added multiprocessing to classifier

README.md

@@ -86,7 +86,7 @@ of representative results. _The subset execution need not be undertaken
 if the full run is performed;_ it is provided as a convenience.
 Hyperparameters for FPDiff executions may be adjusted by editing `header.py`.
 
-### [1.1] Running FPDiff on all discoverable functions (approx. 1.5 hours)
+### [1.1] Running FPDiff on all discoverable functions (approx. 1 hour)
 ```
 $ nohup ./run.sh
 ```
@@ -102,7 +102,7 @@ equivalence classes, and finally perform differential testing to
 discover numerical discrepancies.
 
 
-### [1.2] Running FPDiff on a subset of representative functions (approx. 15 minutes)
+### [1.2] Running FPDiff on a subset of representative functions (approx. 8 minutes)
 ```
 $ nohup ./run.sh subset
 ```

frozenState/subset_testing_files/runTest.sh

@@ -1,5 +1,10 @@
 cd /usr/local/src/fp-diff-testing/workspace
 
+echo ""
+echo "** Loading Subset of Function Signatures"
+
+echo ""
+echo "** Running Driver Generator"
 python3 driverGenerator.py mpmath python
 python3 driverGenerator.py scipy python
 python3 driverGenerator.py gsl c

workspace/classify.py

@@ -112,7 +112,7 @@ def runDriver_parallelly(elementaryInput, driver, key):
 
             # printing progress bar
             sys.stdout.write('\r')
-            sys.stdout.write("[%-100s] %d%%" % ('='*int(i/lengthInput*100), int(i/lengthInput*100)))
+            sys.stdout.write("[%-100s] %d%%" % ('#'*int(i/lengthInput*100), int(i/lengthInput*100)))
             sys.stdout.flush()
             i+=1
 
@@ -345,18 +345,17 @@ def getStats(CLASSES):
     # gather some statistics, save the information
     with open("logs/statistics.txt", 'a') as f:
         f.write("\nTOTAL # OF CLASSES: {}\n".format(allClassTally))
-        f.write("# OF NON-TRIVIAL CLASSES: {}\n".format(nontrivialClassTally))
-
+
 
 def prettyPrintClasses(CLASSES):
-    print("==================================================")
+    print("--------------------------------------------")
 
     for counter, classKey in enumerate(CLASSES.keys()):
         print("Class {}:\n".format(counter))
         for driver in CLASSES[classKey]:
-            print("{:<35} => {}".format(driver.get_driverName(), driver.get_id()))
+            print("{}".format(driver.get_driverName()))
         print()
-        print("==================================================")
+        print("--------------------------------------------")
 
 
 if __name__ == "__main__":
@@ -373,7 +372,7 @@ def prettyPrintClasses(CLASSES):
 
     CLASSES = {}
 
-    print("Running classifier: ")
+    print("\n** Running Classifier: ")
     classify(CLASSES, DRIVER_LIST_MANAGER, ELEMENTARY_INPUTS)
     DRIVER_LIST = dict(DRIVER_LIST_MANAGER)
 

workspace/diffTester.py

@@ -17,15 +17,16 @@
 WD = os.path.dirname(os.path.abspath(__file__))
 os.chdir(WD)
 
+# diffTester ID for progress bar
+diffTesterID = multiprocessing.Value("i", 1)
+numDiffTester = 0
 
 '''
 takes in an equivalence class, gathers all test inputs to be used 
 with that equivalence class, and creates/saves discrepancy objects
 '''
 def diffTester(classNo, classKey, totalClassCount, equivalenceClass, TEST_INPUTS, UNIQUE_DISCREPANCIES, ALL_DISCREPANCIES, USED_INPUTS):
 
-    print("Testing Class {}/{}".format(classNo, totalClassCount - 1))
-
     # grab all inputs tagged with "all_"
     inputNames = [x for x in TEST_INPUTS.keys() if "all_" in x]
 
@@ -68,6 +69,12 @@ def diffTester(classNo, classKey, totalClassCount, equivalenceClass, TEST_INPUTS
                     UNIQUE_DISCREPANCIES[x.get_id()] = x                    
 
                 ALL_DISCREPANCIES.append(x)
+
+    # printing progress bar
+    sys.stdout.write('\r')
+    sys.stdout.write("[%-100s] %d%%" % ('#'*int(diffTesterID.value/numDiffTester*100), int(diffTesterID.value/numDiffTester*100)))
+    sys.stdout.flush()
+    diffTesterID.value += 1
 
 
 ''' function to write out a csv file of the results '''
@@ -163,11 +170,8 @@ def getStats(UNIQUE_DISCREPANCIES, ALL_DISCREPANCIES, USED_INPUTS):
         tempz.reverse()
 
         f.write("\nspecialValue_unique_discrepancyTally: {}\n".format(tempx))
-        #f.write("\t# of Unique Discrepancies/# of Evals (# of Unique Inputs): {}/{} ({})\n".format(sum(list(specialValue_discrepancyTally.values())), specialValueEvals, specialValueInputTotal))
         f.write("testMigration_unique_discrepancyTally: {}\n".format(tempy))
-        #f.write("\t# of Unique Discrepancies/# of Evals (# of Unique Inputs): {}/{} ({})\n".format(sum(list(testMigration_discrepancyTally.values())), testMigrationEvals, testMigrationInputTotal))
         f.write("s3fp_unique_discrepancyTally: {}\n".format(tempz))
-        #f.write("\t# of Unique Discrepancies/# of Evals (# of Unique Inputs): {}/{} ({})\n".format(sum(list(s3fp_discrepancyTally.values())), s3fpEvals, s3fpInputTotal))
 
         f.write("\nUNIQUE DISCREPANCY TOTALS:\n")
         f.write("\tTIMEOUTS: {}\n".format(total_discrepancyTally[6]))
@@ -178,6 +182,14 @@ def getStats(UNIQUE_DISCREPANCIES, ALL_DISCREPANCIES, USED_INPUTS):
         f.write("\tMIX OF EXCEPTIONS AND SPECIAL VALUES: {}\n".format(total_discrepancyTally[1]))
         f.write("\n\t\tTOTAL # OF DISCREPANCIES: {}\n".format(sum(list(total_discrepancyTally.values()))))
 
+    # write contents of statistics file to stdout
+    with open("logs/statistics.txt", "r") as f:
+        line = f.readline()
+        print()
+        while line:
+            print(line, end='')
+            line = f.readline()
+
 
 def inspect(UNIQUE_DISCREPANCIES):
     inspected_discrepancies = {}
@@ -244,20 +256,15 @@ def runDiffTester(start, end):
         for i in range(start, end):
             diffTester(classNoList[i], classKeyList[i], len(list(CLASSES.keys())), CLASSES[classKeyList[i]], TEST_INPUTS, UNIQUE_DISCREPANCIES_MANAGER, ALL_DISCREPANCIES_MANAGER, USED_INPUTS_MANAGER)
 
-    # check user tuned multi-task number
-    if MAX_THREAD == "max":
-        NUM_MULTIPROCESSING = os.cpu_count()
-    elif MAX_THREAD > len(classNoList):
-        NUM_MULTIPROCESSING = os.cpu_count()
-    else:
-        NUM_MULTIPROCESSING = MAX_THREAD
-
     # find break points for chunks of classes
     breakPoints = []
     breakPoints.append(0)
     for i in range(1, NUM_MULTIPROCESSING):
         breakPoints.append(int(len(classNoList) * (i / NUM_MULTIPROCESSING)))
     breakPoints.append(len(classNoList))
+    numDiffTester = len(classNoList)
+
+    print("\n** Running Differential Tester: ")
 
     # create multi-process
     processes = []
@@ -270,6 +277,9 @@ def runDiffTester(start, end):
     for process in processes:
         process.join()
 
+    sys.stdout.write('\n')
+
+
     # convert multiprocess data structure back to normal python data structure
     UNIQUE_DISCREPANCIES = dict(UNIQUE_DISCREPANCIES_MANAGER)
     ALL_DISCREPANCIES = list(ALL_DISCREPANCIES_MANAGER)

workspace/runExperiment.sh

@@ -1,10 +1,14 @@
 cd /usr/local/src/fp-diff-testing/workspace && make reset
 
+echo ""
+echo "** Running Extractor"
 python3 /usr/local/src/fp-diff-testing/workspace/extractor.py mpmath /usr/local/lib/python3.6/dist-packages/mpmath/tests/ python >> /usr/local/src/fp-diff-testing/workspace/logs/__extractedSignatures.txt
 python3 /usr/local/src/fp-diff-testing/workspace/extractor.py scipy /usr/local/lib/python3.6/dist-packages/scipy/special/tests/ python >> /usr/local/src/fp-diff-testing/workspace/logs/__extractedSignatures.txt
 python3 /usr/local/src/fp-diff-testing/workspace/extractor.py gsl /usr/local/lib/gsl/specfunc/ c >> /usr/local/src/fp-diff-testing/workspace/logs/__extractedSignatures.txt
 python3 /usr/local/src/fp-diff-testing/workspace/extractor.py jmat /usr/local/lib/jmat/ javascript >> /usr/local/src/fp-diff-testing/workspace/logs/__extractedSignatures.txt
 
+echo ""
+echo "** Running Driver Generator"
 python3 /usr/local/src/fp-diff-testing/workspace/driverGenerator.py mpmath python
 python3 /usr/local/src/fp-diff-testing/workspace/driverGenerator.py scipy python
 python3 /usr/local/src/fp-diff-testing/workspace/driverGenerator.py gsl c