-- Download the rfbsessions tarball linked to below. Untar it into your
home directory (which should create a directory called
~/rfbsessions.turbo.hiqual).
-- Download the attached test script into your home directory.
-- Shut down all running applications, and make sure your screen
resolution is at least 1280x1024.
-- With the AppleJava viewer installed, open a terminal (if you haven't
already) and execute:
sh ~/vncviewerbench >~/vncviewerbench-AppleJava.log
Uninstall the AppleJava viewer and install the OracleJava version,
then do:
sh ~/vncviewerbench >~/vncviewerbench-OracleJava.log
-- Send me both log files, in addition to the following specs (you can
get this from the System Information application):
* Model name (e.g. "Mac Mini")
* Model ID (e.g. "Macmini7.1")
* Processor Name and Speed (e.g. "Intel Core i7 3 GHz")
* Graphics chipset (e.g. "Intel Iris")
* OS X version and patch level (e.g. "10.10.5")
I strongly suspect that your machine is similar to my newer Mini and
that it doesn't have Quartz acceleration. One of the things I'm trying
to ascertain, though, is whether Java 2D Quartz acceleration is a
function of the graphics chip or the O/S. From my limited dataset, I
know that my Iris-equipped Mini running Yosemite doesn't have Quartz
acceleration, whereas my Intel HD-equipped MacBook running Mavericks
does, and my nVidia-equipped Mini running Mountain Lion does. However,
it's entirely possible that no machine running Yosemite or later will
have Quartz acceleration. Running Mountain Lion and El Cap in Parallels
Desktop on my new Mini seems to confirm this, but that's a somewhat
non-standard graphics environment, so I'm not sure how valid those tests
are.
On 8/24/16 10:56 AM, Brady Koenig wrote:
> Thanks. I'll try.
>
> This is what I'm running: MacBook Pro (Retina, 15-inch, Late 2013)
>
> If you send me the scripts and instructions I'll give it a try.
>
> Thanks again
> Brady
>
> On Wed, Aug 24, 2016 at 9:52 AM, DRC <[email protected]
> <mailto:[email protected]>> wrote:
>
> Have you tried the OracleJava version? Depending on your graphics
> hardware, that might be the fastest version anyhow. I don't claim that
> these results are universal, but here's what the performance situation
> looks like on my test machines:
>
> (Numbers represent average blit time across the "canonical" TurboVNC
> session captures that we use for all of our viewer benchmarking:
> http://www.turbovnc.org/rfbsessions.turbo.hiqual.tar.bz2
> <http://www.turbovnc.org/rfbsessions.turbo.hiqual.tar.bz2>)
>
> Late 2014 Mac Mini w/ 3 GHz i7 and Intel IRIS:
> AppleJava (1.6.0_65): 2D sessions= 232s, 3D sessions= 46s
> OracleJava (1.8.0_45): 2D sessions= 18s, 3D sessions= 9.4s
>
> 2009 Mac Mini w/ 2 GHz Core 2 Duo and nVidia GeForce 9400:
> AppleJava (1.6.0_65): 2D sessions= 8.2s, 3D sessions= 9.7s
> OracleJava (1.8.0_45): 2D sessions= 46s, 3D sessions= 9.3s
>
> 2011 Macbook Pro w/ 2.4 GHz i5 and Intel HD Graphics 3000:
> AppleJava (1.6.0_65): 2D sessions= 5.3s, 3D sessions= 5.1s
> OracleJava (1.8.0_45): 2D sessions= 70s, 3D sessions= 16s
>
> Welcome to the Hell that is Java 2D. On older equipment, Apple Java
> (AKA "Java for OS X") tends to be faster, because it has a
> Quartz-accelerated implementation of Java 2D. However, that
> Quartz-accelerated implementation of Java 2D apparently doesn't support
> newer graphics chips, like the Intel IRIS. On my newer Mac Mini w/
> IRIS, trying to use AppleJava is not only very slow (no Quartz
> acceleration), but it produces visual anomalies like the one you're
> describing (and the gamma doesn't appear to be correct, either.) On
> older machines, Oracle Java (which only uses OpenGL for Java 2D on Mac
> platforms-- no Quartz) is somewhat slower than Apple Java but is still
> usable. It's worth noting that the 2D performance metrics aren't
> necessarily realistic indicators of how the end user will perceive
> performance. Those benchmarks are just throwing tiles at the blitter as
> fast as they can. The actual update rate is still quite fast with
> Oracle Java on older systems, so whether or not an end user could tell
> the difference between Apple and Oracle Java on those systems is an open
> question.
>
> Long and the short of it-- try Oracle Java and see if that works around
> the issue. Apple Java is a deprecated product, and support for it on
> newer OS X releases is limited at best.
>
>
> On 8/24/16 10:24 AM, Brady Koenig wrote:
> >
> > I'm using turboVNC server and client.
> > server: turbovnc-2.0.1.x86_64.rpm
> > client: TurboVNC-2.0.2-AppleJava
> >
> > I recently updated to El Capitan. Everything worked great before the
> > upgrade.
> > I run the turboVNC client in full screen mode on a second screen.
> > When I switch focus back and forth from a window on the primary screen
> > to turboVNC on the secondary screen I get a brief full screen white
> > flash. That's driving me nuts.
> >
> > Has anyone else seen this? Anyone found a work around?
> >
> > Thanks
> > Brady
>
>
> ------------------------------------------------------------------------------
> _______________________________________________
> TurboVNC-Users mailing list
> [email protected]
> <mailto:[email protected]>
> https://lists.sourceforge.net/lists/listinfo/turbovnc-users
> <https://lists.sourceforge.net/lists/listinfo/turbovnc-users>
>
>
>
>
> ------------------------------------------------------------------------------
>
>
>
> _______________________________________________
> TurboVNC-Users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/turbovnc-users
>
set -e
RFBSESSIONS=" \
bugzilla-16.rfb \
compilation-16.rfb \
bars-16.rfb \
kde-hearts-16.rfb \
freshmeat-8.rfb \
slashdot-24.rfb \
photos-24.rfb \
kde-hearts-24.rfb \
3dsmax-04-24.rfb \
catia-02-24.rfb \
ensight-03-24.rfb \
light-08-24.rfb \
maya-02-24.rfb \
proe-04-24.rfb \
sw-01-24.rfb \
tcvis-01-24.rfb \
ugnx-01-24.rfb \
glxspheres-24.rfb \
googleearth-rgb-24.rfb \
q3demo-rgb-24.rfb"
for i in $RFBSESSIONS; do
echo >&2 Testing $i ...
/opt/TurboVNC/bin/vncviewer -bench ~/rfbsessions.turbo.hiqual/$i
-benchiter 5 -benchwarmup 5
done
------------------------------------------------------------------------------
_______________________________________________
TurboVNC-Users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/turbovnc-users
