Some issue with Cuda 10.2 in CLion
Answered
I was try to develop a cuda app in the NVIDIA Jetson TX1. I got those error and notice.
1. Reference to __host__ variable 'threadIdx' in __global__ function
2. Use of undeclared identifier cudaConfigureCall
My cuda version is 10.2, nvcc -V output:
$nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_21:14:42_PDT_2019
Cuda compilation tools, release 10.2, V10.2.89
$uname -a
Linux yuzuki-tx1 4.9.201-tegra #1 SMP PREEMPT Fri Jul 9 08:56:29 PDT 2021 aarch64 aarch64 aarch64 GNU/Linux
But cmake build is fine
/usr/bin/cmake --build /home/yuzuki/cuTEST/cmake-build-debug --target cuTEST -- -j 3
Scanning dependencies of target cuTEST
[ 33%] Building CUDA object CMakeFiles/cuTEST.dir/main.cu.o
[ 66%] Linking CUDA device code CMakeFiles/cuTEST.dir/cmake_device_link.o
[100%] Linking CUDA executable cuTEST
[100%] Built target cuTEST
The CLion is updated
Please sign in to leave a comment.
Hello!
Where is CLion installed? Do you use the full remote mode?
CLion is installed in my x64 PC, I do using full remote mode
CUDA attributes are not yet supported in the full remote mode - https://youtrack.jetbrains.com/issue/CPP-18964. Feel free to comment or upvote the issue. See https://intellij-support.jetbrains.com/hc/en-us/articles/207241135-How-to-follow-YouTrack-issues-and-receive-notifications if you are not familiar with YouTrack.