Forum Replies Created
-
Posts
-
Hello,
please run this command and show the output:
sudo find / -name libnvidia-encode.so -or -name libcuda.so 2>/dev/nullHi,
in some cases, even if the machine is headless, the GPU could be “turned off” if the monitor is switched off. Can this be the case? If so, could you try to switch on the monitor attached to the server and check again?
Also, to narrow things down, you can try to disable hardware encoding in NoMachine (‘EnableHardwareEncoding 0’ in /usr/NX/etc/node.cfg) and see if you still experience worse performance on Nvidia than on Intel.
Hi,
can you send the whole log where you see “Using H.264 software encoder”? You can pack the whole log directory as explained in https://www.nomachine.com/AR10K00697 and send to forum[at]nomachine[dot]com, if you prefer.
Since the machine is headless, probably it can’t use the GPU. Can you try to “turn off” the local X server on the server machine (i.e. stop the graphical session), so NoMachine will use its own embedded virtual frame buffer? You can find a couple of ways how to do it at points 1) and 2) in https://www.nomachine.com/AR03P00973.
Hello,
we investigated and found out that AutoHotKeys intentionally ignores input events injected on the system that are not coming from a physical device. NoMachine can’t overcome this, but you could still forward a keyboard to the server as a generic USB device from the NoMachine menu panel.
Further investigations identified two different issues. One is in the Intel drivers, occurring when the monitor on the client machine has a resolution higher than FullHD (1920×1080). That seems to be fixed in the last drivers but it doesn’t look to be your case.
Another issue was found in the NoMachine software, only occurring when the server machine has multiple monitors. This will be fixed in the next release, but if you wish to test a preliminary patch, just send an e-mail to forum[at]nomachine[dot]com to receive it.
Further investigations identified two different issues. One is in the Intel drivers, occurring when the monitor on the client machine has a resolution higher than FullHD (1920×1080). That seems to be fixed in the last drivers. Another issue was found in the NoMachine software, only occurring when the server machine has multiple monitors. This will be fixed in the next release, but if anyone wishes to test a preliminary patch, just send an e-mail to forum[at]nomachine[dot]com.
Could you send a screenshot showing the issue?
If the issue only occurs when using the hardware encoding, that would easily explain why other products don’t have this problem.
So far this issue was only reported against Intel. Do you confirm that NoMachine is set to use the Nvidia card in the Nvidia control panel and not the integrated one (unless the integrated card is disabled in BIOS or you have an AMD CPU)?
Yes, I’m asking for the client side. So you have an AMD CPU (or the integrated card is disabled in BIOS)?
Hi,
yes, it could be useful to know if there was any change on the Windows side (e.g. Were there other updates? Was the monitor unplugged?). Please also gather logs from the client (https://www.nomachine.com/DT10O00163#2.1) and the server machine (https://www.nomachine.com/AR10K00697). You can attach them here or send to forum[at]nomachine[dot]com.
Do you have both a Nvidia card and the Intel integrated graphics? Can you try to use a different card for NoMachine in the Nvidia control panel? And which one is in use now?
Just to be clear: nothing is changed regarding hardware decoding in version 6.10. The only change is that hardware decoding is now enabled by default also for large screen resolutions (in https://www.nomachine.com/FR04R03970 you can read about the reasons).
Unfortunately it looks like that some card/drivers combinations are still affected by this problem. For some users updating Intel graphics drivers fixed the issue, but not for all (make sure you reboot the machine after updating, by the way). As far as we know the problem is in the drivers and there it should be fixed.