The original OpenGL Core changes only used a single vertex array object
for specifying the data layout of our vertex data. This uses the OpenGL
extension GL_ARB_vertex_attrib_binding to extend that so that each
vertex_layout uses a separate vertex array object which should improve
performance by reducing the CPU overhead while switching the vertex
layout. Since we need to use offsets into the vertex buffers at some
points this can not be done with the core functions since those do not
support specifying an offset when binding a vertex buffer.
GL_ARB_vertex_attrib_binding does support that and also features an
improved API for specifying the vertex layout. If that extension is not
available then the rendering engine will fall back to the current code.
In CPU intensive scenes I observed a frametime improvement of 1-2 ms.
All transformations are now handled by shaders so there is nothing
special going on in the graphics API specific code anymore. Since model
uniform buffers require access to these matrices I moved the matrix
handling out of the graphics API code.
This converts the deferred lighting shader to use uniform buffers for
the dynamic data sent to the GPU while rendering the lighting effects.
This doesn't affect performance but the deferred lighting code is a
nicely isolated piece of rendering code where I could test how the
rendering code could use the uniform buffers.
This is related to #910.
This converts all usages of bmpman textures to OpenGL texture arrays.
That means that all frames of an animation are stored in a single OpenGL
texture object and can be accessed at the same time in a single draw
call. This should significantly reduce the overhead of drawing lots of
particles since all particles of a single animation can be drawn with a
single draw call.
This fixes#1169.
The root issue of #976 was that the code that reset the depth buffer of
the scene framebuffer was actually setting the depth buffer of the bloom
framebuffer since that didn't get reset properly. To fix this I
introduced a system in our GL_state which keeps track of a stack of
"bound" framebuffers which allows us to return to the previous
framebuffer with a simple function call.
I also added support for GL_KHR_debug since that help me while debugging
which texture was in use. In the future I'll probably extend that to
other areas of the OpenGL code as needed.
This fixes#976.
The new design is centered around the Event Trace JSON format but it
should still be possible to adapt it for other profiling tools. The old
profiling system has been removed.
primative_batch_buffer:desired_buffer_size was not properly initialized in vc120 compilers during debugging. Default constructors proved insufficient, and std:map::operator[] seems to ignore them anyway. This fix gets around that by explicitly clearing desired_buffer)size from within batching_init_buffer()
Replaces calls to double-precision floating point trig functions with
calls to single-precision floating point trig functions as follows:
cos() -> cosf()
sin() -> sinf()
tan() -> tanf()
acos() -> acosf()
asin() -> asinf()
atan() -> atanf()
This allows the compiler to avoid generating unnecessary instructions
that convert a float to a double and then a double back to a float.
Signed-off-by: Peter Mitsis <[email protected]>
Don't set PF_PROJECTED in g3_transfer_vertex() as it's not projecting
the vertex
Note: displaying nebula emp/lightning used to depend on unitialised data!
git-svn-id: svn://svn.icculus.org/fs2open/trunk/fs2_open@10828 387891d4-d844-0410-90c0-e4c51a9137d3
asarium