profile --> (profile or print_global_stats

theano/compile/function_module.py

@@ -664,7 +664,7 @@ class Function(object):
         input_storage = [i.value for i in ins]
         # reinitialize new maker and create new function
         if profile is None:
-            profile = config.profile
+            profile = config.profile or config.print_global_stats
             # profile -> True or False
         if profile is True:
             if name:

theano/compile/pfunc.py

@@ -364,7 +364,7 @@ def pfunc(params, outputs=None, mode=None, updates=None, givens=None,
     if givens is None:
         givens = []
     if profile is None:
-        profile = config.profile
+        profile = config.profile or config.print_global_stats
         # profile -> True or False
         if profile is False:
             profile = None

theano/compile/profiling.py

@@ -50,66 +50,67 @@ def _atexit_print_fn():
     Print ProfileStat objects in _atexit_print_list to _atexit_print_file.
 
     """
-    to_sum = []
+    if config.profile:
+        to_sum = []
 
-    if config.profiling.destination == 'stderr':
-        destination_file = sys.stderr
-    elif config.profiling.destination == 'stdout':
-        destination_file = sys.stdout
-    else:
-        destination_file = open(config.profiling.destination, 'w')
-
-    # Reverse sort in the order of compile+exec time
-    for ps in sorted(_atexit_print_list,
-                     key=lambda a:a.compile_time + a.fct_call_time)[::-1]:
-        if ps.fct_callcount >= 1 or ps.compile_time > 1:
-            ps.summary(file=destination_file,
-                       n_ops_to_print=config.profiling.n_ops,
-                       n_apply_to_print=config.profiling.n_apply)
-            if not isinstance(ps, ScanProfileStats):
-                to_sum.append(ps)
+        if config.profiling.destination == 'stderr':
+            destination_file = sys.stderr
+        elif config.profiling.destination == 'stdout':
+            destination_file = sys.stdout
         else:
-            # TODO print the name if there is one!
-            print('Skipping empty Profile')
-    if len(to_sum) > 1:
-        # Make a global profile
-        cum = copy.copy(to_sum[0])
-        msg = ("Sum of all(%d) printed profiles at exit excluding Scan op"
-               " profile." % len(to_sum))
-        cum.message = msg
-        for ps in to_sum[1:]:
-            for attr in ["compile_time", "fct_call_time", "fct_callcount",
-                         "vm_call_time", "optimizer_time", "linker_time",
-                         "validate_time", "import_time",
-                         "linker_node_make_thunks"]:
-                setattr(cum, attr, getattr(cum, attr) + getattr(ps, attr))
-
-            # merge dictonary
-            for attr in ["apply_time", "apply_callcount",
-                         "apply_cimpl", "variable_shape", "variable_strides",
-                         "linker_make_thunk_time"]:
-                cum_attr = getattr(cum, attr)
-                for key, val in iteritems(getattr(ps, attr)):
-                    assert key not in cum_attr
-                    cum_attr[key] = val
-
-            if cum.optimizer_profile and ps.optimizer_profile:
-                try:
-                    merge = cum.optimizer_profile[0].merge_profile(
-                        cum.optimizer_profile[1],
-                        ps.optimizer_profile[1])
-                    assert len(merge) == len(cum.optimizer_profile[1])
-                    cum.optimizer_profile = (cum.optimizer_profile[0], merge)
-                except Exception as e:
-                    print("Got an exception while merging profile")
-                    print(e)
-                    cum.optimizer_profile = None
+            destination_file = open(config.profiling.destination, 'w')
+
+        # Reverse sort in the order of compile+exec time
+        for ps in sorted(_atexit_print_list,
+                         key=lambda a:a.compile_time + a.fct_call_time)[::-1]:
+            if ps.fct_callcount >= 1 or ps.compile_time > 1:
+                ps.summary(file=destination_file,
+                           n_ops_to_print=config.profiling.n_ops,
+                           n_apply_to_print=config.profiling.n_apply)
+                if not isinstance(ps, ScanProfileStats):
+                    to_sum.append(ps)
             else:
-                cum.optimizer_profile = None
+                # TODO print the name if there is one!
+                print('Skipping empty Profile')
+        if len(to_sum) > 1:
+            # Make a global profile
+            cum = copy.copy(to_sum[0])
+            msg = ("Sum of all(%d) printed profiles at exit excluding Scan op"
+                   " profile." % len(to_sum))
+            cum.message = msg
+            for ps in to_sum[1:]:
+                for attr in ["compile_time", "fct_call_time", "fct_callcount",
+                             "vm_call_time", "optimizer_time", "linker_time",
+                             "validate_time", "import_time",
+                             "linker_node_make_thunks"]:
+                    setattr(cum, attr, getattr(cum, attr) + getattr(ps, attr))
+
+                # merge dictonary
+                for attr in ["apply_time", "apply_callcount",
+                             "apply_cimpl", "variable_shape", "variable_strides",
+                             "linker_make_thunk_time"]:
+                    cum_attr = getattr(cum, attr)
+                    for key, val in iteritems(getattr(ps, attr)):
+                        assert key not in cum_attr
+                        cum_attr[key] = val
+
+                if cum.optimizer_profile and ps.optimizer_profile:
+                    try:
+                        merge = cum.optimizer_profile[0].merge_profile(
+                            cum.optimizer_profile[1],
+                            ps.optimizer_profile[1])
+                        assert len(merge) == len(cum.optimizer_profile[1])
+                        cum.optimizer_profile = (cum.optimizer_profile[0], merge)
+                    except Exception as e:
+                        print("Got an exception while merging profile")
+                        print(e)
+                        cum.optimizer_profile = None
+                else:
+                    cum.optimizer_profile = None
 
-        cum.summary(file=destination_file,
-                    n_ops_to_print=config.profiling.n_ops,
-                    n_apply_to_print=config.profiling.n_apply)
+            cum.summary(file=destination_file,
+                        n_ops_to_print=config.profiling.n_ops,
+                        n_apply_to_print=config.profiling.n_apply)
 
     if config.print_global_stats:
         print_global_stats()

theano/gof/vm.py

@@ -482,7 +482,7 @@ class Stack(VM):
                     try:
                         _, dt = self.run_thunk_of_node(current_apply)
                         del _
-                        if config.profile:
+                        if config.profile or config.print_global_stats:
                             current_idx = self.node_idx[current_apply]
                             self.call_counts[current_idx] += 1
                             self.call_times[current_idx] += dt
@@ -596,7 +596,7 @@ class Stack(VM):
                         if current_apply.inputs[r].owner:
                             apply_stack.append(current_apply.inputs[r].owner)
                 else:
-                    if config.profile:
+                    if config.profile or config.print_global_stats:
                         for (idx, o) in enumerate(thunks[
                                 self.node_idx[current_apply]].outputs):
                             var = self.nodes[
@@ -757,7 +757,7 @@ class VM_Linker(link.LocalLinker):
         associated to self, else, a new VM_Linker associated to fgraph.
 
         """
-        if (config.profile and
+        if ((config.profile or config.print_global_stats) and
                 ((hasattr(theano, 'sandbox') and
                   hasattr(theano.sandbox, 'cuda') and
                   theano.sandbox.cuda.cuda_enabled) or
@@ -856,7 +856,7 @@ class VM_Linker(link.LocalLinker):
         pre_call_clear = [storage_map[v] for v in self.no_recycling]
 
         if (self.callback is not None or self.callback_input is not None or
-                (config.profile and config.profile_memory) or
+                ((config.profile or config.print_global_stats) and config.profile_memory) or
                 (self.allow_partial_eval and not self.use_cloop)):
 
             if self.use_cloop and (self.callback is not None or
@@ -1086,7 +1086,7 @@ class VM_Linker(link.LocalLinker):
             lazy = config.vm.lazy
         if lazy is None:
             lazy = not all([(not th.lazy) for th in thunks])
-        if not (lazy or (config.profile and config.profile_memory) or
+        if not (lazy or ((config.profile or config.print_global_stats) and config.profile_memory) or
                 self.use_cloop or self.callback or self.callback_input):
             for pair in itervalues(reallocated_info):
                 storage_map[pair[1]] = storage_map[pair[0]]