11.x builds failing on system without nvcc

[charlesbluca]

When attempting to build UCXX with the CUDA 11.8 conda environment on a system without nvcc pre-installed (i.e. all CTK components being installed through conda), I get the following error at build configuration:

CMake Error at /home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/cmake/rmm/rmm-targets.cmake:61 (set_target_properties):
  The link interface of target "rmm::rmm" contains:

    CUDA::cudart

  but the target was not found.  Possible reasons include:

    * There is a typo in the target name.
    * A find_package call is missing for an IMPORTED target.
    * An ALIAS target is missing.

Call Stack (most recent call first):
  /home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/cmake/rmm/rmm-config.cmake:75 (include)
  build/cmake/CPM_0.40.0.cmake:249 (find_package)
  build/cmake/CPM_0.40.0.cmake:303 (cpm_find_package)
  build/_deps/rapids-cmake-src/rapids-cmake/cpm/find.cmake:189 (CPMFindPackage)
  build/_deps/rapids-cmake-src/rapids-cmake/cpm/rmm.cmake:75 (rapids_cpm_find)
  cmake/thirdparty/get_rmm.cmake:20 (rapids_cpm_rmm)
  cmake/thirdparty/get_rmm.cmake:24 (find_and_configure_rmm)
  CMakeLists.txt:112 (include)

This was somewhat confusing, as the conda install itself raised a warning message that implied I should have libcudart in the conda environment:

To enable CUDA support, UCX requires the CUDA Runtime library (libcudart).
The library can be installed with the appropriate command below:

* For CUDA 11, run:    conda install cudatoolkit cuda-version=11
* For CUDA 12, run:    conda install cuda-cudart cuda-version=12

→ conda list cuda
# packages in environment at /home/charlesb/miniforge3/envs/ucxx-cuda-118:
#
# Name                    Version                   Build  Channel
cuda-version              11.8                 h70ddcb2_3    conda-forge
cudatoolkit               11.8.0              h4ba93d1_13    conda-forge

→ find $CONDA_PREFIX -name "libcudart.so*"
/home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/libcudart.so
/home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/libcudart.so.11.0
/home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/libcudart.so.11.8.89

Saw that these failures were coming up in the configuration of RMM, so tried building that with its accompanying 11.8 conda environment and got a somewhat clearer error that it was unable to find an installation of CTK on my system (nvcc bin was missing):

/home/charlesb/miniforge3/envs/rmm-cuda-118/bin/nvcc: line 9: /bin/nvcc: No such file or directory
...
CMake Error at /home/charlesb/miniforge3/envs/rmm-cuda-118/share/cmake-3.30/Modules/FindPackageHandleStandardArgs.cmake:233 (message):
  Could NOT find CUDAToolkit (missing: CUDAToolkit_INCLUDE_DIRECTORIES)
Call Stack (most recent call first):
  /home/charlesb/miniforge3/envs/rmm-cuda-118/share/cmake-3.30/Modules/FindPackageHandleStandardArgs.cmake:603 (_FPHSA_FAILURE_MESSAGE)
  /home/charlesb/miniforge3/envs/rmm-cuda-118/share/cmake-3.30/Modules/FindCUDAToolkit.cmake:1048 (find_package_handle_standard_args)
  build/_deps/rapids-cmake-src/rapids-cmake/find/package.cmake:125 (find_package)
  CMakeLists.txt:62 (rapids_find_package)

Was unable to reproduce this with the 12.5 environment, which does pull a conda installation of nvcc.

About to do a system installation of CTK on the system to see if this unblocks

[pentschev]

As far as I remember, you need to install nvcc_linux-64=11.8, could you check if that works?

[charlesbluca]

Installing that, it seems like the resulting nvcc bin is just wrapping what I assume is a system installation of nvcc?

→ conda list nvcc
# packages in environment at /home/charlesb/miniforge3/envs/ucxx-cuda-118:
#
# Name                    Version                   Build  Channel
nvcc_linux-64             11.8                h9852d18_24    conda-forge

→ nvcc
/home/charlesb/miniforge3/envs/ucxx-cuda-118/bin/nvcc: line 9: /bin/nvcc: No such file or directory

→ cat /home/charlesb/miniforge3/envs/ucxx-cuda-118/bin/nvcc
#!/bin/bash

for arg in "${@}" ; do
  case ${arg} in -ccbin)
    # If -ccbin argument is already provided, don't add an additional one.
    exec "${CUDA_HOME}/bin/nvcc" "${@}"
  esac
done
exec "${CUDA_HOME}/bin/nvcc" -ccbin "${CXX}" "${@}"

[pentschev]

IIRC, with CUDA 11.x CUDA_HOME is redefined during conda activate. Can you check if deactivating and reactivating your environment changes the behavior?

[pentschev]

By "redefined" I mean it should be redefined to $CONDA_PREFIX.

[charlesbluca]

Ah thanks for that tip - this highlights what I assume is an underlying issue here, in that we aren't able to locate the CUDA_HOME during environment activation:

→ conda activate ucxx-cuda-118
Cannot determine CUDA_HOME: cuda-gdb not in PATH

This warning specifically starts popping up with the installation of nvcc_linux-64 in the environment

[pentschev]

I had this discussion with @robertmaynard in the past, his answer was:

conda nvcc scrip uses cuda-gdb to determine the cuda install location if CUDA_HOME hasn't been explicitly set beforehand so if the machine doesn't have cuda-gdb the conda activation scripts will fail to setup CUDA_HOME, which you will need to do manually

So yeah, I think you need a system install of CTK for CUDA 11.x to be able to compile.

[charlesbluca]

Thanks @pentschev, installed CTK 12.5 on my system (seemingly the oldest version available for ubuntu24.04 right now), and that unblocked builds.

Moving forward, can or should we explicitly encode a CTK dependency similar to what RMM is doing in its CMakeLists.txt?

https://github.com/rapidsai/rmm/blob/c494395e58288cac16321ce90e9b15f3508ae89a/CMakeLists.txt#L62-L65

Or is this too brittle of a solution, with just general documentation of system installing CTK for 11.x builds making more sense?

[charlesbluca]

Also worth noting that it seems like there's more required than just proper setting of CUDA_HOME here, as even manually setting it to the CONDA_PREFIX above that I can see contains libcudart seems to raise the same failures

[robertmaynard]

Also worth noting that it seems like there's more required than just proper setting of CUDA_HOME here, as even manually setting it to the CONDA_PREFIX above that I can see contains libcudart seems to raise the same failures

Can you try setting the env variable CUDA_PATH that is what is used by CMake ( not CUDA_HOME ).

[charlesbluca]

Thanks for the tip - looks like that's still failing. For reference the command I'm working with:

$ CUDA_PATH=/home/charlesb/miniforge3/envs/ucxx-cuda-118 ./build.sh
-- The C compiler identification is GNU 13.3.0
-- The CXX compiler identification is GNU 13.3.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /home/charlesb/miniforge3/envs/ucxx-cuda-118/bin/x86_64-conda-linux-gnu-cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /home/charlesb/miniforge3/envs/ucxx-cuda-118/bin/x86_64-conda-linux-gnu-c++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Failed
-- Check if compiler accepts -pthread
-- Check if compiler accepts -pthread - yes
-- Found Threads: TRUE
-- CPM: Using local package [email protected]
-- Configuring done (2.1s)
CMake Error at /home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/cmake/rmm/rmm-targets.cmake:61 (set_target_properties):
  The link interface of target "rmm::rmm" contains:

    CUDA::cudart

  but the target was not found.  Possible reasons include:

    * There is a typo in the target name.
    * A find_package call is missing for an IMPORTED target.
    * An ALIAS target is missing.

Call Stack (most recent call first):
  /home/charlesb/miniforge3/envs/ucxx-cuda-118/lib/cmake/rmm/rmm-config.cmake:75 (include)
  build/cmake/CPM_0.40.0.cmake:249 (find_package)
  build/cmake/CPM_0.40.0.cmake:303 (cpm_find_package)
  build/_deps/rapids-cmake-src/rapids-cmake/cpm/find.cmake:189 (CPMFindPackage)
  build/_deps/rapids-cmake-src/rapids-cmake/cpm/rmm.cmake:75 (rapids_cpm_find)
  cmake/thirdparty/get_rmm.cmake:20 (rapids_cpm_rmm)
  cmake/thirdparty/get_rmm.cmake:24 (find_and_configure_rmm)
  CMakeLists.txt:112 (include)


-- Generating done (0.0s)
CMake Generate step failed.  Build files cannot be regenerated correctly.

Output of conda list:

# packages in environment at /home/charlesb/miniforge3/envs/ucxx-cuda-118:
#
# Name                    Version                   Build  Channel
_libgcc_mutex             0.1                 conda_forge    conda-forge
_openmp_mutex             4.5                       2_gnu    conda-forge
attrs                     24.2.0             pyh71513ae_0    conda-forge
autoconf                  2.71            pl5321h2b4cb7a_1    conda-forge
automake                  1.17            pl5321ha770c72_0    conda-forge
aws-c-auth                0.7.31               h57bd9a3_0    conda-forge
aws-c-cal                 0.7.4                hfd43aa1_1    conda-forge
aws-c-common              0.9.28               hb9d3cd8_0    conda-forge
aws-c-compression         0.2.19               h756ea98_1    conda-forge
aws-c-event-stream        0.4.3                h29ce20c_2    conda-forge
aws-c-http                0.8.10               h5e77a74_0    conda-forge
aws-c-io                  0.14.18             h4e6ae90_11    conda-forge
aws-c-mqtt                0.10.6               h02abb05_0    conda-forge
aws-c-s3                  0.6.6                h834ce55_0    conda-forge
aws-c-sdkutils            0.1.19               h756ea98_3    conda-forge
aws-checksums             0.1.20               h756ea98_0    conda-forge
aws-crt-cpp               0.28.3               h469002c_5    conda-forge
aws-sdk-cpp               1.11.407             h9f1560d_0    conda-forge
azure-core-cpp            1.13.0               h935415a_0    conda-forge
azure-identity-cpp        1.9.0                hd126650_0    conda-forge
azure-storage-blobs-cpp   12.13.0              h1d30c4a_0    conda-forge
azure-storage-common-cpp  12.8.0               ha3822c6_0    conda-forge
azure-storage-files-datalake-cpp 12.12.0              h0f25b8a_0    conda-forge
binutils                  2.43                 h4852527_1    conda-forge
binutils_impl_linux-64    2.43                 h4bf12b8_1    conda-forge
binutils_linux-64         2.43                 h4852527_1    conda-forge
bokeh                     3.5.2              pyhd8ed1ab_0    conda-forge
brotli-python             1.1.0           py312h2ec8cdc_2    conda-forge
bzip2                     1.0.8                h4bc722e_7    conda-forge
c-ares                    1.33.1               heb4867d_0    conda-forge
c-compiler                1.8.0                h2b85faf_0    conda-forge
ca-certificates           2024.8.30            hbcca054_0    conda-forge
cachetools                5.5.0              pyhd8ed1ab_0    conda-forge
cffi                      1.17.1          py312h06ac9bb_0    conda-forge
cfgv                      3.3.1              pyhd8ed1ab_0    conda-forge
click                     8.1.7           unix_pyh707e725_0    conda-forge
cloudpickle               3.0.0              pyhd8ed1ab_0    conda-forge
cmake                     3.30.4               hf9cb763_0    conda-forge
colorama                  0.4.6              pyhd8ed1ab_0    conda-forge
contourpy                 1.3.0           py312h68727a3_2    conda-forge
cubinlinker               0.3.0           py312hbe86355_1    rapidsai
cuda-python               11.8.3          py312h32b3722_2    conda-forge
cuda-version              11.8                 h70ddcb2_3    conda-forge
cudatoolkit               11.8.0              h4ba93d1_13    conda-forge
cudf                      24.12.00a150    cuda11_py312_241007_gfcff2b6ef7_150    rapidsai-nightly
cupy                      13.3.0          py312h8e83189_0    conda-forge
cupy-core                 13.3.0          py312h53955ab_0    conda-forge
cxx-compiler              1.8.0                h1a2810e_0    conda-forge
cython                    3.0.11          py312h8fd2918_3    conda-forge
cytoolz                   1.0.0           py312h66e93f0_0    conda-forge
dask                      2024.9.0           pyhd8ed1ab_0    conda-forge
dask-core                 2024.9.0           pyhd8ed1ab_0    conda-forge
dask-cuda                 24.12.00a2      py312_241007_gfe16796_2    rapidsai-nightly
dask-cudf                 24.12.00a150    cuda11_py312_241007_gfcff2b6ef7_150    rapidsai-nightly
dask-expr                 1.1.14             pyhd8ed1ab_0    conda-forge
distlib                   0.3.8              pyhd8ed1ab_0    conda-forge
distributed               2024.9.0           pyhd8ed1ab_0    conda-forge
dlpack                    0.8                  h59595ed_3    conda-forge
doxygen                   1.9.1                hb166930_1    conda-forge
exceptiongroup            1.2.2              pyhd8ed1ab_0    conda-forge
fastrlock                 0.8.2           py312h30efb56_2    conda-forge
filelock                  3.16.1             pyhd8ed1ab_0    conda-forge
fmt                       11.0.2               h434a139_0    conda-forge
freetype                  2.12.1               h267a509_2    conda-forge
fsspec                    2024.9.0           pyhff2d567_0    conda-forge
gcc                       13.3.0               h9576a4e_1    conda-forge
gcc_impl_linux-64         13.3.0               hfea6d02_1    conda-forge
gcc_linux-64              13.3.0               hc28eda2_4    conda-forge
gflags                    2.2.2             h5888daf_1005    conda-forge
glog                      0.7.1                hbabe93e_0    conda-forge
gxx                       13.3.0               h9576a4e_1    conda-forge
gxx_impl_linux-64         13.3.0               hdbfa832_1    conda-forge
gxx_linux-64              13.3.0               h6834431_4    conda-forge
h2                        4.1.0              pyhd8ed1ab_0    conda-forge
hpack                     4.0.0              pyh9f0ad1d_0    conda-forge
hyperframe                6.0.1              pyhd8ed1ab_0    conda-forge
icu                       75.1                 he02047a_0    conda-forge
identify                  2.6.1              pyhd8ed1ab_0    conda-forge
importlib-metadata        8.5.0              pyha770c72_0    conda-forge
importlib-resources       6.4.5              pyhd8ed1ab_0    conda-forge
importlib_metadata        8.5.0                hd8ed1ab_0    conda-forge
importlib_resources       6.4.5              pyhd8ed1ab_0    conda-forge
iniconfig                 2.0.0              pyhd8ed1ab_0    conda-forge
jinja2                    3.1.4              pyhd8ed1ab_0    conda-forge
jsonschema                4.23.0             pyhd8ed1ab_0    conda-forge
jsonschema-specifications 2023.12.1          pyhd8ed1ab_0    conda-forge
kernel-headers_linux-64   3.10.0              he073ed8_17    conda-forge
keyutils                  1.6.1                h166bdaf_0    conda-forge
krb5                      1.21.3               h659f571_0    conda-forge
lcms2                     2.16                 hb7c19ff_0    conda-forge
ld_impl_linux-64          2.43                 h712a8e2_1    conda-forge
lerc                      4.0.0                h27087fc_0    conda-forge
libabseil                 20240722.0      cxx17_h5888daf_1    conda-forge
libarrow                  17.0.0          h364f349_19_cpu    conda-forge
libarrow-acero            17.0.0          h5888daf_19_cpu    conda-forge
libarrow-dataset          17.0.0          h5888daf_19_cpu    conda-forge
libarrow-substrait        17.0.0          he882d9a_19_cpu    conda-forge
libblas                   3.9.0           24_linux64_openblas    conda-forge
libbrotlicommon           1.1.0                hb9d3cd8_2    conda-forge
libbrotlidec              1.1.0                hb9d3cd8_2    conda-forge
libbrotlienc              1.1.0                hb9d3cd8_2    conda-forge
libcblas                  3.9.0           24_linux64_openblas    conda-forge
libcrc32c                 1.1.2                h9c3ff4c_0    conda-forge
libcudf                   24.12.00a150    cuda11_241007_gfcff2b6ef7_150    rapidsai-nightly
libcufile                 1.4.0.31                      0    nvidia
libcufile-dev             1.4.0.31                      0    nvidia
libcurl                   8.10.1               hbbe4b11_0    conda-forge
libdeflate                1.22                 hb9d3cd8_0    conda-forge
libedit                   3.1.20191231         he28a2e2_2    conda-forge
libev                     4.33                 hd590300_2    conda-forge
libevent                  2.1.12               hf998b51_1    conda-forge
libexpat                  2.6.3                h5888daf_0    conda-forge
libffi                    3.4.2                h7f98852_5    conda-forge
libgcc                    14.1.0               h77fa898_1    conda-forge
libgcc-devel_linux-64     13.3.0             h84ea5a7_101    conda-forge
libgcc-ng                 14.1.0               h69a702a_1    conda-forge
libgfortran               14.1.0               h69a702a_1    conda-forge
libgfortran-ng            14.1.0               h69a702a_1    conda-forge
libgfortran5              14.1.0               hc5f4f2c_1    conda-forge
libgomp                   14.1.0               h77fa898_1    conda-forge
libgoogle-cloud           2.29.0               h438788a_1    conda-forge
libgoogle-cloud-storage   2.29.0               h0121fbd_1    conda-forge
libgrpc                   1.65.5               hf5c653b_0    conda-forge
libiconv                  1.17                 hd590300_2    conda-forge
libjpeg-turbo             3.0.0                hd590300_1    conda-forge
libkvikio                 24.12.00a       cuda11_241007_ge64c363_20    rapidsai-nightly
liblapack                 3.9.0           24_linux64_openblas    conda-forge
libllvm14                 14.0.6               hcd5def8_4    conda-forge
libnghttp2                1.58.0               h47da74e_1    conda-forge
libnl                     3.10.0               h4bc722e_0    conda-forge
libnsl                    2.0.1                hd590300_0    conda-forge
libopenblas               0.3.27          pthreads_hac2b453_1    conda-forge
libparquet                17.0.0          h6bd9018_19_cpu    conda-forge
libpng                    1.6.44               hadc24fc_0    conda-forge
libprotobuf               5.27.5               h5b01275_2    conda-forge
libre2-11                 2023.09.01           hbbce691_3    conda-forge
librmm                    24.12.00a9      cuda11_241007_gc494395e_9    rapidsai-nightly
libsanitizer              13.3.0               heb74ff8_1    conda-forge
libsqlite                 3.46.1               hadc24fc_0    conda-forge
libssh2                   1.11.0               h0841786_0    conda-forge
libstdcxx                 14.1.0               hc0a3c3a_1    conda-forge
libstdcxx-devel_linux-64  13.3.0             h84ea5a7_101    conda-forge
libstdcxx-ng              14.1.0               h4852527_1    conda-forge
libthrift                 0.21.0               h0e7cc3e_0    conda-forge
libtiff                   4.7.0                he137b08_1    conda-forge
libtool                   2.4.7                he02047a_1    conda-forge
libutf8proc               2.8.0                h166bdaf_0    conda-forge
libuuid                   2.38.1               h0b41bf4_0    conda-forge
libuv                     1.49.0               hb9d3cd8_0    conda-forge
libwebp-base              1.4.0                hd590300_0    conda-forge
libxcb                    1.17.0               h8a09558_0    conda-forge
libxcrypt                 4.4.36               hd590300_1    conda-forge
libxml2                   2.12.7               he7c6b58_4    conda-forge
libzlib                   1.3.1                hb9d3cd8_2    conda-forge
llvmlite                  0.43.0          py312h374181b_1    conda-forge
locket                    1.0.0              pyhd8ed1ab_0    conda-forge
lz4                       4.3.3           py312hb3f7f12_1    conda-forge
lz4-c                     1.9.4                hcb278e6_0    conda-forge
m4                        1.4.18            h516909a_1001    conda-forge
markdown-it-py            3.0.0              pyhd8ed1ab_0    conda-forge
markupsafe                2.1.5           py312h66e93f0_1    conda-forge
mdurl                     0.1.2              pyhd8ed1ab_0    conda-forge
msgpack-python            1.1.0           py312h68727a3_0    conda-forge
ncurses                   6.5                  he02047a_1    conda-forge
ninja                     1.12.1               h297d8ca_0    conda-forge
nodeenv                   1.9.1              pyhd8ed1ab_0    conda-forge
numba                     0.60.0          py312h83e6fd3_0    conda-forge
numba-cuda                0.0.15             pyh267e887_1    conda-forge
numpy                     2.0.2           py312h58c1407_0    conda-forge
nvcomp                    4.0.1                hee583db_0    conda-forge
nvtx                      0.2.10          py312h66e93f0_2    conda-forge
openjpeg                  2.5.2                h488ebb8_0    conda-forge
openssl                   3.3.2                hb9d3cd8_0    conda-forge
orc                       2.0.2                h690cf93_1    conda-forge
packaging                 24.1               pyhd8ed1ab_0    conda-forge
pandas                    2.2.3           py312hf9745cd_1    conda-forge
partd                     1.4.2              pyhd8ed1ab_0    conda-forge
pathspec                  0.12.1             pyhd8ed1ab_0    conda-forge
perl                      5.32.1          7_hd590300_perl5    conda-forge
pillow                    10.4.0          py312h56024de_1    conda-forge
pip                       24.2               pyh8b19718_1    conda-forge
pkg-config                0.29.2            h4bc722e_1009    conda-forge
pkgutil-resolve-name      1.3.10             pyhd8ed1ab_1    conda-forge
platformdirs              4.3.6              pyhd8ed1ab_0    conda-forge
pluggy                    1.5.0              pyhd8ed1ab_0    conda-forge
pre-commit                4.0.0              pyha770c72_0    conda-forge
psutil                    6.0.0           py312h66e93f0_1    conda-forge
pthread-stubs             0.4               hb9d3cd8_1002    conda-forge
ptxcompiler               0.8.1           py312h32b3722_4    conda-forge
pyarrow                   17.0.0          py312h9cebb41_1    conda-forge
pyarrow-core              17.0.0          py312h9cafe31_1_cpu    conda-forge
pyarrow-hotfix            0.6                pyhd8ed1ab_0    conda-forge
pycparser                 2.22               pyhd8ed1ab_0    conda-forge
pygments                  2.18.0             pyhd8ed1ab_0    conda-forge
pylibcudf                 24.12.00a150    cuda11_py312_241007_gfcff2b6ef7_150    rapidsai-nightly
pynvml                    11.4.1             pyhd8ed1ab_0    conda-forge
pysocks                   1.7.1              pyha2e5f31_6    conda-forge
pytest                    7.4.4              pyhd8ed1ab_0    conda-forge
pytest-asyncio            0.23.8             pyhd8ed1ab_0    conda-forge
pytest-rerunfailures      14.0               pyhd8ed1ab_0    conda-forge
python                    3.12.7          hc5c86c4_0_cpython    conda-forge
python-dateutil           2.9.0              pyhd8ed1ab_0    conda-forge
python-tzdata             2024.2             pyhd8ed1ab_0    conda-forge
python_abi                3.12                    5_cp312    conda-forge
pytz                      2024.1             pyhd8ed1ab_0    conda-forge
pyyaml                    6.0.2           py312h66e93f0_1    conda-forge
rapids-build-backend      0.3.2                      py_0    rapidsai
rapids-dask-dependency    24.12.00a6                 py_0    rapidsai-nightly
rapids-dependency-file-generator 1.15.0                     py_0    rapidsai
rdma-core                 54.0                 h5888daf_0    conda-forge
re2                       2023.09.01           h77b4e00_3    conda-forge
readline                  8.2                  h8228510_1    conda-forge
referencing               0.35.1             pyhd8ed1ab_0    conda-forge
rhash                     1.4.4                hd590300_0    conda-forge
rich                      13.9.2             pyhd8ed1ab_0    conda-forge
rmm                       24.12.00a9      cuda11_py312_241007_gc494395e_9    rapidsai-nightly
rpds-py                   0.20.0          py312h12e396e_1    conda-forge
s2n                       1.5.4                h1380c3d_0    conda-forge
scikit-build-core         0.10.7             pyh4afc917_0    conda-forge
setuptools                75.1.0             pyhd8ed1ab_0    conda-forge
six                       1.16.0             pyh6c4a22f_0    conda-forge
snappy                    1.2.1                ha2e4443_0    conda-forge
sortedcontainers          2.4.0              pyhd8ed1ab_0    conda-forge
spdlog                    1.14.1               hed91bc2_1    conda-forge
sysroot_linux-64          2.17                h4a8ded7_17    conda-forge
tblib                     3.0.0              pyhd8ed1ab_0    conda-forge
tk                        8.6.13          noxft_h4845f30_101    conda-forge
tomli                     2.0.2              pyhd8ed1ab_0    conda-forge
tomlkit                   0.13.2             pyha770c72_0    conda-forge
toolz                     1.0.0              pyhd8ed1ab_0    conda-forge
tornado                   6.4.1           py312h66e93f0_1    conda-forge
typing-extensions         4.12.2               hd8ed1ab_0    conda-forge
typing_extensions         4.12.2             pyha770c72_0    conda-forge
tzdata                    2024b                hc8b5060_0    conda-forge
ucx                       1.17.0               h0104b51_3    conda-forge
ukkonen                   1.0.1           py312h68727a3_5    conda-forge
urllib3                   2.2.3              pyhd8ed1ab_0    conda-forge
virtualenv                20.26.6            pyhd8ed1ab_0    conda-forge
wheel                     0.44.0             pyhd8ed1ab_0    conda-forge
xorg-libxau               1.0.11               hb9d3cd8_1    conda-forge
xorg-libxdmcp             1.1.5                hb9d3cd8_0    conda-forge
xyzservices               2024.9.0           pyhd8ed1ab_0    conda-forge
xz                        5.2.6                h166bdaf_0    conda-forge
yaml                      0.2.5                h7f98852_2    conda-forge
zict                      3.0.0              pyhd8ed1ab_0    conda-forge
zipp                      3.20.2             pyhd8ed1ab_0    conda-forge
zstandard                 0.23.0          py312hef9b889_1    conda-forge
zstd                      1.5.6                ha6fb4c9_0    conda-forge

[robertmaynard]

I would need a full trace log from CMake to see what is exactly going wrong.

IIRC the command line would be:

CUDA_PATH=/home/charlesb/miniforge3/envs/ucxx-cuda-118 ./build.sh --cmake-args=\"--trace\" > log

[charlesbluca]

Here's a log with CMake traces enabled:

failure.log

[robertmaynard]

Some clarification. The cuda-gdb detection logic is what conda uses to manage finding a local install of CUDA 11.X

CMake uses different logic for finding nvcc and from the extracting the rest of the CUDA Toolkit libraries and headers. @charlesbluca In the trace you provided the FindCUDAToolkit is failing since it can't find nvcc or the sentinel versions files inside the CUDA Toolkit.

I think the primary issue is that CUDA_PATH needs to point not to your conda env, but the local install of the cuda toolkit. E.g /usr/local/cuda-11.8/

[pentschev]

@charlesbluca do you think there's still anything we should do in UCXX for better UX?