diff --git a/heat/core/__init__.py b/heat/core/__init__.py
index 5a506c572..82326ed5b 100644
--- a/heat/core/__init__.py
+++ b/heat/core/__init__.py
@@ -32,3 +32,4 @@
 from . import version
 from .version import __version__
 from .vmap import *
+from ._config import *
diff --git a/heat/core/_config.py b/heat/core/_config.py
new file mode 100644
index 000000000..da327835a
--- /dev/null
+++ b/heat/core/_config.py
@@ -0,0 +1,62 @@
+"""
+Everything you need to know about the configuration of Heat
+"""
+
+import torch
+import platform
+import mpi4py
+import subprocess
+import os
+import warnings
+import re
+
+PLATFORM = platform.platform()
+MPI_LIBRARY_VERSION = mpi4py.MPI.Get_library_version()
+TORCH_VERSION = torch.__version__
+TORCH_CUDA_IS_AVAILABLE = torch.cuda.is_available()
+CUDA_IS_ACTUALLY_ROCM = "rocm" in TORCH_VERSION
+
+CUDA_AWARE_MPI = False
+ROCM_AWARE_MPI = False
+
+# check whether there is CUDA- or ROCm-aware OpenMPI
+try:
+    buffer = subprocess.check_output(["ompi_info", "--parsable", "--all"])
+    CUDA_AWARE_MPI = b"mpi_built_with_cuda_support:value:true" in buffer
+    pattern = re.compile(r"^MPI extensions:.*", re.MULTILINE)
+    match = pattern.search(buffer)
+    ROCM_AWARE_MPI = "rocm" in match.group(0)
+except:  # noqa E722
+    pass
+
+# do the same for MVAPICH
+CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("MV2_USE_CUDA") == "1"
+CUDA_AWARE_MPI = ROCM_AWARE_MPI or os.environ.get("MV2_USE_ROCM") == "1"
+
+# do the same for MPICH, TODO: outdated?
+CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("MPIR_CVAR_ENABLE_HCOLL") == "1"
+
+# Cray MPICH
+CUDA_AWARE_MPI = os.environ.get("MPICH_GPU_SUPPORT_ENABLED") == "1"
+ROCM_AWARE_MPI = os.environ.get("MPICH_GPU_SUPPORT_ENABLED") == "1"
+
+# do the same for ParaStationMPI, seems to have CUDA-support only
+CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("PSP_CUDA") == "1"
+
+# Intel-MPI?
+
+# warn the user if CUDA/ROCm-aware MPI is not available, but PyTorch can use GPUs with CUDA/ROCm
+if TORCH_CUDA_IS_AVAILABLE:
+    if not CUDA_IS_ACTUALLY_ROCM and not CUDA_AWARE_MPI:
+        warnings.warn(
+            f"Heat has CUDA GPU-support (PyTorch version {TORCH_VERSION} and `torch.cuda.is_available() = True`), but CUDA-awareness of MPI could not be detected. This may lead to performance degradation as direct MPI-communication between GPUs is not possible.",
+            UserWarning,
+        )
+    elif CUDA_IS_ACTUALLY_ROCM and not ROCM_AWARE_MPI:
+        warnings.warn(
+            f"Heat has ROCm GPU-support (PyTorch version {TORCH_VERSION} and `torch.cuda.is_available() = True`), but ROCm-awareness of MPI could not be detected. This may lead to performance degradation as direct MPI-communication between GPUs is not possible.",
+            UserWarning,
+        )
+    GPU_AWARE_MPI = True
+else:
+    GPU_AWARE_MPI = False
diff --git a/heat/core/communication.py b/heat/core/communication.py
index 5cefe5041..22890f2b5 100644
--- a/heat/core/communication.py
+++ b/heat/core/communication.py
@@ -7,25 +7,14 @@
 import numpy as np
 import math
 import ctypes
-import os
-import subprocess
 import torch
+import warnings
 from mpi4py import MPI
 
 from typing import Any, Callable, Optional, List, Tuple, Union
 from .stride_tricks import sanitize_axis
 
-CUDA_AWARE_MPI = False
-# check whether OpenMPI support CUDA-aware MPI
-if "openmpi" in os.environ.get("MPI_SUFFIX", "").lower():
-    buffer = subprocess.check_output(["ompi_info", "--parsable", "--all"])
-    CUDA_AWARE_MPI = b"mpi_built_with_cuda_support:value:true" in buffer
-# MVAPICH
-CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("MV2_USE_CUDA") == "1"
-# MPICH
-CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("MPIR_CVAR_ENABLE_HCOLL") == "1"
-# ParaStationMPI
-CUDA_AWARE_MPI = CUDA_AWARE_MPI or os.environ.get("PSP_CUDA") == "1"
+from ._config import CUDA_AWARE_MPI
 
 
 class MPIRequest: