I was able to verify the CPU usage on my retina MBP and further show
that ES2 on Windows also consumes a similar amount of CPU:
Mac (using ES2): 20-25%
Windows (using D3D): 3%
Windows (using ES2): 20%
so this is definitely related to our use of OpenGL and not a Mac
platform issue (though it did have a couple percent more overhead than
the ES2 pipe on Windows, the bulk of the performance was ES2 vs OGL).
Can you remove the external dependencies on the date formatter from your
test case and then submit a bug so we can track this issue?
...jim
On 7/8/15 7:57 AM, Tobias Bley wrote:
Hi Jim,
please checkout the small app on Github:
https://github.com/tobium/JavaFXPerformanceTest
Mac OS X: 10.10.4
Mac Book Pro Retina (Late 2012), Intel HD Graphics 4000 1024 MB, 8 GB RAM, i7
2,6Ghz
Java FX 1.8.0_60
FPS: Mac and Windows: 60 FPS
CPU usage: Mac: 25-80%, Windows: 0-3%
Best regards,
Tobi
Am 07.07.2015 um 21:23 schrieb Jim Graham <james.gra...@oracle.com>:
Hi Tobi,
Can you share your small clock app? Perhaps file a bug and attach the source?
Also, what version of MacOS are you running on what hardware? (And compared to
what version of Windows on what hardware?)
...jim
On 7/7/15 4:32 AM, Tobias Bley wrote:
Hi,
currently our experiences with JavaFX on Mac are very disappointing. While
JavaFX on Windows runs very good with low cpu usage, JavaFX on Mac via Java 8
doesn’t perform well. I created a little clock app which uses between 25% and
80% cpu usage. But what’s the reason for? I though JavaFX rendering pipeline
works on the graphic device? So why there is such a traffic on the CPU?
Best regards,
Tobi