architecture - When to use Nvidia CUDA instead of plain OpenGL shaders -

architecture - When to use Nvidia CUDA instead of plain OpenGL shaders -

i'm asking myself architectural benefits of these 2 approaches whether using nvidia cuda or opengl shaders computation.

therefore want determine part of application improve implemented on cuda or opengl.

for sure, if platform-dependence no reason, gain more granularity threads , memory using cuda.

to display results, computed cuda, using opengl, have utilize maybe "tricky" interoperabiltiy api.

are there best practices when utilize 1 of these architectures , when utilize combined approach?

most rendering tasks way harder implement using cuda. shaders integrated rendering apis, such opengl, automate , provide mutual , efficient tools rendering geometry. things texture sampling , polygon rasterisation built-in shading languages. if re-implement using cuda, not take long time , effort end running slower shader.

cuda designed, in view, process other complex computational things, physics simulations, in more parallel , efficient manner, using naturally scalable gpu. apart that, rendering of geometry, there little gain utilize of computing framework cuda or opencl.

opengl architecture cuda gpgpu nvidia