nVIDIA CUDA R3D

[1div0]

Description

This is intended to add support of R3D CUDA based decoder, which is much faster than R3D CPU only decoder.

RED One clips are handled by the CPU.

Tests

Checklist:

• I have read the contribution guidelines.
• I have updated the documentation, if applicable.
• I have ensured that the change is tested somewhere in the testsuite
  (adding new test cases if necessary).
• If I added or modified a C++ API call, I have also amended the
  corresponding Python bindings (and if altering ImageBufAlgo functions, also
  exposed the new functionality as oiiotool options).
• My code follows the prevailing code style of this project. If I haven't
  already run clang-format before submitting, I definitely will look at the CI
  test that runs clang-format and fix anything that it highlights as being
  nonconforming.

[1div0]

//Flags used by the linker during the creation of shared libraries
// during all build types.
CMAKE_SHARED_LINKER_FLAGS:STRING=-L/usr/local/cuda -lcudart

[1div0]

There is still needed to add:

CMAKE_SHARED_LINKER_FLAGS -lcudart

[lgritz]

I'm concerned about setting these unconditionally.
Does every system that would have R3DSDK always have the CUDA SDK available?

We already have a CMake variable called OIIO_USE_CUDA to enable CUDA at build time, and when that is enabled and the CUDA SDK is found, it will define the C++ preprocessor symbol OIIO_USE_CUDA, so you should use that rather than making new symbols.

[1div0]

Yes, indeed. These definitions do belong to the build configuration settings. Will modify ASAP.

Within the R3DSDK there are four options: CPU only decoding, nVIDIA CUDA, OpenCL and 🍏Metal.