Melchior FRANZ wrote:

* Curtis L. Olson -- Wednesday 13 October 2004 23:47:


I haven't looked closely at the display list stuff. Are we being careful to free the display lists when we free the associated objects and remove them from the scene graph? Is this something that plib should automatically handle (but maybe isn't because we are the first to stress test it?)



No, it doesn't look like plib accumulated DLists. I flew the Hunter again from KLAX to KSFO along the coast, then quite a while heading W, into the pacific.

 KLAX, right after taking off:      22k  (i.e.: 22.000 DLists)
 coast:                       avg. ~18k  (13K--22k, depending on the
                                          terrain; peaks at 30k)
 KSFO:                              27k
 heading W, slowly falling to:       9k
 exiting from fgfs:              2960


The final 2960 DLists were not freed via glDeleteLists. But in any case: the DLists weren't the cause for excessive memory consumption. It could have been problems with KDE's artsdsp.

m.

_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d



Those numbers (22k...9k) seam enormous to me. You say 13 to 22k depending on terrain...
If I understand well we have one dlist per polygone ?
Each terrain chunk (.btg) is loaded and stored in his ssg tree. A range and transform matrix is added at the top of this tree.
Isn't it possible to have one Dlist for this whole tree ?


Regards.



_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d

Reply via email to