Isaac Sim Startup Error (cannot load cublas dll)

When I launch Isaac Sim from the Omniverse launcher, I get the following error:

2022-06-29 09:00:01 [5,663ms] [Error] [omni.ext.impl._internal] Failed to import python module omni.isaac.core from c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core. Error: [WinError 1114] A dynamic link library (DLL) initialization routine failed. Error loading "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.ml_archive\pip_prebundle\torch\lib\cublas64_11.dll" or one of its dependencies.. Traceback:
Traceback (most recent call last):
  File "c:/nvidia omniverse/pkg/isaac_sim-2022.1.0/kit/plugins/bindings-python\omni\ext\impl\_internal.py", line 151, in _custom_importer
    return _import_public(ext_module.path, ext_module.name)
  File "c:/nvidia omniverse/pkg/isaac_sim-2022.1.0/kit/plugins/bindings-python\omni\ext\impl\_internal.py", line 78, in _import_public
    module = import_module(module_name)
  File "c:/nvidia omniverse/pkg/isaac_sim-2022.1.0/kit/plugins/bindings-python\carb\profiler\__init__.py", line 78, in wrapper
    r = f(*args, **kwds)
  File "c:/nvidia omniverse/pkg/isaac_sim-2022.1.0/kit/plugins/bindings-python\omni\ext\impl\custom_importer.py", line 76, in import_module
    return importlib.import_module(name)
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\kit\python\lib\importlib\__init__.py", line 127, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 1006, in _gcd_import
  File "<frozen importlib._bootstrap>", line 983, in _find_and_load
  File "<frozen importlib._bootstrap>", line 967, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 677, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 728, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core\omni\isaac\core\__init__.py", line 10, in <module>
    from omni.isaac.core.simulation_context.simulation_context import SimulationContext
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core\omni\isaac\core\simulation_context\__init__.py", line 9, in <module>
    from omni.isaac.core.simulation_context.simulation_context import SimulationContext
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core\omni\isaac\core\simulation_context\simulation_context.py", line 24, in <module>
    import omni.isaac.core.utils.torch as torch_utils
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core\omni\isaac\core\utils\torch\__init__.py", line 1, in <module>
    from .rotations import *
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core\omni\isaac\core\utils\torch\rotations.py", line 5, in <module>
    import torch
  File "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.ml_archive\pip_prebundle\torch\__init__.py", line 127, in <module>
    raise err
OSError: [WinError 1114] A dynamic link library (DLL) initialization routine failed. Error loading "c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.ml_archive\pip_prebundle\torch\lib\cublas64_11.dll" or one of its dependencies.

2022-06-29 09:00:01 [5,665ms] [Error] [carb.scripting-python.plugin] Exception: Extension python module: 'omni.isaac.core' in 'c:\nvidia omniverse\pkg\isaac_sim-2022.1.0\exts\omni.isaac.core' failed to load.

Apparently, he cublas64_11.dll fails to load when importing torch in some files of the isaac core extension. Isaac Sim starts nevertheless and I can use it to some extent, but I can’t use, for example, writers as they depend on the core module.
I’m working on a Windows 10 operating system with a RTX 3090. Here is the first line of the output of nvidia-smi:
NVIDIA-SMI 511.79 Driver Version: 511.79 CUDA Version: 11.6

I’ve got CUDA 11.6 installed, as I need it for other applications. Could there be a problem with conflicting versions? If yes, how could I resolve this?

Using a regular python install (or anaconda) if you install
torch==1.11.0+cu113 does it import correctly?

Yes, if I use anaconda (python 3.7 environment, conda install pytorch==1.11.0 cudatoolkit=11.3 -c pytorch) and open a python shell, I can use import torch without problems.

I’m not sure what the issue is currently, will see if I can find a better way to reproduce it on our end