[BUG] CUTLASS_NVCC_ARCHS_SUPPORTED has wrong values leading to compilation failure

[irshadcc]

Which component has the problem?

CMake files

Bug Report

A clear and concise description of what the bug is.
The variable CUTLASS_NVCC_ARCHS_SUPPORTED stores the list of architectures that is supported according to the cuda version.
Code reference : https://github.com/NVIDIA/cutlass/blob/main/CMakeLists.txt#L175

This prevents CUTLASS from building its own extensive examples/tests

set(CUTLASS_ENABLE_EXAMPLES OFF CACHE BOOL "" FORCE)
set(CUTLASS_ENABLE_TESTS OFF CACHE BOOL "" FORCE)

set(CUTLASS_ENABLE_LIBRARY OFF CACHE BOOL "" FORCE)

FetchContent_MakeAvailable(cutlass)

> nvcc fatal   : Unsupported gpu architecture 'compute_70'


I would like to open a pull request to fix this issue.

[irshadcc]

A simple fix is to add one more condition to check the cuda_version < 13.0 to see if cuda 70 can be added or not

# Support for Volta (70, 72) is only available in CUDA < 13.0
if (CUDA_VERSION VERSION_GREATER_EQUAL 11.4 AND CUDA_VERSION VERSION_LESS 13.0)
    list(APPEND CUTLASS_NVCC_ARCHS_SUPPORTED 70 72)
endif()

[Flink-ddd]

Hi @hwu36 , Could you please assign this task to me? I want to resolve it, Thanks.