Clang: error: no library 'libomptarget-new-nvptx-sm_35.bc' found in the default clang

Runtimes CompilerRT
, ,
1

Hello,

I hope you are all fine and look forward to your help.

I am trying to compile and run, on a GPU NVIDIA GeForce GTX 1060, a C program using Linux Ubuntu 22.04.1 LTS.

First I followed these steps to install LLVM: Install LLVM on Ubuntu 22.04 (linuxhint.com)

I tried to compile this program:

#include

int main(){
int N = 1e8;
float *x = new float[N];
float *y = new float[N];

#pragma omp target teams distribute parallel for map(to: x[0:N]) map(from: y[0:N])
for(int i = 0; i < N; i++){
for(int j = 0; j < 1000; j++){
y[i] += 3*x[i];
}
}

printf(“%g\n”, y[5]);

delete x;
delete y;
}

Script to compile: clang++ -fopenmp -fopenmp-targets=nvptx64 -O3 run.cpp && time ./a.out

And I received the error on the subject.

Then I followed these steps: Build Clang with OpenMP Target Offloading for NVIDIA GPUs · GitHub

And I received the same error.

So now, my questions are:

  1. How can I compile and run, on the indicated GPU, the program above, using LLVM?
  2. How can I determine which versions of LLVM (clang) I have installed in my Linux?
  3. How can I remove from my Linux the versions of LLVM that I am not interested in having?

Thank you very much in advance.

Best regards,
David.

2

The build steps there are outdated and may not work anymore, use the CMake flags described in our FAQ:
https://openmp.llvm.org/SupportAndFAQ.html#q-how-to-build-an-openmp-gpu-offload-capable-compiler