[libomptarget][nfc][nvptx] Less verbose debug build

[JonChesterfield]

Summary

[libomptarget][nfc][nvptx] Less verbose debug build

When compiling with LIBOMPTARGET_NVPTX_DEBUG, verbose ptxas
provides approximately 300 instances of diagnostics of the form:
ptxas info : Function properties for some_function
x bytes stack frame, y bytes spill stores, z bytes spill loads

While interesting if tuning the performance of the runtime, it doesn't seem
worth having this in the build output by default.

[grokos]

I guess it's a reasonable thought to disable such verbose output by default, but I can also see why these diagnostics could be useful l in some cases. Can we introduce a CMake variable to toggle --ptxas-options=-v on and off? (off by default)

[jdoerfert]

This comment was removed by jdoerfert.

[jdoerfert]

I did not see @grokos comment, I let him do the review.

openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt

if(LIBOMPTARGET_DEP_CUDA_FOUND)
   foreach(sm ${nvptx_sm_list})
     set(CUDA_ARCH ${CUDA_ARCH} -gencode arch=compute_${sm},code=sm_${sm})
   endforeach()
 
   # Activate RTL message dumps if requested by the user.
   set(LIBOMPTARGET_NVPTX_DEBUG FALSE CACHE BOOL
     "Activate NVPTX device RTL debug messages.")
   if(${LIBOMPTARGET_NVPTX_DEBUG})
-    set(CUDA_DEBUG -DOMPTARGET_NVPTX_DEBUG=-1 -g --ptxas-options=-v)
+    set(CUDA_DEBUG -DOMPTARGET_NVPTX_DEBUG=-1 -g )
   endif()
 
   # NVPTX runtime library has to be statically linked. Dynamic linking is not
   # yet supported by the CUDA toolchain on the device.
   set(BUILD_SHARED_LIBS OFF)
   set(CUDA_SEPARABLE_COMPILATION ON)
   list(APPEND CUDA_NVCC_FLAGS -I${devicertl_base_directory}
                               -I${devicertl_nvptx_directory}/src)