DirectX v OpenGL which one to use under windows

Accelerated Computing CUDA CUDA Programming and Performance
1

Let us assume we do not care about the portability of the system, its just under Windows.

Of the two options - DirectX 9 and OpenGL, which one has more advantages?
Here are a few I can think of right away -

OpenGL:

  1. You can work with both vertex buffers as well as pixel buffers.
  2. All advantages OpenGL has over DirectX :D (that covers quite a bit eh?)

DirectX:

  1. Only vertex buffers
  2. Lesser load on the CPU. (If you run the fluidsGL ~50% CPU is utilized, while in case of fluidsD3D, CPU utilization is 0% !!! )
  3. What if newer versions of DirectX work differently than the current. (Like deprecating DirectDraw)

What else?

A side-question:
Does anyone know why OpenGL leads to higher CPU utilization?

2

I don’t think your analisys is exact on this matter: the programming model between OpenGL (both 2.0 or ES) and DX9/10 is quite different.

In GL you should think your process as a main loop instanciated by your app: for this reason the CPU usage is higher than a DX app (as the DX event loop is handled by another software layer, more connected to the underlying OS).

The advantage is you can handle your events with a your own event procedure. The drawback is you have to handle everything by yourself ! (uh… it seems the same than before… sounds strange!!!).

BTW, if you work with CUDA, stay with GL and you’ll preserve the maximum functionalities range and cross OS compatibility.

Bye!

3

Thanks zzambia! Will certainly find out more about what you have said.