When I switched from Slackware 14.0 to 14.1, one of the things I noticed was a couple of performance problems. Today I decided to look at the vnc server performance problem. I use vncserver to start a desktop session that I can remote into, been doing this for quite a while and it works well. With Slackware 14.1, I experienced a noticeable drop in performance when accessing the remote desktop. I also noticed that they switched from tightvnc to tigervnc when going from 14.0 to 14.1. After playing with this for a while I've discovered that that tigervnc version of vncserver is slower than that tightvnc version, and switching back to the tightvnc version fixed the performance problem I was observing.

Now before y'all get your stop watches out, let me say that the difference isn't huge, it's just noticeable. I spend a lot of time on remote desktops, and little things like this bug me. Those that don't do this much may never notice any difference.

Has anyone else noticed any difference between tigervnc and tightvnc vncserver performance?

At this time I'm using Slackware 14.1 32 bit. AMD FX 8 core 4GHz cpu, 8GB ram (yes, 32 bit Slackware really does see all 8GB ram), nVidia 660.

I use tigervnc across several machines (all slack 64bit 14/14.1), both LAN and ssh tunneled WAN, with zero network/cpu impact.

tightvnc has been depreciated for some time now: http://en.wikipedia.org/wiki/TigerVNC
and you probably need to define 'performance', launch parameters and running environment to get any real help.

Cheers,

Yeah, "performance" is really highly subjective. There are a few more factors I've not looked at, such as bandwidth usage, etc. I also updated my mother board at the same time, and I had to use Slackware 14.1 because 14.0 won't run on this board - but that is another story entirely. But my cpu went from 6 core 3600MHz phenom to 8 core 4000MHz FX, you would think it would be faster but maybe the low end motherboard is choking it? I should probably look into that also. But regardless I have found in this specific application that tightvnc is faster than tigervnc.

I have also noticed a very significant performance reduction when using ssh tunnel over internet - this is the one that has really bit me, it want from pretty good to almost unusable, and I'm not sure why. This I can't blame on tigervnc because tightvnc does it also. Did ssh client/daemon change? Did my Internet latency/bandwith drop at the same time I upgraded to 14.1? Does my motherboard suck? My old board went bad so I don't have a test bed to compare against. Too many questions, not enough answers.

On the remote, I first ssh in and start a desktop session like this:

vncserver :0 -geometry 1800x950 -depth 16

My client has not changed. From this point there are a several ways I access this box:

1) I start an ssh tunnel: ssh -L5901:localhost:5900 -N -l <user> <remote ip> -p 13567
Then I connect with vncviewer to localhost port 5901. This works but is horribly slow.
I can connect with tightvnc-jviewer.jar to localhost port 5901. Likewise horribly slow.
2) I connect directly to machine instead of using ssh tunnel
tightvnc viewer - nice and fast.
tigetvnc viewer - still fast, but if you use both you can tell the difference
tightvnc-jviewer.jar - still fast.
3) I use tightvnc-jviewer.jar to create the ssh tunnel, and then connect via tunnel. Works but slow.

The thing that is killing me is that on my old motherboard with Slackware 14.0, connecting via ssh tunnel was very fast. The *only* thing that has changed is I now have a newer motherboard, faster cpu, and I'm using Slackware 14.1. I'm doing everything the same way as before, and my client has not changed.

So now that I've confused the issue with all of the above verbiage, these are the issues:

1) tightvnc vncviewer is (subjectively) faster than tigervnc vncviewer, but they are both fast. This was my original topic. If you don't compare them side by side you would probably not notice the difference.

2) vnc via ssh tunnel performance went from good to almost unusable when I 1) replaced mother board 2) upgraded from Slackware 14.0 to 14.1. Idk if it is the newer board or not because everything else performs well. I don't have an answer for this one.

Interesting problem. If I remember correctly, tightvnc leaves some startup file garbage (etc/rc.d/rc.vncserver*) even after uninstalling, which may or may not impact tigervnc, Alien Bob's tigervnc 1.3.0 package does not put anything in /etc/rc.d.

Having both packages installed at the same time could also be an issue.

I start the server from rc.local as su -c "vncserver :1 -name <computer> -extension RANDR" <user> &, use
ciphers=arcfour and -fNT ssh options for the tunnel, vncviewer localhost:1 for the connect.

tigervnc by default shows connection performance adjustements in real time if you launch vncviewer in a terminal.

Network latency and available bandwidth will determine vnc response time so also try a LAN based compare with relatively quiet server/client while monitoring CPU and network on both (such as gkrellm).