[DOC]: Documentation for `getLocalRuntimeVersion()` is probably incorrect

[carterbox]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this bug and that I agree to the Code of Conduct

Is this for new documentation, or an update to existing docs?

Update

Describe the incorrect/future/missing documentation

The documentation for cudaRuntimeGetVersion() and getLocalRuntimeVersion() is exactly the same. It is not explained what the difference between these two functions is.

If this is a correction, please provide a link to the incorrect documentation. If this is a new documentation request, please link to where you have looked.

https://nvidia.github.io/cuda-python/cuda-bindings/latest/module/runtime.html#cuda.bindings.runtime.getLocalRuntimeVersion

[leofang]

I think it is correct, it's just a lot of duplication. The short description for one is

Returns the CUDA Runtime version.

and the other is

Returns the CUDA Runtime version of local shared library.

[carterbox]

I propose more explanatory detail which I have gleaned from discussions here on GitHub. Please correct me if I'm wrong. e.g.

Returns the CUDA Runtime version statically linked into `cuda.bindings`.

[Insert a sentence about why this is useful.]

TBH, it's unclear to me why you would query version that cuda.bindings was compiled with in this way. Wouldn't you just inspect cuda.bindings.__version__?

Returns the CUDA Runtime version of a local shared library.

Returns the version of whatever cudart shared library is located by `cuda.pathfinder`.

[kkraus14]

TBH, it's unclear to me why you would query version that cuda.bindings was compiled with in this way. Wouldn't you just inspect cuda.bindings.__version__?

The patch versions of cuda.bindings and the underlying cuda runtime library are not aligned. So if there's a bug in the cuda runtime library that is fixed in a patch version you can check I guess.