Add additional performance optimizations to cythonized Buffer

Co-authored-by: leofang <5534781+leofang@users.noreply.github.com>

Author: Copilot

cuda_core/cuda/core/experimental/_memory.pyx

@@ -101,22 +101,25 @@ cdef class Buffer:
     @property
     def is_device_accessible(self) -> bool:
         """Return True if this buffer can be accessed by the GPU, otherwise False."""
-        if self._mnff.mr is not None:
-            return self._mnff.mr.is_device_accessible
+        cdef object mr = self._mnff.mr
+        if mr is not None:
+            return mr.is_device_accessible
         raise NotImplementedError("WIP: Currently this property only supports buffers with associated MemoryResource")
 
     @property
     def is_host_accessible(self) -> bool:
         """Return True if this buffer can be accessed by the CPU, otherwise False."""
-        if self._mnff.mr is not None:
-            return self._mnff.mr.is_host_accessible
+        cdef object mr = self._mnff.mr
+        if mr is not None:
+            return mr.is_host_accessible
         raise NotImplementedError("WIP: Currently this property only supports buffers with associated MemoryResource")
 
     @property
     def device_id(self) -> int:
         """Return the device ordinal of this buffer."""
-        if self._mnff.mr is not None:
-            return self._mnff.mr.device_id
+        cdef object mr = self._mnff.mr
+        if mr is not None:
+            return mr.device_id
         raise NotImplementedError("WIP: Currently this property only supports buffers with associated MemoryResource")
 
     def copy_to(self, dst: Buffer = None, *, stream: Stream) -> Buffer:
@@ -141,11 +144,14 @@ cdef class Buffer:
             if self._mnff.mr is None:
                 raise ValueError("a destination buffer must be provided (this buffer does not have a memory_resource)")
             dst = self._mnff.mr.allocate(self._mnff.size, stream)
-        if dst._mnff.size != self._mnff.size:
+        
+        cdef int src_size = self._mnff.size
+        cdef int dst_size = dst._mnff.size
+        if dst_size != src_size:
             raise ValueError(
-                f"buffer sizes mismatch between src and dst (sizes are: src={self._mnff.size}, dst={dst._mnff.size})"
+                f"buffer sizes mismatch between src and dst (sizes are: src={src_size}, dst={dst_size})"
             )
-        handle_return(driver.cuMemcpyAsync(dst._mnff.ptr, self._mnff.ptr, self._mnff.size, stream.handle))
+        handle_return(driver.cuMemcpyAsync(dst._mnff.ptr, self._mnff.ptr, src_size, stream.handle))
         return dst
 
     def copy_from(self, src: Buffer, *, stream: Stream):
@@ -162,11 +168,14 @@ cdef class Buffer:
         """
         if stream is None:
             raise ValueError("stream must be provided")
-        if src._mnff.size != self._mnff.size:
+        
+        cdef int src_size = src._mnff.size
+        cdef int dst_size = self._mnff.size
+        if src_size != dst_size:
             raise ValueError(
-                f"buffer sizes mismatch between src and dst (sizes are: src={src._mnff.size}, dst={self._mnff.size})"
+                f"buffer sizes mismatch between src and dst (sizes are: src={src_size}, dst={dst_size})"
             )
-        handle_return(driver.cuMemcpyAsync(self._mnff.ptr, src._mnff.ptr, self._mnff.size, stream.handle))
+        handle_return(driver.cuMemcpyAsync(self._mnff.ptr, src._mnff.ptr, src_size, stream.handle))
 
     def __dlpack__(
         self,