Clarify that num_threads does not set the OpenCL work-group size.

Author: lohedges

src/loch/_platforms/__init__.py

@@ -112,7 +112,8 @@ def create_backend(
         Total number of atoms in the system.
 
     num_threads : int
-        Number of threads per block/work-group.
+        Number of threads per block (CUDA) or granularity of the global
+        work size (OpenCL, where the runtime picks the work-group size).
 
     nvcc : str, optional
         Path to NVCC compiler (CUDA only).

src/loch/_platforms/_base.py

@@ -72,7 +72,8 @@ def __init__(
             Total number of atoms in the system.
 
         num_threads : int
-            Number of threads per block (CUDA) or work-group size (OpenCL).
+            Number of threads per block (CUDA) or granularity of the global
+            work size (OpenCL, where the runtime picks the work-group size).
 
         nvcc : str, optional
             Path to NVCC compiler (CUDA only). If None, uses default.

src/loch/_platforms/_opencl.py

@@ -81,7 +81,8 @@ def __init__(
             Total number of atoms in the system.
 
         num_threads : int
-            Work-group size (threads per work-group).
+            Granularity of the global work size. The work-group size is
+            chosen by the OpenCL runtime.
 
         nvcc : str, optional
             Ignored for OpenCL (included for API compatibility).