1

I am trying to write a basic program to read a csv into a cudf with libcudf. I have installed libcudf by installing the RAPIDS conda environment. I am trying to use this command to build my executable:

nvcc -c csv_read.cu -o csv_read.o -I$CONDA_PREFIX/include -L$CONDA_PREFIX/lib*

I keep producing the error fatal error: spdlog/fmt/bundled/core.h: No such file or directory. with regards to this line: include <spdlog/fmt/bundled/core.h>. Upon inspection of the spdlog/fmt directory only these files exist: bin_to_hex.h chrono.h compile.h fmt.h ostr.h ranges.h std.h xchar.h

No sign of a bundled directory.

These are my include statements for reference:

#include <cudf/io/csv.hpp>
#include <cudf/table/table_view.hpp>
#include <cudf/table/table.hpp>
#include <cudf/column/column_view.hpp>

Any advice on how to resolve this issue would be greatly appreciated.

I have tried using commands such as this to build the code: nvcc -c multi_gpu_sort.cu -o multi_gpu_sort.o -I$CONDA_PREFIX/include -L$CONDA_PREFIX/lib -lspdlog -lfmt

I have also tried specifying the exact include paths to spdlog and fmt.

Improve this question
6
  • can you post the output of conda list for the environment from which you are trying to compile this? The RAPIDS environment is quite unclear in terms of reproducibility of your issue Commented Jun 17, 2024 at 13:09
  • Yes, it would be important to know the versions of spdlog and fmt that are being picked up. These dependencies should be coming from the conda environment, so conda list should help. Commented Jun 17, 2024 at 20:52
  • @FlyingTeller I will edit the OP to show output of conda list. I am using the conda environment you can create from the nvidia rapids AI suite found at this link [](docs.rapids.ai/install?_gl=1). Here is the command I used to create the conda environment conda create -n rapids-24.06 -c rapidsai -c conda-forge -c nvidia \ rapids=24.06 python=3.11 cuda-version=11.8 Commented Jun 18, 2024 at 20:32
  • 1
    @FlyingTeller the output of conda list is quite long, but if you want to reproduce the issue you can create the conda environment from my other comment and try to compile a .cu file with the include statements from OP. A command such as nvcc -c your_file.cu -o your_file.o -I$CONDA_PREFIX/include -L$CONDA_PREFIX/lib should do the trick. Commented Jun 18, 2024 at 20:46
  • @BradleyDice the output of conda list is quite long, but if you want to reproduce the issue you can check link and try to create your own environment. Here is the exact command I used: conda create -n rapids-24.06 -c rapidsai -c conda-forge -c nvidia \ rapids=24.06 python=3.11 cuda-version=11.8. Btw, from conda list: spdlog 1.12.0 hd2e6256_2 conda-forge, fmt 10.2.1 h00ab1b0_0 conda-forge Commented Jun 18, 2024 at 20:54
Related questions
0
I am trying to install cudf from source for conda, I cannot use cmake to install it
0
cuDF installation failure: "Didn't find wheel for cudf-cu12 24.6.1"
0
I'm having trouble understanding what's wrong with my linking to the include folder of spdlog in Visual Studio 2022
Load 6 more related questions Show fewer related questions

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.