Improve buffer initialization, skip pinning for CPU device

src/calng/Gotthard2Correction.py

@@ -200,6 +200,7 @@ class Gotthard2Correction(BaseCorrection):
     _image_data_path = "data.adc"
     _cell_table_path = "data.memoryCell"
     _warn_memory_cell_range = False  # for now, receiver always writes 255
+    _cuda_pin_buffers = False
 
     @staticmethod
     def expectedParameters(expected):

src/calng/base_correction.py

@@ -197,6 +197,7 @@ class BaseCorrection(PythonDevice):
     _image_data_path = "image.data"  # customize for *some* subclasses
     _cell_table_path = "image.cellId"
     _warn_memory_cell_range = True  # can be disabled for some detectors
+    _cuda_pin_buffers = True
 
     def _load_constant_to_runner(self, constant_name, constant_data):
         """Subclass must define how to process constants into correction maps and store
@@ -894,8 +895,9 @@ class BaseCorrection(PythonDevice):
                 self.output_data_dtype,
                 shmem_buffer_name,
             )
-            self.log.INFO("Trying to pin the shmem buffer memory")
-            self._shmem_buffer.cuda_pin()
+            if self._cuda_pin_buffers:
+                self.log.INFO("Trying to pin the shmem buffer memory")
+                self._shmem_buffer.cuda_pin()
             self.log.INFO("Done, shmem buffer is ready")
         else:
             self._shmem_buffer.change_shape(self.output_data_shape)

src/calng/shmem_utils.py

@@ -67,8 +67,6 @@ class ShmemCircularBuffer:
         self._buffer_ary = None
         self._update_shape(array_shape, dtype)
         self._cuda_pinned = False
-        # important for performance and pinning: touch memory to actually allocate
-        self._buffer_ary.fill(0)
 
     def _update_shape(self, array_shape, dtype):
         array_shape = tuple(array_shape)
@@ -101,16 +99,16 @@ class ShmemCircularBuffer:
 
     def cuda_pin(self):
         import cupy
+
         self._memory_pointer = self._buffer_ary.ctypes.get_data()
         cupy.cuda.runtime.hostRegister(
-            self._memory_pointer,
-            self._shared_memory.size,
-            0
+            self._memory_pointer, self._shared_memory.size, 0
         )
 
     def __del__(self):
         if self._cuda_pinned:
             import cupy
+
             cupy.cuda.runtime.hostUnregister(self._memory_pointer)
         del self._buffer_ary
         del self._shared_memory