From: Kenneth B. R. <kbr...@al...> - 2001-02-08 08:52:27
|
> experiences with: > > XFree86 4.0.2, GL4Java 2.5.0, KDE2.1b2 + konqueror 2.1b2 and > - jdk1.2-sun: no display .. > - jdk1.3-sun: no display .. > > - jdk1.3-IBM: runs well > ================== I've been running GL4Java on top of Sun's JDK 1.3 on Windows 2000 successfully with an NVidia GeForce. I had some problems porting some code to GL4Java. Specifically, in order to get animation working at all and with good performance, it was necessary to subclass GLAnimCanvas and use the following settings: setUseRepaint(true); setUseFpsSleep(false); setSuspended(false); I found that calling sDisplay() in my main thread reliably crashed the machine (apparently doing something relating to multithreaded access to the OpenGL context the GeForce didn't like). I should note that I played a little bit with the non-free Magician binding and believe that there are some unresolved issues with GL4Java's interaction with the window system. Specifically, a friend and I did an experimental project in which we shoehorned some of GL4Java's sources into the Magician APIs. Unfortunately, it wasn't stable (crashed frequently on his machine), likely due to my inexperience with the multithreading issues in interacting properly with the AWT. However, performance of this system appeared to be significantly better than my first attempts with GL4Java, probably because we were able to avoid the repaint() mechanism and associated thread synchronization. I'm planning on doing some more work on this over the next few weekends -- if anybody has experiences to share or would like to collaborate please send mail. -Ken _____________________________________________________________________________ Kenneth B. Russell (kbr...@al...) http://www.media.mit.edu/~kbrussel |
From: Sven G. <sgo...@ja...> - 2001-02-08 09:43:00
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday 08 February 2001 09:52, you wrote: > > experiences with: > > > > XFree86 4.0.2, GL4Java 2.5.0, KDE2.1b2 + konqueror 2.1b2 and > > - jdk1.2-sun: no display .. > > - jdk1.3-sun: no display .. > > > > - jdk1.3-IBM: runs well > > ================== > > I've been running GL4Java on top of Sun's JDK 1.3 on Windows 2000 > successfully with an NVidia GeForce. > well, me either, but thats OT :-) > I had some problems porting some code to GL4Java. Specifically, > in order to get animation working at all and with good > performance, it was necessary to subclass GLAnimCanvas and use > the following settings: > setUseRepaint(true); > setUseFpsSleep(false); > setSuspended(false); > > I found that calling sDisplay() in my main thread reliably > crashed the machine (apparently doing something relating to > multithreaded access to the OpenGL context the GeForce didn't > like). Well for all the demos where I tried those flags, a crash never happend. So please send a little nice example (tar.bz2 or zip). Be sure to free and makeCurrent the GLContext, especially in a multithreading application ! I am thinking about a more precise native animation sheduler for a next version. So the fps can be more accurate. "Artiste on the Web" told me success with the yield() method, to allow other threads being run. Anyway, todays JVM's implements the native threading very well. The rest of the posting is something I do not understand, because it uses word's outta my universe ;-)) Well, I appreciate to make GL4Java more stable and better. I does not love a magician shoe - no i do not, but the world is a free one, at least the GNU world. I would prefere an own free solution for a common API. bye, sven > > I should note that I played a little bit with the non-free > Magician binding and believe that there are some unresolved > issues with GL4Java's interaction with the window system. > Specifically, a friend and I did an experimental project in which > we shoehorned some of GL4Java's sources into the Magician > APIs. Unfortunately, it wasn't stable (crashed frequently on his > machine), likely due to my inexperience with the multithreading > issues in interacting properly with the AWT. However, performance > of this system appeared to be significantly better than my first > attempts with GL4Java, probably because we were able to avoid the > repaint() mechanism and associated thread synchronization. > > I'm planning on doing some more work on this over the next few > weekends -- if anybody has experiences to share or would like to > collaborate please send mail. > > -Ken > > > ___________________________________________________________________________ >__ Kenneth B. Russell (kbr...@al...) > http://www.media.mit.edu/~kbrussel > > _______________________________________________ > gl4java-usergroup mailing list > gl4...@li... > http://lists.sourceforge.net/lists/listinfo/gl4java-usergroup - -- mailto:sgo...@ja... www : http://www.jausoft.com ; pgp: http://www.jausoft.com/gpg/ voice : +49-521-2399440, +49-170-2115963; fax: +49-521-2399442 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE6gmoOHdOA30NoFAARAv26AKCl7bW89ZcRwcsyNvkq1XyjKXFOawCgoQhl ruji1jrgN9jncbE2HQi+RLE= =Lz3U -----END PGP SIGNATURE----- |
From: Sven G. <sgo...@ja...> - 2001-02-11 03:46:47
|
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - - -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday 08 February 2001 10:42, you wrote: > > I had some problems porting some code to GL4Java. Specifically, > > in order to get animation working at all and with good > > performance, it was necessary to subclass GLAnimCanvas and use > > the following settings: > > setUseRepaint(true); > > setUseFpsSleep(false); > > setSuspended(false); > > > > I found that calling sDisplay() in my main thread reliably > > crashed the machine (apparently doing something relating to > > multithreaded access to the OpenGL context the GeForce didn't > > like). > I have checked this bug out while playing around with the glDemosCvs ! I just pressed the repaint:=false & fpsSleep=false buttons ! Also I have changed the GLLandscape Demo: - added those buttons - added two more threads, which just makecurrent/Free and reads some GL strings without any pause ! Well, the specification of glXMakeCurrent says, that if another threads uses the GLXContext, glXMakeCurrent should fail, so gljMakeCurrent will fail and the display method finishes ! I have played arround with an own manual java-thread locking (monitor) mechanism within gljMakeCurrent/gljFree ! So this looks like a missbehavior of NVidia's glXMakeCurrent (X11 driver for Linux) implementation -> not thread safe !? With the new java monitor for within gljMakeCurrent/gljFree the responsiveness and reliabel rised: - no more mouse-click -> openGL loss (thread is now waiting) - with a kind of tri-state, better java-thread switching .. I have also added in gl4java.applet.SimpleAnimApplet1 a feature for mouse button 3, where (if pressed), the GLCanvas is moved to a new Frame. Pressed again, it is moved back ;-) It works, I have just added a glj = null, in the cvsDispose method .. well looks more stable now .. (tested with above described demos) but it is an NVidia - X11 driver implementation problem ! Not happend within NVidia - Win32 and the DRI XFree86 drivers ! later, sven - - - -- mailto:sgo...@ja... www : http://www.jausoft.com ; pgp: http://www.jausoft.com/gpg/ voice : +49-521-2399440, +49-170-2115963; fax: +49-521-2399442 - - -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE6hLdZHdOA30NoFAARAowsAJ4r1Kgg6qTax/XVwergv6nyec8MuQCfQPSu Sn1qBrCE0OBKPTFA4oDU2TY= =67RA - - -----END PGP SIGNATURE----- - -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE6hS9NHdOA30NoFAARAnXdAJ9th+udycrJ5xWhQIGdXtMWf6oH7ACfeX65 p4a0tnBvYw4uXpSGjW4HeQw= =VgWM - -----END PGP SIGNATURE----- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE6hgsdHdOA30NoFAARAoEXAJ0RE1cx6W577JxrUVEwNX6v7oEHhQCgv8aj iRtyzL32zHughNFLXc7liG8= =MFmT -----END PGP SIGNATURE----- |