It would be nice to have a longer video that would show all the steps.

[Author]

Posts