Update nvbench configuration to use C++20 standard

[cliffburdick]

No description provided.

[cliffburdick]

/build

[greptile-apps]

Greptile Summary

This PR aligns nvbench's compilation standard with the rest of the MatX project by forwarding CMAKE_CXX_STANDARD 20 and CMAKE_CUDA_STANDARD 20 as CPM options when fetching nvbench. The main library already mandates C++20 via target_compile_features(matx INTERFACE cxx_std_20 cuda_std_20) (line 139), so this change ensures the benchmark dependency is compiled consistently under the same standard.

• Sets CMAKE_CXX_STANDARD 20 so nvbench's host C++ translation units use C++20.
• Sets CMAKE_CUDA_STANDARD 20 so nvbench's device/CUDA translation units also use C++20, keeping parity with the main project's cuda_std_20 feature requirement.
• No other files are affected; the bench subdirectory (bench/CMakeLists.txt) links against nvbench::main and matx::matx and is unmodified.

Confidence Score: 5/5

• This PR is safe to merge — it is a one-line functional change that aligns nvbench's compilation standard with the rest of the project.
• The change is minimal (4 lines), correct in syntax and semantics, and directly addresses the pre-existing standard mismatch. Both CMAKE_CXX_STANDARD 20 and CMAKE_CUDA_STANDARD 20 are now forwarded, matching the cxx_std_20/cuda_std_20 pair required by matx::matx. No logic, API surface, or other configuration is affected.
• No files require special attention.

Important Files Changed

Filename	Overview
CMakeLists.txt	Adds CPM_ARGS OPTIONS "CMAKE_CXX_STANDARD 20" "CMAKE_CUDA_STANDARD 20" to rapids_cpm_nvbench() so that the fetched nvbench dependency is compiled under the same C++20/CUDA 20 standard already required by the main matx library.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[cmake configure: MATX_BUILD_BENCHMARKS=ON] --> B[rapids_cpm_nvbench with CPM_ARGS OPTIONS]
    B --> C{nvbench already found?}
    C -- No --> D[CPM fetches nvbench source]
    D --> E[Configure nvbench with CMAKE_CXX_STANDARD=20 and CMAKE_CUDA_STANDARD=20]
    E --> F[nvbench built with C++20 / CUDA 20]
    C -- Yes --> G[Use pre-installed nvbench]
    F --> H[add_subdirectory bench]
    G --> H
    H --> I[matx_bench links nvbench::main + matx::matx both requiring cxx_std_20 and cuda_std_20]

Loading

_{Last reviewed commit: "Update CMakeLists.tx..."}

[cliffburdick]

/build

[cliffburdick]

/build