fix(ggml): preload bundled OpenMP runtime before loading ggml-base

- Preload the packaged libomp140.x86_64.dll on Windows before initializing
ggml-base to ensure CPU backend DLLs can resolve their OpenMP runtime
dependency.

- This only applies to Windows builds with llama-cpp-python >= 0.3.39 and
uses the bundled runtime from the package lib directory, avoiding the need
for users to configure system PATH or install additional OpenMP runtimes.

Signed-off-by: JamePeng <jame_peng@sina.com>

Author: JamePeng

llama_cpp/_ctypes_extensions.py

@@ -5,6 +5,7 @@
 import ctypes
 import functools
 import pathlib
+import importlib.metadata
 from ctypes.util import find_library
 from typing import (
     Any,
@@ -19,6 +20,15 @@
 )
 from typing_extensions import TypeAlias
 
+def _version_at_least(version: str) -> bool:
+    """Check whether installed llama-cpp-python version meets requirement."""
+    try:
+        current = importlib.metadata.version("llama-cpp-python")
+        from packaging.version import Version
+        return Version(current) >= Version(version)
+    except Exception:
+        return False
+
 def _format_library_dir_contents(base_paths: list[pathlib.Path]) -> str:
     """Format directory contents for diagnostics after library loading fails."""
     sections = []

llama_cpp/_ggml.py

@@ -6,10 +6,9 @@
 import enum
 import os
 import pathlib
-
 from llama_cpp._ctypes_extensions import (
+    _version_at_least,
     load_shared_library,
-    byref,
     ctypes_function_for_shared_library,
 )
 
@@ -21,12 +20,46 @@
     TYPE_CHECKING,
 )
 
+def _preload_openmp_runtime():
+    """Preload bundled OpenMP runtime before loading ggml-base.
+
+    This is required on Windows when CPU backends depend on the packaged
+    OpenMP runtime DLL.
+    """
+
+    # Only Windows DLL loading requires this workaround.
+    if os.name != "nt":
+        return
+
+    # Keep compatibility with older package versions.
+    if not _version_at_least("0.3.39"):
+        return
+
+    libomp_path = (pathlib.Path(__file__).parent / "lib" / "libomp140.x86_64.dll")
+
+    if not libomp_path.exists():
+        print(f"[llama-cpp-python] WARNING: bundled OpenMP runtime not found: {libomp_path}")
+        return
+
+    try:
+        ctypes.CDLL(str(libomp_path), winmode=ctypes.RTLD_GLOBAL)
+        print(f"[llama-cpp-python] loaded bundled OpenMP runtime: {libomp_path}")
+    except Exception as e:
+        print(
+            "[llama-cpp-python] WARNING: failed to load bundled OpenMP runtime:\n"
+            f"  path: {libomp_path}\n"
+            f"  error: {e}"
+        )
+
 libggml_base_path = pathlib.Path(os.path.abspath(os.path.dirname(__file__)))
 libggml_base_paths = [
     libggml_base_path / "lib",
     libggml_base_path / "bin",
 ]
 
+# Load bundled OpenMP runtime before ggml-base on Windows.
+_preload_openmp_runtime()
+
 libggml_base = load_shared_library("ggml-base", libggml_base_paths)
 
 ggml_base_function = ctypes_function_for_shared_library(libggml_base)