discopop 3.1.0__tar.gz → 3.2.0__tar.gz

{discopop-3.1.0/discopop.egg-info → discopop-3.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: discopop
-Version: 3.1.0
+Version: 3.2.0
 Summary: DiscoPoP is a tool that helps software developers parallelize their programs with threads. It discovers potential parallelism in a sequential program and makes recommendations on how to exploit it.
 Home-page: https://www.discopop.tu-darmstadt.de/
 Author: TU Darmstadt and Iowa State University

{discopop-3.1.0 → discopop-3.2.0}/README.md

@@ -12,15 +12,15 @@
  -->
 
 # DiscoPoP - Discovery of Potential Parallelism
-DiscoPoP is an open-source tool that helps software developers parallelize their programs with threads. It is a joint project of the [Laboratory for Parallel Programming @ TU Darmstadt](https://github.com/tuda-parallel) and the Iowa State University.
+DiscoPoP is an open-source tool that helps software developers parallelize their programs with threads. It is a joint project of the [Laboratory for Parallel Programming @ TU Darmstadt](https://github.com/tuda-parallel) and the [Software Analytics and Pervasive Parallelism Lab](https://www.cs.iastate.edu/swapp/) at Iowa State University.
 
 In a nutshell, DiscoPoP performs the following steps:
 * detect parts of the code (computational units or CUs) with little to no internal parallelization potential,
 * find data dependences among them,
 * identify parallel patterns that can be used to parallelize a code region,
-* and finally suggest corresponding OpenMP parallelization constructs and clauses to programmers.
+* and finally, suggest corresponding OpenMP parallelization constructs and clauses to programmers.
 
-DiscoPoP is built on top of LLVM. Therefore, DiscoPoP can perform the above-mentioned steps on any source code which can be transferred into the LLVM IR.
+DiscoPoP is built on top of LLVM. Therefore, DiscoPoP can perform the above-mentioned steps on any source code that can be transferred into the LLVM IR.
 
 A more comprehensive overview of DiscoPoP can be found on our [project website](https://www.discopop.tu-darmstadt.de/).
 
@@ -28,13 +28,13 @@ A more comprehensive overview of DiscoPoP can be found on our [project website](
 Follow the steps in [setup](https://discopop-project.github.io/discopop/setup/discopop/) to install DiscoPoP.
 To setup the [Visual Studio Code Extension](https://marketplace.visualstudio.com/items?itemName=TUDarmstadt-LaboratoryforParallelProgramming.discopop) (recommended for general use of the framework), please follow [these steps](https://discopop-project.github.io/discopop/setup/vscx/).
 
-For a brief introduction into the [VSCode Extension](https://marketplace.visualstudio.com/items?itemName=TUDarmstadt-LaboratoryforParallelProgramming.discopop), please follow the [walk-through example](https://discopop-project.github.io/discopop/examples/walk_through_gui/).
-For a brief introduction into the command line tools, please refer to the [tools overview](https://discopop-project.github.io/discopop/Tools) and follow the [command-line walk-through example](https://discopop-project.github.io/discopop/examples/walk_through/).
+For a brief introduction to the [VSCode Extension](https://marketplace.visualstudio.com/items?itemName=TUDarmstadt-LaboratoryforParallelProgramming.discopop), please follow the [walk-through example](https://discopop-project.github.io/discopop/examples/walk_through_gui/).
+For a brief introduction to the command line tools, please refer to the [tools overview](https://discopop-project.github.io/discopop/Tools) and follow the [command-line walk-through example](https://discopop-project.github.io/discopop/examples/walk_through/).
 
 For detailed information on the gathered and stored data as well as the tools themselves, please refer to [data](https://discopop-project.github.io/discopop/Data) and the pages of the individual tools in the [tools overview](https://discopop-project.github.io/discopop/Tools).
 
 ## TL;DR
-This example installs DiscoPoP, instruments and builds the provided example, analyzes the results and prints the identified parallelization suggestions to the console.
+This example installs DiscoPoP, instruments and builds the provided example, analyzes the results, and prints the identified parallelization suggestions to the console.
 In case any issues arise during the process, please refer to the detailed [setup instructions](https://discopop-project.github.io/discopop/Setup), contact us via GitHub messages, or get in contact by mail to [discopop-support@lists.parallel.informatik.tu-darmstadt.de](mailto:discopop-support@lists.parallel.informatik.tu-darmstadt.de).
 ```
 # setup DiscoPoP
@@ -42,7 +42,7 @@ git clone git@github.com:discopop-project/discopop.git
 cd discopop
 mkdir build && cd build
 DP_BUILD=$(pwd)
-cmake .. && make 
+cmake .. && make
 # instrument and build the example code
 cd ../example
 mkdir build && cd build && cmake -DCMAKE_CXX_COMPILER=${DP_BUILD}/scripts/CXX_wrapper.sh .. && make
@@ -56,7 +56,7 @@ discopop_patch_generator
 # print patches to the console
 for f in $(find patch_generator -maxdepth 1 -type d); do
     echo "SUGGESTION: $f"
-    cat $f/1.patch 
+    cat $f/1.patch
     echo ""
 done
 # apply patch with id 1
@@ -67,19 +67,19 @@ discopop_patch_applicator -C
 
 
 ## Exemplary output
-The following is an automatically generated, exemplary output patch file generated and applicable as show in the provided [examples](https://discopop-project.github.io/discopop/Examples).
+The following is an automatically generated, exemplary output patch file generated and applicable as shown in the provided [examples](https://discopop-project.github.io/discopop/Examples).
 ```
  --- /home/lukas/temp/discopop_tmp/discopop/example/example.cpp	2024-01-09 10:11:50.369555235 +0100
 +++ /home/lukas/temp/discopop_tmp/discopop/example/example.cpp.discopop_patch_generator.temp	2024-01-09 11:14:20.904823624 +0100
 @@ -20,6 +20,7 @@
          Arr[i] = i % 13;
      }
- 
-+    #pragma omp parallel for shared(Arr,N) reduction(+:sum) 
+
++    #pragma omp parallel for shared(Arr,N) reduction(+:sum)
      for(int i = 0; i < N; i++){
          sum += Arr[i];
      }
 ```
 
 ## License
-© DiscoPoP is available under the terms of the BSD-3-Clause license, as specified in the LICENSE file.
+© DiscoPoP is available under the terms of the BSD-3-Clause license, as specified in the LICENSE file.

{discopop-3.1.0 → discopop-3.2.0/discopop.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: discopop
-Version: 3.1.0
+Version: 3.2.0
 Summary: DiscoPoP is a tool that helps software developers parallelize their programs with threads. It discovers potential parallelism in a sequential program and makes recommendations on how to exploit it.
 Home-page: https://www.discopop.tu-darmstadt.de/
 Author: TU Darmstadt and Iowa State University

{discopop-3.1.0 → discopop-3.2.0}/discopop.egg-info/SOURCES.txt

@@ -104,6 +104,7 @@ discopop_explorer/plugins/pipeline.py
 discopop_explorer/test/__init__.py
 discopop_library/__init__.py
 discopop_library/py.typed
+discopop_library/ArgumentClasses/GeneralArguments.py
 discopop_library/CodeGenerator/CodeGenerator.py
 discopop_library/CodeGenerator/CodeGeneratorArguments.py
 discopop_library/CodeGenerator/__init__.py
@@ -117,6 +118,11 @@ discopop_library/ConfigProvider/ConfigProviderArguments.py
 discopop_library/ConfigProvider/__main__.py
 discopop_library/ConfigProvider/config_provider.py
 discopop_library/ConfigProvider/assets/.gitignore
+discopop_library/GlobalLogger/setup.py
+discopop_library/HostpotLoader/HotspotLoaderArguments.py
+discopop_library/HostpotLoader/HotspotNodeType.py
+discopop_library/HostpotLoader/HotspotType.py
+discopop_library/HostpotLoader/hostpot_loader.py
 discopop_library/JSONHandler/JSONHandler.py
 discopop_library/JSONHandler/__init__.py
 discopop_library/LineMapping/__init__.py
@@ -157,7 +163,12 @@ discopop_library/discopop_optimizer/CostModels/utilities.py
 discopop_library/discopop_optimizer/CostModels/DataTransfer/DataTransferCosts.py
 discopop_library/discopop_optimizer/CostModels/DataTransfer/__init__.py
 discopop_library/discopop_optimizer/DataTransfers/DataTransfers.py
+discopop_library/discopop_optimizer/DataTransfers/NewDataTransfers.py
 discopop_library/discopop_optimizer/DataTransfers/__init__.py
+discopop_library/discopop_optimizer/DataTransfers/calculate_configuration_data_movement.py
+discopop_library/discopop_optimizer/DataTransfers/prepare_force_branch_end_data_movement.py
+discopop_library/discopop_optimizer/DataTransfers/prepare_forced_data_movement.py
+discopop_library/discopop_optimizer/DataTransfers/prepare_forced_data_movement_prior_to_call.py
 discopop_library/discopop_optimizer/Microbench/DelaunayInterpolatedMicrobench.py
 discopop_library/discopop_optimizer/Microbench/ExtrapInterpolatedMicrobench.py
 discopop_library/discopop_optimizer/Microbench/Microbench.py
@@ -171,9 +182,11 @@ discopop_library/discopop_optimizer/PETParser/__init__.py
 discopop_library/discopop_optimizer/PETParser/DataAccesses/CalculateUpdates.py
 discopop_library/discopop_optimizer/PETParser/DataAccesses/FromCUs.py
 discopop_library/discopop_optimizer/PETParser/DataAccesses/__init__.py
+discopop_library/discopop_optimizer/UpdateOptimization/AddRangesToUpdates.py
 discopop_library/discopop_optimizer/UpdateOptimization/LoopInitializationUpdates.py
 discopop_library/discopop_optimizer/UpdateOptimization/RemoveDuplicatedUpdates.py
 discopop_library/discopop_optimizer/UpdateOptimization/RemoveLoopIndexUpdates.py
+discopop_library/discopop_optimizer/UpdateOptimization/RemoveSameDeviceUpdates.py
 discopop_library/discopop_optimizer/UpdateOptimization/main.py
 discopop_library/discopop_optimizer/Variables/Experiment.py
 discopop_library/discopop_optimizer/Variables/ExperimentUtils.py
@@ -185,6 +198,7 @@ discopop_library/discopop_optimizer/classes/context/ContextObject.py
 discopop_library/discopop_optimizer/classes/context/ContextObjectUtils.py
 discopop_library/discopop_optimizer/classes/context/Update.py
 discopop_library/discopop_optimizer/classes/context/__init__.py
+discopop_library/discopop_optimizer/classes/edges/CallEdge.py
 discopop_library/discopop_optimizer/classes/edges/ChildEdge.py
 discopop_library/discopop_optimizer/classes/edges/DataFlowEdge.py
 discopop_library/discopop_optimizer/classes/edges/GenericEdge.py
@@ -202,9 +216,12 @@ discopop_library/discopop_optimizer/classes/nodes/ContextRestore.py
 discopop_library/discopop_optimizer/classes/nodes/ContextSave.py
 discopop_library/discopop_optimizer/classes/nodes/ContextSnapshot.py
 discopop_library/discopop_optimizer/classes/nodes/ContextSnapshotPop.py
+discopop_library/discopop_optimizer/classes/nodes/DeviceSwitch.py
+discopop_library/discopop_optimizer/classes/nodes/FunctionReturn.py
 discopop_library/discopop_optimizer/classes/nodes/FunctionRoot.py
 discopop_library/discopop_optimizer/classes/nodes/GenericNode.py
 discopop_library/discopop_optimizer/classes/nodes/Loop.py
+discopop_library/discopop_optimizer/classes/nodes/SynchronizationTrigger.py
 discopop_library/discopop_optimizer/classes/nodes/Workload.py
 discopop_library/discopop_optimizer/classes/nodes/__init__.py
 discopop_library/discopop_optimizer/classes/system/Network.py
@@ -229,9 +246,11 @@ discopop_library/discopop_optimizer/gui/queries/ValueTableQuery.py
 discopop_library/discopop_optimizer/gui/queries/__init__.py
 discopop_library/discopop_optimizer/gui/widgets/ScrollableFrame.py
 discopop_library/discopop_optimizer/gui/widgets/__init__.py
+discopop_library/discopop_optimizer/interactive/interactive_optimizer.py
 discopop_library/discopop_optimizer/optimization/evaluate.py
 discopop_library/discopop_optimizer/optimization/evaluate_all_decision_combinations.py
 discopop_library/discopop_optimizer/optimization/evolutionary_algorithm.py
+discopop_library/discopop_optimizer/optimization/greedy.py
 discopop_library/discopop_optimizer/optimization/validation.py
 discopop_library/discopop_optimizer/scheduling/__init__.py
 discopop_library/discopop_optimizer/scheduling/workload_delta.py
@@ -244,6 +263,7 @@ discopop_library/discopop_optimizer/suggestions/optimizers/loop_collapse.py
 discopop_library/discopop_optimizer/suggestions/optimizers/main.py
 discopop_library/discopop_optimizer/utilities/MOGUtilities.py
 discopop_library/discopop_optimizer/utilities/__init__.py
+discopop_library/discopop_optimizer/utilities/insert_device_switch_nodes.py
 discopop_library/discopop_optimizer/utilities/simple_utilities.py
 discopop_library/discopop_optimizer/utilities/optimization/__init__.py
 discopop_library/discopop_optimizer/utilities/optimization/GlobalOptimization/RandomSamples.py
@@ -261,20 +281,99 @@ discopop_library/global_data/version/utils.py
 discopop_library/global_data/version/__pycache__/__init__.cpython-38.pyc
 discopop_library/global_data/version/__pycache__/utils.cpython-38.pyc
 discopop_library/result_classes/DetectionResult.py
+discopop_library/result_classes/MergedPattern.py
 discopop_library/result_classes/OptimizerOutputPattern.py
+discopop_library/result_classes/PatternStorage.py
 discopop_library/result_classes/__init__.py
 test/__init__.py
 test/do_all/__init__.py
+test/do_all/backwards_array_access/__init__.py
+test/do_all/backwards_array_access/test.py
+test/do_all/calls/__init__.py
+test/do_all/calls/allowing/__init__.py
+test/do_all/calls/allowing/test.py
+test/do_all/calls/preventing/__init__.py
+test/do_all/calls/preventing/simple/__init__.py
+test/do_all/calls/preventing/simple/test.py
+test/do_all/calls/second_order/__init__.py
+test/do_all/calls/second_order/allowing/__init__.py
+test/do_all/calls/second_order/allowing/test.py
+test/do_all/calls/second_order/allowing_2/__init__.py
+test/do_all/calls/second_order/allowing_2/test.py
+test/do_all/calls/second_order/preventing/__init__.py
+test/do_all/calls/second_order/preventing/test.py
+test/do_all/calls/second_order/preventing_2/__init__.py
+test/do_all/calls/second_order/preventing_2/test.py
 test/do_all/daxpy/__init__.py
 test/do_all/daxpy/test.py
+test/do_all/negative/__init__.py
+test/do_all/negative/nested/__init__.py
+test/do_all/negative/nested/OMPSCR/__init__.py
+test/do_all/simple/__init__.py
+test/do_all/simple/global_vars/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_arrays/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_arrays/test.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_struct/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_struct/test.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_vectors/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_vectors/test.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_vectors_and_struct/__init__.py
+test/do_all/simple/global_vars/miniFE_CSRMatrix_proxy/global_vectors_and_struct/test.py
+test/do_all/simple/nested/__init__.py
+test/do_all/simple/nested/depth_2/__init__.py
+test/do_all/simple/nested/depth_2/positive/__init__.py
+test/do_all/simple/nested/depth_2/positive/both_loops_doall/__init__.py
+test/do_all/simple/nested/depth_2/positive/both_loops_doall/test.py
+test/do_all/simple/nested/depth_2/positive/inner_loop_doall/__init__.py
+test/do_all/simple/nested/depth_2/positive/inner_loop_doall/test.py
+test/do_all/simple/nested/depth_2/positive/outer_loop_doall/__init__.py
+test/do_all/simple/nested/depth_2/positive/outer_loop_doall/test.py
+test/do_all/simple/not_nested/__init__.py
+test/do_all/simple/not_nested/negative/__init__.py
+test/do_all/simple/not_nested/negative/minimal/__init__.py
+test/do_all/simple/not_nested/negative/minimal/test.py
+test/do_all/simple/not_nested/negative/stack_access/__init__.py
+test/do_all/simple/not_nested/negative/stack_access/test.py
+test/do_all/simple/not_nested/positive/__init__.py
+test/do_all/simple/not_nested/positive/test.py
+test/do_all/stack_access/__init__.py
+test/do_all/stack_access/nested/__init__.py
+test/do_all/stack_access/nested/positive/__init__.py
+test/do_all/stack_access/nested/positive/test.py
+test/do_all/stack_access/various/__init__.py
+test/do_all/stack_access/various/case_0/__init__.py
+test/do_all/stack_access/various/case_0/test.py
+test/do_all/stack_access/various/case_1/__init__.py
+test/do_all/stack_access/various/case_1/test.py
+test/do_all/stack_access/various/case_2/__init__.py
+test/do_all/stack_access/various/case_2/test.py
+test/do_all/stack_access/various/case_3/__init__.py
+test/do_all/stack_access/various/case_3/test.py
+test/do_all/stack_access/various/case_4/__init__.py
+test/do_all/stack_access/various/case_4/test.py
+test/do_all/stack_access/various/case_5/__init__.py
+test/do_all/stack_access/various/case_5/test.py
+test/do_all/struct/__init__.py
+test/do_all/struct/positive/__init__.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/__init__.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/access_outside_struct/__init__.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/access_outside_struct/test.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/std_data_types/__init__.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/std_data_types/test.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/template_data_types/__init__.py
+test/do_all/struct/positive/miniFE_CSRMatrix_proxy/template_data_types/test.py
 test/optimizer/__init__.py
 test/optimizer/loop_collapse/__init__.py
-test/optimizer/loop_collapse/negative/__init__.py
-test/optimizer/loop_collapse/negative/simple_1/__init__.py
-test/optimizer/loop_collapse/negative/simple_1/test.py
 test/optimizer/loop_collapse/positive/__init__.py
 test/optimizer/loop_collapse/positive/simple_1/__init__.py
 test/optimizer/loop_collapse/positive/simple_1/test.py
+test/reduction_pattern/__init__.py
+test/reduction_pattern/positive/__init__.py
+test/reduction_pattern/positive/nested/__init__.py
+test/reduction_pattern/positive/nested/OMPSCR/__init__.py
+test/reduction_pattern/positive/nested/OMPSCR/c_Mandelbrot/__init__.py
+test/reduction_pattern/positive/nested/OMPSCR/c_Mandelbrot/test.py
 test/utils/__init__.py
 test/utils/validator_classes/DoAllInfoForValidation.py
 test/utils/validator_classes/__init__.py

{discopop-3.1.0 → discopop-3.2.0}/discopop_explorer/PEGraphX.py

@@ -17,7 +17,9 @@ import jsonpickle  # type:ignore
 import matplotlib.pyplot as plt  # type:ignore
 import networkx as nx  # type:ignore
 from alive_progress import alive_bar  # type: ignore
-from lxml.objectify import ObjectifiedElement  # type:ignore
+from lxml.objectify import ObjectifiedElement  # type: ignore
+
+from discopop_library.HostpotLoader.HotspotNodeType import HotspotNodeType  # type:ignore
 
 from .parser import LoopData, readlineToCUIdMap, writelineToCUIdMap, DependenceItem
 from .variable import Variable
@@ -158,9 +160,9 @@ class Node:
     end_line: int
     type: NodeType
     name: str
-    parent_function_id: Optional[
-        NodeID
-    ] = None  # metadata to speedup some calculations (TODO FunctionNodes have themselves as parent)
+    parent_function_id: Optional[NodeID] = (
+        None  # metadata to speedup some calculations (TODO FunctionNodes have themselves as parent)
+    )
     workload: Optional[int] = None
 
     # properties of CU Nodes
@@ -693,8 +695,6 @@ class PEGraphX(object):
             # as a result, comparing variable names to match memory regions is valid
             for _, _, d1 in out_deps:
                 for _, _, d2 in out_deps:
-                    if d1 == d2:
-                        continue
                     if d1.var_name == d2.var_name:
                         if d1.memory_region != d2.memory_region:
                             if d1.memory_region not in mem_reg_mappings:
@@ -724,10 +724,26 @@ class PEGraphX(object):
 
         print("Done.")
 
-    def calculateFunctionMetadata(self) -> None:
+    def calculateFunctionMetadata(self, hotspot_information=None, func_nodes=None) -> None:
         # store id of parent function in each node
         # and store in each function node a list of all children ids
-        func_nodes = self.all_nodes(FunctionNode)
+        if func_nodes is None:
+            func_nodes = self.all_nodes(FunctionNode)
+
+            if hotspot_information is not None:
+                all_hotspot_functions: Set[Tuple[int, str]] = set()
+                for key in hotspot_information:
+                    for entry in hotspot_information[key]:
+                        if entry[2] == HotspotNodeType.FUNCTION:
+                            all_hotspot_functions.add((entry[0], entry[3]))
+
+                filtered_func_nodes = [
+                    func_node
+                    for func_node in func_nodes
+                    if (func_node.file_id, func_node.name) in all_hotspot_functions
+                ]
+                func_nodes = filtered_func_nodes
+
         print("Calculating local metadata results for functions...")
         import tqdm  # type: ignore
         from multiprocessing import Pool
@@ -752,37 +768,39 @@ class PEGraphX(object):
         print("\tMetadata calculation done.")
 
         # cleanup dependencies (remove dependencies, if it is overwritten by a more specific Intra-iteration dependency
-        print("Cleaning duplicated dependencies...")
-        to_be_removed = []
-        for cu_node in self.all_nodes(CUNode):
-            out_deps = self.out_edges(cu_node.id, EdgeType.DATA)
-            for dep_1 in out_deps:
-                for dep_2 in out_deps:
-                    if dep_1 == dep_2:
-                        continue
-                    if (
-                        dep_1[2].dtype == dep_2[2].dtype
-                        and dep_1[2].etype == dep_2[2].etype
-                        and dep_1[2].memory_region == dep_2[2].memory_region
-                        and dep_1[2].sink_line == dep_2[2].sink_line
-                        and dep_1[2].source_line == dep_2[2].source_line
-                        and dep_1[2].var_name == dep_2[2].var_name
-                    ):
-                        if not dep_1[2].intra_iteration and dep_2[2].intra_iteration:
-                            # dep_2 is a more specific duplicate of dep_1
-                            # remove dep_1
-                            to_be_removed.append(dep_1)
-
-        to_be_removed_with_keys = []
-        for dep in to_be_removed:
-            graph_edges = self.g.out_edges(dep[0], keys=True, data="data")
-
-            for s, t, key, data in graph_edges:
-                if dep[0] == s and dep[1] == t and dep[2] == data:
-                    to_be_removed_with_keys.append((s, t, key))
-        for edge in set(to_be_removed_with_keys):
-            self.g.remove_edge(edge[0], edge[1], edge[2])
-        print("Cleaning dependencies done.")
+        # note: this can introduce false positives! Keep the analysis pessimistic to ensure correctness
+        if False:
+            print("Cleaning duplicated dependencies...")
+            to_be_removed = []
+            for cu_node in self.all_nodes(CUNode):
+                out_deps = self.out_edges(cu_node.id, EdgeType.DATA)
+                for dep_1 in out_deps:
+                    for dep_2 in out_deps:
+                        if dep_1 == dep_2:
+                            continue
+                        if (
+                            dep_1[2].dtype == dep_2[2].dtype
+                            and dep_1[2].etype == dep_2[2].etype
+                            and dep_1[2].memory_region == dep_2[2].memory_region
+                            and dep_1[2].sink_line == dep_2[2].sink_line
+                            and dep_1[2].source_line == dep_2[2].source_line
+                            and dep_1[2].var_name == dep_2[2].var_name
+                        ):
+                            if not dep_1[2].intra_iteration and dep_2[2].intra_iteration:
+                                # dep_2 is a more specific duplicate of dep_1
+                                # remove dep_1
+                                to_be_removed.append(dep_1)
+
+            to_be_removed_with_keys = []
+            for dep in to_be_removed:
+                graph_edges = self.g.out_edges(dep[0], keys=True, data="data")
+
+                for s, t, key, data in graph_edges:
+                    if dep[0] == s and dep[1] == t and dep[2] == data:
+                        to_be_removed_with_keys.append((s, t, key))
+            for edge in set(to_be_removed_with_keys):
+                self.g.remove_edge(edge[0], edge[1], edge[2])
+            print("Cleaning dependencies done.")
 
         # cleanup dependencies II : only consider the Intra-iteration dependencies with the highest level
         print("Cleaning duplicated dependencies II...")
@@ -941,12 +959,10 @@ class PEGraphX(object):
     NodeT = TypeVar("NodeT", bound=Node)
 
     @overload
-    def all_nodes(self) -> List[Node]:
-        ...
+    def all_nodes(self) -> List[Node]: ...
 
     @overload
-    def all_nodes(self, type: Union[Type[NodeT], Tuple[Type[NodeT], ...]]) -> List[NodeT]:
-        ...
+    def all_nodes(self, type: Union[Type[NodeT], Tuple[Type[NodeT], ...]]) -> List[NodeT]: ...
 
     def all_nodes(self, type=Node):
         """List of all nodes of specified type
@@ -989,12 +1005,10 @@ class PEGraphX(object):
             return [t for t in self.g.in_edges(node_id, data="data") if t[2].etype == etype]
 
     @overload
-    def subtree_of_type(self, root: Node) -> List[Node]:
-        ...
+    def subtree_of_type(self, root: Node) -> List[Node]: ...
 
     @overload
-    def subtree_of_type(self, root: Node, type: Union[Type[NodeT], Tuple[Type[NodeT], ...]]) -> List[NodeT]:
-        ...
+    def subtree_of_type(self, root: Node, type: Union[Type[NodeT], Tuple[Type[NodeT], ...]]) -> List[NodeT]: ...
 
     def subtree_of_type(self, root, type=Node):
         """Gets all nodes in subtree of specified type including root
@@ -1006,14 +1020,12 @@ class PEGraphX(object):
         return self.subtree_of_type_rec(root, set(), type)
 
     @overload
-    def subtree_of_type_rec(self, root: Node, visited: Set[Node]) -> List[Node]:
-        ...
+    def subtree_of_type_rec(self, root: Node, visited: Set[Node]) -> List[Node]: ...
 
     @overload
     def subtree_of_type_rec(
         self, root: Node, visited: Set[Node], type: Union[Type[NodeT], Tuple[Type[NodeT], ...]]
-    ) -> List[NodeT]:
-        ...
+    ) -> List[NodeT]: ...
 
     def subtree_of_type_rec(self, root, visited, type=Node):
         """recursive helper function for subtree_of_type"""
@@ -1400,6 +1412,21 @@ class PEGraphX(object):
         """
         if isinstance(node, FunctionNode):
             return node
+        if node.parent_function_id is None:
+            # no precalculated information found.
+            current_node = node
+            parent_node: Optional[Node] = node
+            while parent_node is not None:
+                current_node = parent_node
+                if type(self.node_at(current_node.id)) == FunctionNode:
+                    node.parent_function_id = current_node.id
+                    break
+                parents = [e[0] for e in self.in_edges(current_node.id, etype=EdgeType.CHILD)]
+                if len(parents) == 0:
+                    parent_node = None
+                else:
+                    parent_node = self.node_at(parents[0])
+
         assert node.parent_function_id
         return cast(FunctionNode, self.node_at(node.parent_function_id))
 

{discopop-3.1.0 → discopop-3.2.0}/discopop_explorer/__main__.py

@@ -8,6 +8,7 @@
 
 from argparse import ArgumentParser
 from pathlib import Path
+from discopop_library.GlobalLogger.setup import setup_logger
 
 from discopop_library.PathManagement.PathManagement import get_path, get_path_or_none
 from .discopop_explorer import ExplorerArguments, run
@@ -71,6 +72,9 @@ def parse_args() -> ExplorerArguments:
         "--enable-patterns", type=str, nargs="?", default="*",
         help="Specify comma-separated list of pattern types to be identified. Options: reduction,doall,pipeline,geodec,simplegpu. Default: *",
     )
+    parser.add_argument("--load-existing-doall-and-reduction-patterns", action="store_true", help="Skip pattern detection and insert existing patterns.json contents into the created detection_result.json")
+    parser.add_argument("--log", type=str, default="WARNING", help="Specify log level: DEBUG, INFO, WARNING, ERROR, CRITICAL")
+    parser.add_argument("--write-log", action="store_true", help="Create Logfile.")
 
     # EXPERIMENTAL FLAGS:
     # temporary flag for microbenchmark file
@@ -149,11 +153,15 @@ def parse_args() -> ExplorerArguments:
         cu_inst_result_file=arguments.cu_inst_res,
         llvm_cxxfilt_path=arguments.llvm_cxxfilt_path,
         enable_json_file=arguments.json,
+        log_level=arguments.log.upper(),
+        write_log=arguments.write_log,
+        load_existing_doall_and_reduction_patterns=arguments.load_existing_doall_and_reduction_patterns,
     )
 
 
 def main():
     arguments = parse_args()
+    setup_logger(arguments)
     run(arguments)