DGXOS-7.3.1 cuda_13.0.r13.0/compiler.36424714_0 gcc-13/clang-14? "__SVFloat32_t" is undefined

Accelerated Computing DGX Spark / GB10 User Forum DGX Spark / GB10
1

Is there a known issue with glibc-2.39 on DGX-spark under DGXOS 7.3.1 for compiling anything?

  /usr/include/aarch64-linux-gnu/bits/math-vector.h(106): error: identifier
  "__SVFloat32_t" is undefined

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -S /usr/include/aarch64-linux-gnu/bits/math-vector.h
libc6-dev:arm64: /usr/include/aarch64-linux-gnu/bits/math-vector.h
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -l libc6-dev:arm64
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name            Version         Architecture Description
+++-===============-===============-============-=====================================================
ii  libc6-dev:arm64 2.39-0ubuntu8.6 arm64        GNU C Library: Development Libraries and Header Files

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ cat /etc/dgx-release 
DGX_NAME="DGX Spark"
DGX_PRETTY_NAME="NVIDIA DGX Spark"
DGX_SWBUILD_DATE="2025-09-10-13-50-03"
DGX_SWBUILD_VERSION="7.2.3"
DGX_COMMIT_ID="833b4a7"
DGX_PLATFORM="DGX Server for KVM"
DGX_SERIAL_NUMBER="1984025003364"

DGX_OTA_VERSION="7.3.1"
DGX_OTA_DATE="Thu Nov 20 10:34:12 AM CET 2025"

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2025 NVIDIA Corporation
Built on Wed_Aug_20_01:57:39_PM_PDT_2025
Cuda compilation tools, release 13.0, V13.0.88
Build cuda_13.0.r13.0/compiler.36424714_0

example with cuda-samples:

git clone https://github.com/NVIDIA/cuda-samples.git
cd cuda-samples/
cmake -B build

is failing whatever I am trying..

with gcc:

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ command -v gcc
/usr/bin/gcc
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ realpath /usr/bin/gcc
/usr/bin/aarch64-linux-gnu-gcc-13
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -S /usr/bin/aarch64-linux-gnu-gcc-13
gcc-13-aarch64-linux-gnu: /usr/bin/aarch64-linux-gnu-gcc-13
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -l gcc-13-aarch64-linux-gnu
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                     Version               Architecture Description
+++-========================-=====================-============-=====================================================
ii  gcc-13-aarch64-linux-gnu 13.3.0-6ubuntu2~24.04 arm64        GNU C compiler for the aarch64-linux-gnu architecture


tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ cmake -B build
CMake Error at /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:780 (message):
  Compiling the CUDA compiler identification source file
  "CMakeCUDACompilerId.cu" failed.

  Compiler: /usr/bin/nvcc

  Build flags:

  Id flags: --keep;--keep-dir;tmp -v



  The output was:

  1

  #$ _NVVM_BRANCH_=nvvm

  #$ _SPACE_=

  #$ _CUDART_=cudart

  #$ _HERE_=/usr/lib/nvidia-cuda-toolkit/bin

  #$ _THERE_=/usr/lib/nvidia-cuda-toolkit/bin

  #$ _TARGET_SIZE_=

  #$ _TARGET_DIR_=

  #$ _TARGET_SIZE_=64

  #$ NVVMIR_LIBRARY_DIR=/usr/lib/nvidia-cuda-toolkit/libdevice

  #$
  PATH=/usr/lib/nvidia-cuda-toolkit/bin:/home/tru/bin:/usr/local/cuda/bin:/opt/bin/:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin


  #$ LIBRARIES= -L/usr/lib/aarch64-linux-gnu/stubs
  -L/usr/lib/aarch64-linux-gnu

  #$ rm tmp/a_dlink.reg.c

  #$ gcc -D__CUDA_ARCH_LIST__=520 -E -x c++ -D__CUDACC__ -D__NVCC__
  -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=0
  -D__CUDACC_VER_BUILD__=140 -D__CUDA_API_VER_MAJOR__=12
  -D__CUDA_API_VER_MINOR__=0 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include
  "cuda_runtime.h" "CMakeCUDACompilerId.cu" -o
  "tmp/CMakeCUDACompilerId.cpp4.ii"

  ERROR: No supported gcc/g++ host compiler found, but clang-14 is available.

         Use 'nvcc -ccbin clang-14' to use that instead.

  # --error 0x1 --





Call Stack (most recent call first):
  /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:8 (CMAKE_DETERMINE_COMPILER_ID_BUILD)
  /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:53 (__determine_compiler_id_test)
  /usr/share/cmake-3.28/Modules/CMakeDetermineCUDACompiler.cmake:135 (CMAKE_DETERMINE_COMPILER_ID)
  CMakeLists.txt:3 (project)


-- Configuring incomplete, errors occurred!

with clang-14:

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ command -v clang-14
/usr/bin/clang-14
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ realpath /usr/bin/clang-14
/usr/lib/llvm-14/bin/clang
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -S /usr/lib/llvm-14/bin/clang
clang-14: /usr/lib/llvm-14/bin/clang
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ dpkg -l clang-14
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name           Version           Architecture Description
+++-==============-=================-============-=================================
ii  clang-14       1:14.0.6-19build4 arm64        C, C++ and Objective-C compiler

tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ export NVCC_CCBIN='clang-14'
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ export NVCC_PREPEND_FLAGS='-ccbin clang-14' 
tru@spark-50df:~/git/github/NVIDIA/cuda-samples$ cmake -B build
CMake Error at /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:780 (message):
  Compiling the CUDA compiler identification source file
  "CMakeCUDACompilerId.cu" failed.

  Compiler: /usr/bin/nvcc

  Build flags:

  Id flags: --keep;--keep-dir;tmp -v



  The output was:

  1

  #$ NVCC_PREPEND_FLAGS="-ccbin clang-14"

  #$ _NVVM_BRANCH_=nvvm

  #$ _SPACE_=

  #$ _CUDART_=cudart

  #$ _HERE_=/usr/lib/nvidia-cuda-toolkit/bin

  #$ _THERE_=/usr/lib/nvidia-cuda-toolkit/bin

  #$ _TARGET_SIZE_=

  #$ _TARGET_DIR_=

  #$ _TARGET_SIZE_=64

  #$ NVVMIR_LIBRARY_DIR=/usr/lib/nvidia-cuda-toolkit/libdevice

  #$
  PATH=/usr/lib/nvidia-cuda-toolkit/bin:/home/tru/bin:/usr/local/cuda/bin:/opt/bin/:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin


  #$ LIBRARIES= -L/usr/lib/aarch64-linux-gnu/stubs
  -L/usr/lib/aarch64-linux-gnu

  #$ rm tmp/a_dlink.reg.c

  #$ clang-14 -D__CUDA_ARCH_LIST__=520 -E -x c++ -D__CUDACC__ -D__NVCC__
  -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=0
  -D__CUDACC_VER_BUILD__=140 -D__CUDA_API_VER_MAJOR__=12
  -D__CUDA_API_VER_MINOR__=0 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include
  "cuda_runtime.h" "CMakeCUDACompilerId.cu" -o
  "tmp/CMakeCUDACompilerId.cpp4.ii"

  #$ cudafe++ --c++14 --clang --clang_version=140000 --display_error_number
  --orig_src_file_name "CMakeCUDACompilerId.cu" --orig_src_path_name
  "/home/tru/git/github/NVIDIA/cuda-samples/build/CMakeFiles/3.28.3/CompilerIdCUDA/CMakeCUDACompilerId.cu"
  --unicode_source_kind=UTF-8 --allow_managed --unsigned_chars
  --unsigned_wchar_t --m64 --parse_templates --gen_c_file_name
  "tmp/CMakeCUDACompilerId.cudafe1.cpp" --stub_file_name
  "CMakeCUDACompilerId.cudafe1.stub.c" --gen_module_id_file
  --module_id_file_name "tmp/CMakeCUDACompilerId.module_id"
  "tmp/CMakeCUDACompilerId.cpp4.ii"

  #$ clang-14 -D__CUDA_ARCH__=520 -D__CUDA_ARCH_LIST__=520 -E -x c++
  -DCUDA_DOUBLE_MATH_FUNCTIONS -D__CUDACC__ -D__NVCC__
  -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=0
  -D__CUDACC_VER_BUILD__=140 -D__CUDA_API_VER_MAJOR__=12
  -D__CUDA_API_VER_MINOR__=0 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include
  "cuda_runtime.h" "CMakeCUDACompilerId.cu" -o
  "tmp/CMakeCUDACompilerId.cpp1.ii"

  #$ cicc --c++14 --clang --clang_version=140000 --display_error_number
  --orig_src_file_name "CMakeCUDACompilerId.cu" --orig_src_path_name
  "/home/tru/git/github/NVIDIA/cuda-samples/build/CMakeFiles/3.28.3/CompilerIdCUDA/CMakeCUDACompilerId.cu"
  --unicode_source_kind=UTF-8 --allow_managed --unsigned_chars
  --unsigned_wchar_t -arch compute_52 -m64 --no-version-ident -ftz=0
  -prec_div=1 -prec_sqrt=1 -fmad=1 --include_file_name
  "CMakeCUDACompilerId.fatbin.c" -tused --module_id_file_name
  "tmp/CMakeCUDACompilerId.module_id" --gen_c_file_name
  "tmp/CMakeCUDACompilerId.cudafe1.c" --stub_file_name
  "tmp/CMakeCUDACompilerId.cudafe1.stub.c" --gen_device_file_name
  "tmp/CMakeCUDACompilerId.cudafe1.gpu" "tmp/CMakeCUDACompilerId.cpp1.ii" -o
  "tmp/CMakeCUDACompilerId.ptx"

  /usr/include/aarch64-linux-gnu/bits/math-vector.h(106): error: identifier
  "__SVFloat32_t" is undefined



  /usr/include/aarch64-linux-gnu/bits/math-vector.h(107): error: identifier
  "__SVFloat64_t" is undefined



  /usr/include/aarch64-linux-gnu/bits/math-vector.h(108): error: identifier
  "__SVBool_t" is undefined



  3 errors detected in the compilation of "CMakeCUDACompilerId.cu".

  # --error 0x1 --





Call Stack (most recent call first):
  /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:8 (CMAKE_DETERMINE_COMPILER_ID_BUILD)
  /usr/share/cmake-3.28/Modules/CMakeDetermineCompilerId.cmake:53 (__determine_compiler_id_test)
  /usr/share/cmake-3.28/Modules/CMakeDetermineCUDACompiler.cmake:135 (CMAKE_DETERMINE_COMPILER_ID)
  CMakeLists.txt:3 (project)


-- Configuring incomplete, errors occurred!
2

No issues. I was able to compile the cuda-samples using nvcr.io/nvidia/cuda:13.0.2-devel-ubuntu24.04 container.

3

Thanks for the quick reply :P

This is something I caused on the bare metal machine, I also don’t have the issue when using the nvcr.io/nvidia/cuda:13.0.2-devel-ubuntu24.04 docker based image with apptainer.

now diffing the installed packages..

Tru

4

The root cause was some leftover of nvidia-cuda-toolkit from version 12… once apt remove nvidia-cuda-toolkit and apt auto-remove steps later, the bare metal installation is fixed.

Thanks

Tru