Skip to content

P
pytensor
提交 28785b40 authored 作者: Frederic Bastien's avatar Frederic Bastien
浏览文件
操作

Remove ProfileMode

上级 c193af5c
隐藏空白字符变更
内嵌 并排
正在显示 包含 41 行增加 和 369 行删除
Theano.pyproj
浏览文件 @ 28785b40
......@@ -34,7 +34,6 @@
<Compile Include="theano\compile\mode.py" />
<Compile Include="theano\compile\module.py" />
<Compile Include="theano\compile\pfunc.py" />
<Compile Include="theano\compile\profilemode.py" />
<Compile Include="theano\compile\profiling.py" />
<Compile Include="theano\compile\sandbox\__init__.py" />
<Compile Include="theano\compile\sharedvalue.py" />
......
benchmark/autoencoder/aa.py
浏览文件 @ 28785b40
......@@ -8,7 +8,7 @@ import theano
import theano.tensor as T
import theano.sandbox
from six.moves import xrange
from theano.compile import module, Mode, ProfileMode
from theano.compile import module, Mode
from theano import gof, Op, Apply
from theano.tensor import blas, opt
......@@ -191,7 +191,6 @@ class M(module.Module):
mod = M()
mode = 'FAST_RUN'
#mode = ProfileMode(optimizer='fast_run', linker=theano.gof.OpWiseCLinker())
mode = Mode(optimizer='fast_run', linker=theano.gof.OpWiseCLinker(nice_errors=True))
mode = Mode(optimizer='fast_run', linker='c')
mode = Mode(optimizer='fast_run', linker='c|py')
......
benchmark/regression/regression.py
浏览文件 @ 28785b40
......@@ -91,7 +91,6 @@ class PrintEverythingMode(theano.Mode):
def test_module_advanced_example():
profmode = theano.ProfileMode(optimizer='fast_run', linker=theano.gof.OpWiseCLinker())
profmode = PrintEverythingMode(theano.gof.OpWiseCLinker(), 'fast_run')
data_x = N.random.randn(4, 10)
......
doc/library/compile/index.txt
浏览文件 @ 28785b40
......@@ -19,7 +19,6 @@
ops
mode
debugmode
profilemode
nanguardmode
......
doc/library/compile/mode.txt
浏览文件 @ 28785b40
......@@ -21,10 +21,8 @@ Theano defines the following modes by name:
- ``'FAST_COMPILE'``: Apply just a few graph optimizations and only use Python implementations.
- ``'FAST_RUN'``: Apply all optimizations, and use C implementations where possible.
- ``'DebugMode'``: A mode for debugging. See :ref:`DebugMode <debugmode>` for details.
- ``'ProfileMode'``: Deprecated, use the Theano flag :attr:`config.profile`.
- ``'NanGuardMode``: :ref:`Nan detector <nanguardmode>`
- ``'DEBUG_MODE'``: Deprecated. Use the string DebugMode.
- ``'PROFILE_MODE'``: Deprecated, use the Theano flag :attr:`config.profile`.
The default mode is typically ``FAST_RUN``, but it can be controlled via the
configuration variable :attr:`config.mode`, which can be
......
doc/library/compile/profilemode.txt
浏览文件 @ 28785b40
......@@ -16,203 +16,4 @@ Guide
.. note::
ProfileMode is deprecated. Use :attr:`config.profile` instead.
To profile a Theano graph, a special mode called ProfileMode, must be passed as
an argument when compiling your graph. Using ProfileMode is a three-step
process.
Creating a ProfileMode Instance
-------------------------------
First create a ProfileMode instance.
>>> import theano
>>> from theano import ProfileMode
>>> profmode = theano.ProfileMode(optimizer='fast_run', linker=theano.gof.OpWiseCLinker())
The ProfileMode constructor takes as input an optimizer and a
linker. Which optimizer and linker to use will depend on the
application. For example, a user wanting to profile the Python
implementation only, should use the gof.PerformLinker (or "py" for
short). On the other hand, a user wanting to profile his graph using C
implementations wherever possible should use the ``gof.OpWiseCLinker``
(or "c|py").
In the same manner, modifying which optimizer is passed to ProfileMode
will decide which optimizations are applied to the graph, prior to
profiling. Changing the optimizer should be especially useful when
developing new graph optimizations, in order to evaluate their impact
on performance. Also keep in mind that optimizations might change the
computation graph a lot, meaning that you might not recognize some of
the operations that are profiled (you did not use them explicitly but
an optimizer decided to use it to improve performance or numerical
stability). If you cannot easily relate the output of ProfileMode with
the computations you defined, you might want to try setting optimizer
to None (but keep in mind the computations will be slower than if they
were optimized).
Note that most users will want to use ProfileMode to optimize their
graph and find where most of the computation time is being spent. In
this context, 'fast_run' optimizer and ``gof.OpWiseCLinker`` are the
most appropriate choices.
Compiling your Graph with ProfileMode
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Once the ProfileMode instance is created, simply compile your graph as you
would normally, by specifying the mode parameter.
.. testsetup::
import theano
input1, input2 = theano.tensor.scalars(2)
output1 = input1+input2
>>> # with functions
>>> f = theano.function([input1,input2],[output1], mode=profmode)
Retrieving Timing Information
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Once your graph is compiled, simply run the program or operation you wish to
profile, then call ``profmode.print_summary()``. This will provide you with
the desired timing information, indicating where your graph is spending most
of its time.
This is best shown through an example.
Lets use the example of logistic
regression. (Code for this example is in the file
``benchmark/regression/regression.py``.)
Compiling the module with ProfileMode and calling ``profmode.print_summary()``
generates the following output:
.. code-block:: python
"""
ProfileMode.print_summary()
---------------------------
local_time 0.0749197006226 (Time spent running thunks)
Apply-wise summary: <fraction of local_time spent at this position> (<Apply position>, <Apply Op name>)
0.069 15 _dot22
0.064 1 _dot22
0.053 0 InplaceDimShuffle{x,0}
0.049 2 InplaceDimShuffle{1,0}
0.049 10 mul
0.049 6 Elemwise{ScalarSigmoid{output_types_preference=<theano.scalar.basic.transfer_type object at 0x171e650>}}[(0, 0)]
0.049 3 InplaceDimShuffle{x}
0.049 4 InplaceDimShuffle{x,x}
0.048 14 Sum{0}
0.047 7 sub
0.046 17 mul
0.045 9 sqr
0.045 8 Elemwise{sub}
0.045 16 Sum
0.044 18 mul
... (remaining 6 Apply instances account for 0.25 of the runtime)
Op-wise summary: <fraction of local_time spent on this kind of Op> <Op name>
0.139 * mul
0.134 * _dot22
0.092 * sub
0.085 * Elemwise{Sub{output_types_preference=<theano.scalar.basic.transfer_type object at 0x1779f10>}}[(0, 0)]
0.053 * InplaceDimShuffle{x,0}
0.049 * InplaceDimShuffle{1,0}
0.049 * Elemwise{ScalarSigmoid{output_types_preference=<theano.scalar.basic.transfer_type object at 0x171e650>}}[(0, 0)]
0.049 * InplaceDimShuffle{x}
0.049 * InplaceDimShuffle{x,x}
0.048 * Sum{0}
0.045 * sqr
0.045 * Sum
0.043 * Sum{1}
0.042 * Elemwise{Mul{output_types_preference=<theano.scalar.basic.transfer_type object at 0x17a0f50>}}[(0, 1)]
0.041 * Elemwise{Add{output_types_preference=<theano.scalar.basic.transfer_type object at 0x1736a50>}}[(0, 0)]
0.039 * Elemwise{Second{output_types_preference=<theano.scalar.basic.transfer_type object at 0x1736d90>}}[(0, 1)]
... (remaining 0 Ops account for 0.00 of the runtime)
(*) Op is running a c implementation
"""
.. note::
***TODO***
The following text was recovered from a recent version of the source
file... hopefully things haven't gotten too out-of-sync!
The first show an Apply-wise summary, the second show an Op-wise summary, the third show an type-Op-wise summary.
The Apply-wise summary print the timing information for the worst
offending Apply nodes. This corresponds to individual Op applications
within your graph which take the longest to execute (so if you use dot
twice, you will see two entries there).
The Op-wise summary print the execution time of all Apply nodes
executing the same Op are grouped together and the total execution
time per Op is shown (so if you use dot twice, you will see only one
entry there corresponding to the sum of the time spent in each of
them). If two Op have different hash value, they will be separate.
The type-Op-wise summary group the result by type of op. So event if
two Op have different hash value, they will be merged.
Their is an hack with the Op-wise summary. Go see it if you want to know more.
The summary has two components to it. In the first section called the
Apply-wise summary, timing information is provided for the worst
offending Apply nodes. This corresponds to individual Op applications
within your graph which take the longest to execute (so if you use
``dot`` twice, you will see two entries there). In the second portion,
the Op-wise summary, the execution time of all Apply nodes executing
the same Op are grouped together and the total execution time per Op
is shown (so if you use ``dot`` twice, you will see only one entry
there corresponding to the sum of the time spent in each of them).
Note that the ProfileMode also shows which Ops were running a c
implementation.
Developers wishing to optimize the performance of their graph should
focus on the worst offending Ops and Apply nodes -- either by optimizing an
implementation, providing a missing C implementation, or by writing a graph
optimization that eliminates the offending Op altogether.
You should strongly consider emailing one of our lists about your issue before
spending too much time on this.
Reference
=========
.. class:: ProfileMode(Mode)
.. method:: print_summary(n_apply_to_print=None, n_ops_to_print=None)
Print three summaries to stdout that show where cpu time is spent during theano function executions (for all functions using this object instance).
:param n_apply_to_print: the number of apply nodes to print.
The default 15, but can be configured via ``ProfileMode.n_ops_to_print`` in :envvar:`THEANO_FLAGS`.
:param n_ops_to_print: the number of ops to print.
Default 20, or but can be configured via ``ProfileMode.n_apply_to_print`` in :envvar:`THEANO_FLAGS`.
:returns: None
.. method:: print_diff_summary(self, other, n_apply_to_print=None, n_ops_to_print=None):
""" As print_summary, but print the difference on two different profile mode.
TODO: Also we don't print the Apply-wise summary as it don't work for now.
TODO: make comparaison with gpu code.
:param other: the other instance of ProfileMode that we want to be compared to.
:param n_apply_to_print: the number of apply nodes to print.
The default 15, but can be configured via ``ProfileMode.n_ops_to_print`` in :envvar:`THEANO_FLAGS`.
:param n_ops_to_print: the number of ops to print.
Default 20, or but can be configured via ``ProfileMode.n_apply_to_print`` in :envvar:`THEANO_FLAGS`.
:returns: None
ProfileMode is removed. Use :attr:`config.profile` instead.
doc/library/config.txt
浏览文件 @ 28785b40
......@@ -315,7 +315,7 @@ import theano and print the config variable, as in:
.. attribute:: mode
String value: ``'Mode'``, ``'ProfileMode'`` (deprecated), ``'DebugMode'``, ``'FAST_RUN'``,
String value: ``'Mode'``, ``'DebugMode'``, ``'FAST_RUN'``,
``'FAST_COMPILE'``
Default: ``'Mode'``
......
doc/tutorial/debug_faq.txt
浏览文件 @ 28785b40
......@@ -284,7 +284,7 @@ First, make sure you're running in ``FAST_RUN`` mode. Even though
to ``theano.function`` (or ``theano.make``) or by setting :attr:`config.mode`
to ``FAST_RUN``.
Second, try the Theano :ref:`using_profilemode`. This will tell you which
Second, try the Theano :ref:`profiling <tut_profiling>`. This will tell you which
``Apply`` nodes, and which ops are eating up your CPU cycles.
Tips:
......
doc/tutorial/modes.txt
浏览文件 @ 28785b40
......@@ -248,13 +248,3 @@ constructor arguments. The keyword version of DebugMode (which you get by using
is quite strict.
For more detail, see :ref:`DebugMode<debugmode>` in the library.
.. _using_profilemode:
ProfileMode
===========
.. note::
ProfileMode is deprecated. Use :attr:`config.profile` instead.
theano/__init__.py
浏览文件 @ 28785b40
......@@ -73,7 +73,7 @@ from theano.compile import (
Mode,
predefined_modes, predefined_linkers, predefined_optimizers,
FunctionMaker, function, function_dump, OpFromGraph,
ProfileMode, ProfileStats,
ProfileStats,
Param, shared, as_op)
from theano.misc.safe_asarray import _asarray
......
theano/compile/__init__.py
浏览文件 @ 28785b40
......@@ -19,8 +19,6 @@ from theano.compile.monitormode import MonitorMode
from theano.compile.profiling import ProfileStats, ScanProfileStats
from theano.compile.profilemode import ProfileMode
from theano.compile.sharedvalue import (shared, shared_constructor,
SharedVariable)
from theano.compile.pfunc import pfunc, Param, rebuild_collect_shared
......
theano/compile/function_module.py
浏览文件 @ 28785b40
......@@ -1381,17 +1381,11 @@ class FunctionMaker(object):
output_keys=None):
mode = theano.compile.mode.get_mode(mode)
# figure out which profile object to use (if any)
# to help with forward-porting ProfileMode,
# we allow ProfileMode to provide a ProfileStats object
# using this somewhat awkward mechanism.
mode_profile = getattr(mode, 'profile', None)
if (profile is not None and
profile is not False and
mode_profile is not None):
# Assert old way of working isn't used
if getattr(mode, 'profile', None):
raise TypeError(
'profile passed via both "mode" and "profile" arguments')
self.profile = profile = profile or mode_profile
"profile passed via 'mode'. This isn't supported anymore")
self.profile = profile
if profile:
# This is very important:
# 1) We preload the cache here to don't have its timming
......@@ -1748,9 +1742,6 @@ def orig_function(inputs, outputs, mode=None, accept_inplace=False,
- FAST_COMPILE (minimal optimization)
- ProfileMode(deprecated): allow to print a profile mode with
mode.print_summary
- DebugMode: verify many internal conditions that are normally assumed
(slow)
......
theano/compile/monitormode.py
浏览文件 @ 28785b40
......@@ -79,7 +79,7 @@ class MonitorMode(Mode):
Create a new instance of this Mode.
Keyword arguments can be provided for the linker, but they will be
ignored, because ProfileMode needs to use its own linker.
ignored, because MonitorMode needs to use its own linker.
"""
if optimizer == "":
......
theano/compile/profiling.py
浏览文件 @ 28785b40
......@@ -3,8 +3,6 @@ ProfileStats object for runtime and memory profiling.
"""
#
# TODO: measure memory usage like ProfileMode did
# TODO: put the optimization tips into a tips section??
# TODO: add tip to use specify_shape (is specify_shape even in library doc?)
# TODO: ensure field width for string fields makes columns line up
# TODO: what to do about 'diff summary'? (ask Fred?)
......@@ -378,7 +376,7 @@ class ProfileStats(object):
else:
local_time = 0
if local_time == 0:
print(('ProfileMode.summary_class: total time 0'
print(('ProfileStats.summary_class: total time 0'
' (did you forget to enable counters?)'), file=file)
return
class_time = self.class_time()
......@@ -462,7 +460,7 @@ class ProfileStats(object):
else:
local_time = 0
if local_time == 0:
print(('ProfileMode.summary_ops: total time 0'
print(('ProfileStats.summary_ops: total time 0'
' (did you forget to enable counters?)'), file=file)
return
op_time = self.op_time()
......@@ -540,7 +538,7 @@ class ProfileStats(object):
else:
local_time = 0
if local_time == 0:
print(('ProfileMode.summary_nodes: total time 0'
print(('ProfileStats.summary_nodes: total time 0'
' (did you forget to enable counters?)'), file=file)
return
......
theano/compile/tests/test_modes.py
浏览文件 @ 28785b40
......@@ -7,7 +7,7 @@ import unittest
import theano
import theano.tensor as T
from theano.compile import Mode, ProfileMode
from theano.compile import Mode
class T_bunch_of_modes(unittest.TestCase):
......@@ -18,9 +18,6 @@ class T_bunch_of_modes(unittest.TestCase):
linker_classes_involved = []
predef_modes = ['FAST_COMPILE', 'FAST_RUN', 'DEBUG_MODE']
# Use a new instance of ProfileMode instead of 'ProfileMode' to
# avoid printing a profile mode summary in nose output
predef_modes.append(ProfileMode())
# Linkers to use with regular Mode
if theano.config.cxx:
......@@ -43,20 +40,14 @@ class T_bunch_of_modes(unittest.TestCase):
# there should be
# - VM_Linker
# - OpWiseCLinker (FAST_RUN)
# - WrapLinker ("ProfileMode")
# - WrapLinker
# - PerformLinker (FAST_COMPILE)
# - DebugMode's Linker (DEBUG_MODE)
assert 5 == len(set(linker_classes_involved))
class T_ProfileMode_WrapLinker(unittest.TestCase):
class T_WrapLinker(unittest.TestCase):
def test_1(self):
# First, compile a function with a new ProfileMode() object
# No need to call that function
x = T.matrix()
mode = ProfileMode()
theano.function([x], x * 2, mode=mode)
# Then, build a mode with the same linker, and a modified optimizer
default_mode = theano.compile.mode.get_default_mode()
modified_mode = default_mode.including('specialize')
......
theano/configdefaults.py
浏览文件 @ 28785b40
......@@ -405,9 +405,9 @@ AddConfigVar(
AddConfigVar(
'mode',
"Default compilation mode",
EnumStr('Mode', 'ProfileMode', 'DebugMode', 'FAST_RUN',
EnumStr('Mode', 'DebugMode', 'FAST_RUN',
'NanGuardMode',
'FAST_COMPILE', 'PROFILE_MODE', 'DEBUG_MODE'),
'FAST_COMPILE', 'DEBUG_MODE'),
in_c_key=False)
param = "g++"
......@@ -463,8 +463,7 @@ del param
if rc == 0 and config.cxx != "":
# Keep the default linker the same as the one for the mode FAST_RUN
AddConfigVar('linker',
("Default linker used if the theano flags mode is Mode "
"or ProfileMode(deprecated)"),
"Default linker used if the theano flags mode is Mode ",
EnumStr('cvm', 'c|py', 'py', 'c', 'c|py_nogc',
'vm', 'vm_nogc', 'cvm_nogc'),
in_c_key=False)
......@@ -472,8 +471,7 @@ else:
# g++ is not present or the user disabled it,
# linker should default to python only.
AddConfigVar('linker',
("Default linker used if the theano flags mode is Mode "
"or ProfileMode(deprecated)"),
"Default linker used if the theano flags mode is Mode ",
EnumStr('vm', 'py', 'vm_nogc'),
in_c_key=False)
try:
......@@ -501,8 +499,7 @@ AddConfigVar('allow_gc',
# Keep the default optimizer the same as the one for the mode FAST_RUN
AddConfigVar(
'optimizer',
("Default optimizer. If not None, will use this linker with the Mode "
"object (not ProfileMode(deprecated) or DebugMode)"),
"Default optimizer. If not None, will use this optimizer with the Mode ",
EnumStr('fast_run', 'merge', 'fast_compile', 'None'),
in_c_key=False)
......@@ -951,27 +948,6 @@ AddConfigVar('NanGuardMode.action',
EnumStr('raise', 'warn', 'pdb'),
in_c_key=False)
AddConfigVar('ProfileMode.n_apply_to_print',
"Number of apply instances to print by default",
IntParam(15, lambda i: i > 0),
in_c_key=False)
AddConfigVar('ProfileMode.n_ops_to_print',
"Number of ops to print by default",
IntParam(20, lambda i: i > 0),
in_c_key=False)
AddConfigVar('ProfileMode.min_memory_size',
"For the memory profile, do not print apply nodes if the size "
"of their outputs (in bytes) is lower then this threshold",
IntParam(1024, lambda i: i >= 0),
in_c_key=False)
AddConfigVar('ProfileMode.profile_memory',
"""Enable profiling of memory used by Theano functions""",
BoolParam(False),
in_c_key=False)
AddConfigVar('optimizer_excluding',
("When using the default mode, we will remove optimizer with "
"these tags. Separate tags with ':'."),
......
theano/d3viz/formatting.py
浏览文件 @ 28785b40
......@@ -11,7 +11,6 @@ from six import iteritems, itervalues
import theano
from theano import gof
from theano.compile.profilemode import ProfileMode
from theano.compile import Function
from theano.compile import builders
from theano.printing import pydot_imported, pydot_imported_msg
......@@ -123,14 +122,7 @@ class PyDotFormatter(object):
profile = None
if isinstance(fct, Function):
mode = fct.maker.mode
if (not isinstance(mode, ProfileMode) or
fct not in mode.profile_stats):
mode = None
if mode:
profile = mode.profile_stats[fct]
else:
profile = getattr(fct, "profile", None)
profile = getattr(fct, "profile", None)
outputs = fct.maker.fgraph.outputs
topo = fct.maker.fgraph.toposort()
elif isinstance(fct, gof.FunctionGraph):
......
theano/printing.py
浏览文件 @ 28785b40
......@@ -20,7 +20,6 @@ from theano import gof
from theano import config
from theano.gof import Op, Apply
from theano.compile import Function, debugmode, SharedVariable
from theano.compile.profilemode import ProfileMode
pydot_imported = False
pydot_imported_msg = ""
......@@ -746,15 +745,10 @@ def pydotprint(fct, outfile=None,
config.device + '.' + format)
if isinstance(fct, Function):
mode = fct.maker.mode
profile = getattr(fct, "profile", None)
if (not isinstance(mode, ProfileMode) or
fct not in mode.profile_stats):
mode = None
outputs = fct.maker.fgraph.outputs
topo = fct.maker.fgraph.toposort()
elif isinstance(fct, gof.FunctionGraph):
mode = None
profile = None
outputs = fct.outputs
topo = fct.toposort()
......@@ -767,7 +761,6 @@ def pydotprint(fct, outfile=None,
assert all(isinstance(v, gof.Variable) for v in fct)
fct = gof.FunctionGraph(inputs=gof.graph.inputs(fct),
outputs=fct)
mode = None
profile = None
outputs = fct.outputs
topo = fct.toposort()
......@@ -855,19 +848,7 @@ def pydotprint(fct, outfile=None,
if node in apply_name_cache:
return apply_name_cache[node], apply_name_id[node]
prof_str = ''
if mode:
time = mode.profile_stats[fct].apply_time.get(node, 0)
# second, % total time in profiler, %fct time in profiler
if mode.local_time == 0:
pt = 0
else:
pt = time * 100 / mode.local_time
if mode.profile_stats[fct].fct_callcount == 0:
pf = 0
else:
pf = time * 100 / mode.profile_stats[fct].fct_call_time
prof_str = ' (%.3fs,%.3f%%,%.3f%%)' % (time, pt, pf)
elif profile:
if profile:
time = profile.apply_time.get(node, 0)
# second, %fct time in profiler
if profile.fct_callcount == 0:
......
theano/sandbox/cuda/blas.py
浏览文件 @ 28785b40
......@@ -907,7 +907,7 @@ class BaseGpuCorrMM(GpuOp):
def flops(self, inp, outp):
"""
Useful with the hack in profilemode to print the MFlops.
Useful with the hack in profiling to print the MFlops.
"""
# if the output shape is correct, then this gives the correct
......@@ -1421,7 +1421,7 @@ class BaseGpuCorr3dMM(GpuOp):
self.pad)
def flops(self, inp, outp):
""" Useful with the hack in profilemode to print the MFlops"""
""" Useful with the hack in profiling to print the MFlops"""
# if the output shape is correct, then this gives the correct
# flops for any direction, sampling, padding, and border mode
inputs, filters = inp
......@@ -2101,7 +2101,7 @@ class GpuConv(GpuOp):
return Apply(self, [img, kern], [CudaNdarrayType(broadcastable)()])
def flops(self, inputs, outputs):
""" Useful with the hack in profilemode to print the MFlops"""
""" Useful with the hack in profiling to print the MFlops"""
images, kerns = inputs
out, = outputs
assert images[1] == kerns[1]
......
theano/sandbox/cuda/tests/test_basic_ops.py
浏览文件 @ 28785b40
......@@ -1370,12 +1370,12 @@ def speed_adv_sub1():
vec = tensor.lvector()
for batch_size in [100, 1000, 10000, 100000]:
idx = numpy.random.randint(0, 50000, batch_size)
mode_with_gpu = theano.compile.ProfileMode().including('gpu')
f = theano.function([vec], var[vec], mode=mode_with_gpu)
mode_with_gpu = theano.compile.get_default_mode().including('gpu')
f = theano.function([vec], var[vec], mode=mode_with_gpu, profile=True)
for i in range(100):
f(idx)
print("ProfileMode with batch size", batch_size)
mode_with_gpu.print_summary()
print("profile with batch size", batch_size)
mode_with_gpu.summary()
def speed_reduce10():
......
theano/sandbox/cuda/tests/test_mlp.py
浏览文件 @ 28785b40
......@@ -19,7 +19,7 @@ import theano.sandbox.cuda as tcn
import theano.tests.unittest_tools as utt
if theano.config.mode not in ['FAST_RUN', 'Mode', 'ProfileMode']:
if theano.config.mode not in ['FAST_RUN', 'Mode']:
raise SkipTest('Skip test_mlp when not in normal optimization mode as '
'otherwise it is too slow!')
......@@ -48,8 +48,6 @@ def get_mode(use_gpu, check_isfinite=True):
ret = theano.compile.get_default_mode()
else:
ret = theano.compile.mode.get_mode('FAST_RUN')
if isinstance(ret, theano.compile.ProfileMode):
ret = copy.copy(ret)
if isinstance(ret, theano.compile.DebugMode):
ret = copy.copy(ret)
ret.check_isfinite = check_isfinite
......@@ -60,19 +58,6 @@ def get_mode(use_gpu, check_isfinite=True):
return ret
def print_mode(mode):
if mode is not None and isinstance(mode, (theano.compile.ProfileMode,)):
mode.print_summary()
def print_diff_mode(a, b):
if (a is not None and
isinstance(a, (theano.compile.ProfileMode,)) and
isinstance(b, (theano.compile.ProfileMode,))):
a.print_diff_summary(b)
def run_nnet(use_gpu, n_batch=60, n_in=1024, n_hid=2048, n_out=10,
n_train=100):
......@@ -123,7 +108,6 @@ def run_nnet(use_gpu, n_batch=60, n_in=1024, n_hid=2048, n_out=10,
rval.append(train(xval, yval, lr))
dt = time.time() - t0
print_mode(mode)
return numpy.asarray(rval), dt
......@@ -220,7 +204,6 @@ def run_conv_nnet1(use_gpu):
for i in xrange(n_train):
rval = train(xval, yval, lr)
# print 'training done'
print_mode(mode)
return rval
......@@ -316,7 +299,6 @@ def run_conv_nnet2(use_gpu): # pretend we are training LeNet for MNIST
for i in xrange(n_train):
rval = train(xval, yval, lr)
print_mode(mode)
return rval
......@@ -428,7 +410,6 @@ def build_conv_nnet2_classif(use_gpu, isize, ksize, n_batch,
def run_conv_nnet2_classif(use_gpu, seed, isize, ksize, bsize,
n_train=10,
check_isfinite=True,
pickle=False,
verbose=0,
version=-1):
"""Run the train function returned by build_conv_nnet2_classif on one device.
......@@ -456,17 +437,6 @@ def run_conv_nnet2_classif(use_gpu, seed, isize, ksize, bsize,
rvals = my_zeros(n_train)
for i in xrange(n_train):
rvals[i] = train(xval, yval, lr)[0]
print_mode(mode)
if pickle and isinstance(mode, theano.compile.ProfileMode):
import pickle
print("BEGIN %s profile mode dump" % device)
print(pickle.dumps(mode))
print("END %s profile mode dump" % device)
# print "%s time: %.3f" % (device, t1-t0)
# print "estimated time for one pass through MNIST with %s: %f" % (
# device, (t1-t0) * (60000.0 / (n_train*bsize)))
def cmp_run_conv_nnet2_classif(seed, isize, ksize, bsize,
......@@ -476,7 +446,6 @@ def cmp_run_conv_nnet2_classif(seed, isize, ksize, bsize,
cpu_only=False,
float_atol=1e-06,
check_isfinite=True,
pickle=False,
verbose=0,
version=-1):
"""Run the nnet2 function on 1 or 2 devices, and compares the results.
......@@ -512,7 +481,6 @@ def cmp_run_conv_nnet2_classif(seed, isize, ksize, bsize,
seed=seed, isize=isize, ksize=ksize, bsize=bsize,
n_train=n_train,
check_isfinite=check_isfinite,
pickle=pickle,
verbose=verbose,
version=version)
......
theano/scan_module/scan_op.py
浏览文件 @ 28785b40
......@@ -175,7 +175,7 @@ class Scan(PureOp):
mode_instance = compile.mode.get_mode(self.mode)
# Clone mode_instance, altering "allow_gc" for the linker,
# and adding a message if the mode is a ProfileMode.
# and adding a message if we profile
if self.name:
message = self.name + " sub profile"
else:
......@@ -1564,14 +1564,6 @@ class Scan(PureOp):
if hasattr(self.fn.fn, 'update_profile'):
self.fn.fn.update_profile(profile)
#/* Old ProfileMode
# if hasattr(self.fn.maker.mode,'fct_call_time'):
# self.fn.maker.mode.fct_call_time[self.fn] += t_fn
# self.fn.maker.mode.fct_call[self.fn] += n_steps
#self.fn.maker.mode.call_time += t_fn
#self.fn.maker.mode.fn_time += t_fn
# Old Profile Mode */
self.t_call = t_call
self.t_fn = t_fn
......
theano/tensor/nnet/abstract_conv.py
浏览文件 @ 28785b40
......@@ -719,7 +719,7 @@ class BaseAbstractConv2d(Op):
self.filter_dilation = tuple(filter_dilation)
def flops(self, inp, outp):
""" Useful with the hack in profilemode to print the MFlops"""
""" Useful with the hack in profiling to print the MFlops"""
# if the output shape is correct, then this gives the correct
# flops for any direction, sampling, padding, and border mode
inputs, filters = inp
......
theano/tensor/nnet/conv.py
浏览文件 @ 28785b40
......@@ -609,7 +609,7 @@ class ConvOp(OpenMPOp):
def flops(self, inputs, outputs):
"""
Useful with the hack in profilemode to print the MFlops.
Useful with the hack in profiling to print the MFlops.
"""
images, kerns = inputs
......
theano/tensor/tests/test_opt.py
浏览文件 @ 28785b40
......@@ -1394,11 +1394,7 @@ class test_fusion(unittest.TestCase):
def speed_log_exp(self):
s = slice(31, 36)
# linker=gof.CLinker
linker = gof.OpWiseCLinker
mode = compile.Mode(linker(), copy.copy(compile.mode.OPT_FAST_RUN))
mode = compile.ProfileMode()
print("time", self.do(mode, shared, shp=(1000, 1000), gpu=False,
print("time", self.do(None, shared, shp=(1000, 1000), gpu=False,
assert_len_topo=False, slice=s, nb_repeat=100))
def tes_memory_leak(self, mode=compile.mode.Mode('c', 'merge'),
......
theano/tests/test_printing.py
浏览文件 @ 28785b40
......@@ -115,15 +115,19 @@ def test_pydotprint_long_name():
def test_pydotprint_profile():
"""Just check that pydotprint does not crash with ProfileMode."""
"""Just check that pydotprint does not crash with profile."""
# Skip test if pydot is not available.
if not theano.printing.pydot_imported:
raise SkipTest('pydot not available')
A = tensor.matrix()
f = theano.function([A], A + 1, mode='ProfileMode')
prof = theano.compile.ProfileStats(atexit_print=False)
f = theano.function([A], A + 1, profile=prof)
theano.printing.pydotprint(f, print_output_file=False)
f([[1]])
theano.printing.pydotprint(f, print_output_file=False)
def test_min_informative_str():
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论