| Age | Commit message (Collapse) | Author |
|---|
|
treated as error)
|
|
reviewed by davep
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
actual installed and available VRAM.
|
|
|
|
vertex buffer objects when available vram is under 256MB.
|
|
|
|
treated as error)
|
|
reviewed by davep
|
|
|
|
need to apply these manually.
|
|
as part of STORM-1011.
reviewed by davep.
|
|
|
|
queries if GL is set to not use VBOs for stream draw.
|
|
|
|
leaking.
|
|
* Now called 'HeadlessClient' instead of 'DisableRendering'
* Removed most cases where we skipped certain behaviors in the client when in this mode. This gets us closer to a 'true' client, for testing purposes.
|
|
data from gDEBugger.
Reviewed by Nyx.
|
|
leaking.
|
|
usage changes from SNOW-643 and properly occlusion cull void water patches for machines without depth clamp.
|
|
Lock Up
originally done in http://bitbucket.org/leyla/leyla-mesh-development
|
|
|
|
|
|
|
|
available and don't read back result if it will block).
|
|
A copy constructor was implemented which did a memcpy,
which included the vtable pointer which was to another object of another
class (same child though). This resulted in the wrong destructor
being called.
The reason for the memcpy was for alignment purposes.
The solution was to move to LLVector4a, which is intrinsicly aligned.
Also, did some performance optimizations based on the LLVector4a optimizations.
The solution was to re-implement the
|
|
|
|
parameters.
|
|
|
|
|
|
with ARB.
|
|
Get rid of RenderUseFBO debug setting (only use FBO when rendering deferred). Use GL_ARB_framebuffer_objects instead of GL_EXT_...
|
|
|
|
|
|
assert when FBOs are destroyed out of order.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
