Installing CUDA-Q using the script does not work

[diehlpk]

Required prerequisites

• Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
• Make sure you've read the documentation. Your issue may be addressed there.
• Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.
• If possible, make a PR with a failing test to give us a starting point to work on!

Describe the bug

After installing CUDA-Q, I can not compile one of the examples.

I copied the code from here to the file text.cxx

// Compile and run with:
// ```
// nvq++ cuquantum_backends.cpp -o dyn.x --target nvidia && ./dyn.x
// ```

// This example is meant to demonstrate the cuQuantum
// GPU-accelerated backends and their ability to easily handle
// a larger number of qubits compared the CPU-only backend.

// On CPU-only backends, this seems to hang, i.e. it takes a long
// time to handle this number of qubits.

#include <cudaq.h>

// Define a quantum kernel with a runtime parameter
struct ghz {
  auto operator()(const int N) __qpu__ {

    // Dynamically sized vector of qubits
    cudaq::qvector q(N);
    h(q[0]);
    for (int i = 0; i < N - 1; i++) {
      x<cudaq::ctrl>(q[i], q[i + 1]);
    }
    mz(q);
  }
};

int main() {
  auto counts = cudaq::sample(/*shots=*/100, ghz{}, 28);

  if (!cudaq::mpi::is_initialized() || cudaq::mpi::rank() == 0) {
    counts.dump();

    // Fine grain access to the bits and counts
    for (auto &[bits, count] : counts) {
      printf("Observed: %s, %lu\n", bits.data(), count);
    }
  }

  return 0;
}

and get the following error:

 /vast/home/diehlpk/Compile/cudaq/bin/nvq++ --target nvidia test.cc -o test.x
Could not find a valid clang resource-dir.

Steps to reproduce the bug

Install CUDA-Q using ./install_cuda_quantum_cu12.aarch64 --accept --target ./cudaq and use the installed nvq++ compiler.

Expected behavior

The example should compile.

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

• CUDA-Q version:
• Python version:
• C++ compiler:
• Operating system:

Suggestions

No response

[bmhowe23]

Hi @diehlpk - thank you for using CUDA-Q. According to the docs (quoted below), admin privileges are required to install CUDA-Q using the installer method that you described above. When you installed it, did you have admin privileges?

The installation location of CUDA-Q is not currently configurable and using the installer hence requires admin privileges on the system. We may revise that in the future; please see and upvote the corresponding GitHub issue.

If you do not have admin privileges, you can often use our Docker image (either through the Docker daemon or through Singularity as described on the aforementioned webpage). Do those options work for you? In either case, you may want to comment on issue #1075.

[diehlpk]

@bmhowe23 I do not have root access on our Grace Hopper cluster and tried to install into user space. I do not want a system wide installation yet since I explore the software. Containers are not an option yet. I tried to install from source but that failed as well.

[diehlpk]

Hi @bmhowe23 I used the container as recommend, however, that does not work as well.

nvq++ test.cc                  
mv: cannot stat 'test.cc.ll': No such file or directory
ld.lld: error: undefined symbol: main
>>> referenced by /lib/aarch64-linux-gnu/Scrt1.o:(_start)
>>> referenced by /lib/aarch64-linux-gnu/Scrt1.o:(_start)
clang-16: error: linker command failed with exit code 1 (use -v to see invocation)
failed: "/usr/local/llvm/bin/clang++ --ld-path=/usr/local/llvm/bin/ld.lld -Wl,-rpath,/opt/nvidia/cudaq/lib -Wl,-rpath,/opt/nvidia/cudaq/lib/plugins -Wl,-rpath,/home/git/distributed-quantum-simualtor/build -L/opt/nvidia/cudaq/lib -L/opt/nvidia/cudaq/lib/plugins test.cc.o -lcudaq -lcudaq-common -lcudaq-ensmallen -lcudaq-nlopt -lcudaq-spin -lcudaq-comm-plugin -lcudaq-pyscf -lcudaq-mlir-runtime -lcudaq-builder -lcudaq-em-default -lcudaq-platform-default -lnvqir -lnvqir-qpp"

and with --target nvidia, I get

nvq++ --target nvidia  test.cc 
mv: cannot stat 'test.cc.ll': No such file or directory
ld.lld: error: undefined symbol: main
>>> referenced by /lib/aarch64-linux-gnu/Scrt1.o:(_start)
>>> referenced by /lib/aarch64-linux-gnu/Scrt1.o:(_start)
clang-16: error: linker command failed with exit code 1 (use -v to see invocation)
failed: "/usr/local/llvm/bin/clang++ --ld-path=/usr/local/llvm/bin/ld.lld -Wl,-rpath,/opt/nvidia/cudaq/lib -Wl,-rpath,/opt/nvidia/cudaq/lib/plugins -Wl,-rpath,/home/git/distributed-quantum-simualtor/build -L/opt/nvidia/cudaq/lib -L/opt/nvidia/cudaq/lib/plugins test.cc.o -lcudaq -lcudaq-common -lcudaq-ensmallen -lcudaq-nlopt -lcudaq-spin -lcudaq-comm-plugin -lcudaq-pyscf -lcudaq-mlir-runtime -lcudaq-builder -lcudaq-em-default -lcudaq-platform-default -lnvqir -lnvqir-qpp"

[diehlpk]

Ok, the file has to end with .cpp and the compiler works, you can not use .cxx or .cc.

[diehlpk]

Ok, that issue can be closed since the Docker image works. However, it would be nice to have the installer working without root.