Merge pull request #2243 from RoyXue/fix_crash_and_GPU_support

theano/compile/tests/test_profiling.py

@@ -8,6 +8,7 @@ import numpy
 
 import theano
 import theano.tensor as T
+from theano.ifelse import ifelse
 
 
 def test_profiling():
@@ -20,11 +21,11 @@ def test_profiling():
         theano.config.profile_memory = True
         theano.config.profiling.min_peak_memory = True
 
-        x = [T.dvector("val%i" % i) for i in range(3)]
+        x = [T.fvector("val%i" % i) for i in range(3)]
 
         z = []
-        z += [T.outer(x[i], x[i+1]).sum(axis=1) for i in range(len(x)-1)]
-        z += [x[i] + x[i+1] for i in range(len(x)-1)]
+        z += [T.outer(x[i], x[i + 1]).sum(axis=1) for i in range(len(x) - 1)]
+        z += [x[i] + x[i + 1] for i in range(len(x) - 1)]
 
         p = theano.ProfileStats(False)
 
@@ -36,7 +37,7 @@ def test_profiling():
         f = theano.function(x, z, profile=p, name="test_profiling",
                             mode=m)
 
-        inp = [numpy.arange(1024) + 1 for i in range(len(x))]
+        inp = [numpy.arange(1024, dtype='float32') + 1 for i in range(len(x))]
         output = f(*inp)
 
         buf = StringIO.StringIO()
@@ -46,8 +47,16 @@ def test_profiling():
         the_string = buf.getvalue()
         lines1 = [l for l in the_string.split("\n") if "Max if linker" in l]
         lines2 = [l for l in the_string.split("\n") if "Minimum peak" in l]
-        assert "Max if linker=cvm(default): 8224KB (16408KB)" in the_string, (lines1, lines2)
-        assert "Minimum peak from all valid apply node order is 8208KB" in the_string, (lines1, lines2)
+        if theano.config.device == 'cpu':
+            assert "Max if linker=cvm(default): 4112KB (8204KB)" in the_string, (
+                lines1, lines2)
+            assert "Minimum peak from all valid apply node order is 4104KB" in the_string, (
+                lines1, lines2)
+        else:
+            assert "Max if linker=cvm(default): 8220KB (8220KB)" in the_string, (
+                lines1, lines2)
+            assert "Minimum peak from all valid apply node order is 4116KB" in the_string, (
+                lines1, lines2)
 
     finally:
         theano.config.profile = config1
@@ -55,5 +64,41 @@ def test_profiling():
         theano.config.profiling.min_peak_memory = config3
 
 
+def test_ifelse():
+    config1 = theano.config.profile
+    config2 = theano.config.profile_memory
+
+    try:
+        theano.config.profile = True
+        theano.config.profile_memory = True
+
+        a, b = T.scalars('a', 'b')
+        x, y = T.scalars('x', 'y')
+
+        z = ifelse(T.lt(a, b), x * 2, y * 2)
+
+        p = theano.ProfileStats(False)
+
+        if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]:
+            m = "FAST_RUN"
+        else:
+            m = None
+
+        f_ifelse = theano.function([a, b, x, y], z, profile=p, name="test_ifelse",
+                                   mode=m)
+
+        val1 = 0.
+        val2 = 1.
+        big_mat1 = 10
+        big_mat2 = 11
+
+        out = f_ifelse(val1, val2, big_mat1, big_mat2)
+
+    finally:
+        theano.config.profile = config1
+        theano.config.profile_memory = config2
+
+
 if __name__ == '__main__':
     test_profiling()
+    test_ifelse()

theano/gof/vm.py

@@ -56,6 +56,7 @@ raise_with_op = link.raise_with_op
 
 
 class VM(object):
+
     """
     A VM object's __call__ method evaluates a Theano program.
 
@@ -83,6 +84,7 @@ class VM(object):
         storage. False means it *must not* repeat that feedback.
 
     """
+
     def __init__(self, nodes, thunks, pre_call_clear):
         """
         Allocate a virtual machine.
@@ -159,10 +161,12 @@ class VM(object):
 
 
 class Loop(VM):
+
     """
     Unconditional start-to-finish program execution in Python.
     No garbage collection is allowed on intermediate results.
     """
+
     def __call__(self):
         if self.time_thunks:
             for cont in self.pre_call_clear:
@@ -188,10 +192,12 @@ class Loop(VM):
 
 
 class LoopGC(VM):
+
     """
     Unconditional start-to-finish program execution in Python.
     Garbage collection is possible on intermediate results.
     """
+
     def __init__(self, nodes, thunks, pre_call_clear, post_thunk_clear):
         super(LoopGC, self).__init__(nodes, thunks, pre_call_clear)
         self.post_thunk_clear = post_thunk_clear
@@ -231,6 +237,7 @@ class LoopGC(VM):
 
 
 class Stack(VM):
+
     """
     Finish-to-start evalution order of thunks.
 
@@ -340,7 +347,7 @@ class Stack(VM):
         apply_stack = list(self.base_apply_stack)
         last_apply_stack_len = -1
 
-        #This record all function inputs/shared varibles and constants
+        # This record all function inputs/shared varibles and constants
         for var, data in self.storage_map.iteritems():
             if data[0] is None:
                 continue
@@ -396,7 +403,7 @@ class Stack(VM):
                             current_idx = self.node_idx[current_apply]
                             self.call_counts[current_idx] += 1
                             self.call_times[current_idx] += dt
-                            ## Computing the memory footprint of the the op
+                            # Computing the memory footprint of the the op
                             # ?? What about inplace .. if the op is inplace
                             # you don't actually ask for more memory!
                             for (idx, o) in enumerate(
@@ -436,15 +443,16 @@ class Stack(VM):
                                 if all(compute_map[v][0]
                                         for v in dependencies[i]):
                                     storage_map[i][0] = None
-                                    input_index.append(current_apply.inputs.index(i))
+                                    input_index.append(
+                                        current_apply.inputs.index(i))
 
-                                    #DO NOT set compute_map to 0
+                                    # DO NOT set compute_map to 0
 
-                                    #If values become False and the
+                                    # If values become False and the
                                     #current_apply is still in the
-                                    #stack, this will cause it to be
-                                    #recomputed! This can cause wrong value
-                                    #with some combination of inplace op.
+                                    # stack, this will cause it to be
+                                    # recomputed! This can cause wrong value
+                                    # with some combination of inplace op.
                                     compute_map[i][0] = 2
                                     if (config.warn.vm_gc_bug and
                                         current_apply in apply_stack and
@@ -456,7 +464,8 @@ class Stack(VM):
                                             " only in the development version between July 5th 2012"
                                             " and July 30th 2012. This was not in a released version."
                                             " The bug was affecting this script.",
-        #The stack level is not good when inside a Scan.
+                                            # The stack level is not good when
+                                            # inside a Scan.
                                             stacklevel=3
                                         )
                     self.node_cleared_order.append(input_index)
@@ -468,7 +477,6 @@ class Stack(VM):
                                        for inp in current_deps
                                        if inp.owner)
 
-
             elif not computed_outs:
                 #
                 # stack loop: Lazy Evaluation Case
@@ -531,9 +539,10 @@ class Stack(VM):
                                         break
                                 if empty_storage_map:
                                     storage_map[i][0] = None
-                                    input_index.append(current_apply.inputs.index(i)) 
-                                    #See the not lazy gc code for explanations
-                                    #of compute_map change
+                                    input_index.append(
+                                        current_apply.inputs.index(i))
+                                    # See the not lazy gc code for explanations
+                                    # of compute_map change
                                     compute_map[i][0] = 2
 
                     self.node_cleared_order.append(input_index)
@@ -560,6 +569,7 @@ try:
     import lazylinker_c
 
     class CVM(lazylinker_c.CLazyLinker, VM):
+
         def __init__(self, *args, **kwargs):
             lazylinker_c.CLazyLinker.__init__(self, *args, **kwargs)
             # skip VM.__init__
@@ -576,6 +586,7 @@ except (OSError, theano.gof.cmodule.MissingGXX), e:
 
 
 class VM_Linker(link.LocalLinker):
+
     """
     Class that satisfies the Linker interface by acting as a VM factory.
     """