Merge branch 'main' into multi-volume

examples/old_symbolics/dagrt-fusion.py

@@ -243,7 +243,7 @@ def isolate_function_calls_in_phase(phase, stmt_id_gen, var_name_gen):
             stmt_id_gen=stmt_id_gen,
             var_name_gen=var_name_gen)
 
-    for stmt in sorted(phase.statements, key=lambda stmt: stmt.id):
+    for stmt in sorted(phase.statements, key=lambda stmt_: stmt_.id):
         new_deps = []
 
         from dagrt.language import Assign

grudge/array_context.py

@@ -88,6 +88,19 @@ class PyOpenCLArrayContext(_PyOpenCLArrayContextBase):
     to understand :mod:`grudge`-specific transform metadata. (Of which there isn't
     any, for now.)
     """
+    def __init__(self, queue: "pyopencl.CommandQueue",
+            allocator: Optional["pyopencl.tools.AllocatorInterface"] = None,
+            wait_event_queue_length: Optional[int] = None,
+            force_device_scalars: bool = False) -> None:
+
+        if allocator is None:
+            from warnings import warn
+            warn("No memory allocator specified, please pass one. "
+                 "(Preferably a pyopencl.tools.MemoryPool in order "
+                 "to reduce device allocations)")
+
+        super().__init__(queue, allocator,
+                         wait_event_queue_length, force_device_scalars)
 
 # }}}
 
@@ -99,6 +112,13 @@ class PytatoPyOpenCLArrayContext(_PytatoPyOpenCLArrayContextBase):
     Extends it to understand :mod:`grudge`-specific transform metadata. (Of
     which there isn't any, for now.)
     """
+    def __init__(self, queue, allocator=None):
+        if allocator is None:
+            from warnings import warn
+            warn("No memory allocator specified, please pass one. "
+                 "(Preferably a pyopencl.tools.MemoryPool in order "
+                 "to reduce device allocations)")
+        super().__init__(queue, allocator)
 
 # }}}
 
@@ -210,6 +230,7 @@ def __call__(self, arg_id_to_arg) -> ArrayContainer:
         out_dict = execute_distributed_partition(
                 self.distributed_partition, self.part_id_to_prg,
                 self.actx.queue, self.actx.mpi_communicator,
+                allocator=self.actx.allocator,
                 input_args=input_args_for_prg)
 
         def to_output_template(keys, _):
@@ -224,6 +245,12 @@ class MPIPytatoArrayContextBase(MPIBasedArrayContext):
     def __init__(
             self, mpi_communicator, queue, *, mpi_base_tag, allocator=None
             ) -> None:
+        if allocator is None:
+            from warnings import warn
+            warn("No memory allocator specified, please pass one. "
+                 "(Preferably a pyopencl.tools.MemoryPool in order "
+                 "to reduce device allocations)")
+
         super().__init__(queue, allocator)
 
         self.mpi_communicator = mpi_communicator

test/test_mpi_communication.py

@@ -151,22 +151,24 @@ def _test_func_comparison_mpi_communication_entrypoint(actx):
     bdry_faces_func = op.project(dcoll, BTAG_ALL, dd_af,
                                  op.project(dcoll, dd_vol, BTAG_ALL, myfunc))
 
-    hopefully_zero = (
-        op.project(
-            dcoll, "int_faces", "all_faces",
-            dcoll.opposite_face_connection(
-                dof_desc.BoundaryDomainTag(
-                    dof_desc.FACE_RESTR_INTERIOR, dof_desc.VTAG_ALL)
-                )(int_faces_func)
-        )
-        + sum(op.project(dcoll, tpair.dd, "all_faces", tpair.int)
-              for tpair in op.cross_rank_trace_pairs(dcoll, myfunc,
-                  comm_tag=SimpleTag))
-    ) - (all_faces_func - bdry_faces_func)
+    def hopefully_zero():
+        return (
+            op.project(
+                dcoll, "int_faces", "all_faces",
+                dcoll.opposite_face_connection(
+                    dof_desc.BoundaryDomainTag(
+                        dof_desc.FACE_RESTR_INTERIOR, dof_desc.VTAG_ALL)
+                    )(int_faces_func)
+            )
+            + sum(op.project(dcoll, tpair.dd, "all_faces", tpair.ext)
+                  for tpair in op.cross_rank_trace_pairs(dcoll, myfunc,
+                      comm_tag=SimpleTag))
+        ) - (all_faces_func - bdry_faces_func)
+
+    hopefully_zero_result = actx.compile(hopefully_zero)()
 
-    error = actx.to_numpy(flat_norm(hopefully_zero, ord=np.inf))
+    error = actx.to_numpy(flat_norm(hopefully_zero_result, ord=np.inf))
 
-    print(__file__)
     with np.printoptions(threshold=100000000, suppress=True):
         logger.debug(hopefully_zero)
     logger.info("error: %.5e", error)