[Bug]: profiler crashes when profiling with torch multi processing

[itej89]

Describe the bug

I tried to profile a script that contains a main method like this

if __name__ == '__main__':
    num_processes = 8
    torch.multiprocessing.spawn(test_loop, args=(num_processes, ), nprocs=num_processes)

command

/opt/rocm/bin/rocprof-compute  profile -n perf_data  -- python3 ./test.py

Error:
The profiler crashes after one iteration saying "An instance of rocprof is already running"

Workaround(or single rank hack) for now

# Launch following on RANK 0
/opt/rocm/bin/rocprof-compute  profile -n perf_data  -- python3 ./test.py

# Keep closing and relaunching following on the reamining launch for every iteration of /opt/rocm/bin/rocprof-compute
python3 ./test.py

Linux Distribution

Ubuntu 22.04

ROCm Compute Profiler Version

3.0.0

GPU

MI300X

ROCm Version

No response

Cluster name (if applicable)

No response

Reproducer

a script that contains a main method like this

if __name__ == '__main__':
    num_processes = 8
    torch.multiprocessing.spawn(test_loop, args=(num_processes, ), nprocs=num_processes)

command

/opt/rocm/bin/rocprof-compute  profile -n perf_data  -- python3 ./test.py

Expected behavior

No response

Relevant log output

Screenshots

No response

Additional Context

No response