d3d12 vs ogl

[monstercameron]

GeeXLab, the successor of GLSL Hacker, comes with the support of Direct3D 12. It’s not a full support but most of the basic things are available: command lists (CL), pipeline state objects (PSO), constant buffers (CB) and HLSL shaders.


An introduction to Direct3D programming with GeeXLab is available HERE.

Since GeeXLab is now available with an OpenGL and a Direct3D 12 renderers, here is a quick benchmark that shows the difference of performance between Direct3D 12 and OpenGL 3.2 in a very simple scene: a phong-textured mesh (a torus to be original) is rendered with various polygon density. This test uses one command list, one PSO, one HLSL program, one mesh and one texture.

source: http://www.geeks3d.com/20160105/dir...campaign=Feed:+TheGeeksOf3d+(The+Geeks+Of+3D)

 

[dogen1]

Simple benchmark. but it's better than nothing. I wonder if OGL 4.0+ would make a difference for this over 3.2.

 

[Gikaseixas]

Good for perspective
DX12 = higher loads for the Radeon, OGL = higher loads for Geforce

 

[Piroko]

This test uses one command list, one PSO, one HLSL program, one mesh and one texture.

That's as synthetic as it gets imho, nothing to see here...

 

[monstercameron]

Simple benchmark. but it's better than nothing. I wonder if OGL 4.0+ would make a difference for this over 3.2.

AFAIK ogl4+ is just a bunch of extensions...

 

[Mercennarius]

Another test showing AMD seeing more of a benefit in DX12 than Nvidia.

 

[Headfoot]

An interesting test but very hard to draw any conclusions, since its such a narrow test the specific way the code was written will likely have massive impacts in the numbers. A data point but certainly not dispositive. I'm sure we'll see everyone bandying it about as proof their team is better, depending which one they're on.