def testIsGoogleCudaEnabled(self):
# The test doesn't assert anything. It ensures the py wrapper
# function is generated correctly.
if test_util.IsGoogleCudaEnabled():
print("GoogleCuda is enabled")
else:
print("GoogleCuda is disabled")
def testTimelineGpu(self):
if not test_util.IsGoogleCudaEnabled():
return
run_options = tf.RunOptions(trace_level=tf.RunOptions.FULL_TRACE)
run_metadata = tf.RunMetadata()
with self.test_session(force_gpu=True) as sess:
const1 = tf.constant(1.0, name='const1')
const2 = tf.constant(2.0, name='const2')
result = tf.add(const1, const2) + const1 * const2
sess.run(result, options=run_options, run_metadata=run_metadata)
self.assertTrue(run_metadata.HasField('step_stats'))
step_stats = run_metadata.step_stats
devices = [d.device for d in step_stats.dev_stats]
self.assertTrue('/job:localhost/replica:0/task:0/gpu:0' in devices)
self.assertTrue('/gpu:0/stream:all' in devices)
tl = timeline.Timeline(step_stats)
ctf = tl.generate_chrome_trace_format()
self._validateTrace(ctf)
tl = timeline.Timeline(step_stats)
ctf = tl.generate_chrome_trace_format(show_dataflow=False)
self._validateTrace(ctf)
tl = timeline.Timeline(step_stats)
ctf = tl.generate_chrome_trace_format(show_memory=False)
self._validateTrace(ctf)
tl = timeline.Timeline(step_stats)
ctf = tl.generate_chrome_trace_format(show_memory=False,
show_dataflow=False)
self._validateTrace(ctf)
def GetTestConfigs():
"""Get all the valid tests configs to run.
Returns:
all the valid test configs as tuples of data_format and use_gpu.
"""
test_configs = [("NHWC", False), ("NHWC", True)]
if test_util.IsGoogleCudaEnabled():
# "NCHW" format is not currently supported on CPU.
test_configs += [("NCHW", True)]
return test_configs
def is_built_with_cuda():
"""Returns whether TensorFlow was built with CUDA (GPU) support.
This method should only be used in tests written with `tf.test.TestCase`. A
typical usage is to skip tests that should only run with CUDA (GPU).
>>> class MyTest(tf.test.TestCase):
...
... def test_add_on_gpu(self):
... if not tf.test.is_built_with_cuda():
... self.skipTest("test is only applicable on GPU")
...
... with tf.device("GPU:0"):
... self.assertEqual(tf.math.add(1.0, 2.0), 3.0)
TensorFlow official binary is built with CUDA.
"""
return _test_util.IsGoogleCudaEnabled()
def _parse_target(self, arith):
# Check for the target-augmented entry first. If no match,
# fall back to the simple name. e.g.:
# 1. 'int:gpu' , then
# 2. 'int'
g = test_util.IsGoogleCudaEnabled()
if g:
fulltype = arith + ":gpu"
else:
fulltype = arith + ":cpu"
# print("fulltype: %s" % (fulltype))
if fulltype in self._fe:
# print("Found support")
return fulltype
assert arith in self._fe, "Fatal: can't find support for arithmetic %s" % (
arith)
return arith
def is_built_with_cuda(): """Returns whether TensorFlow was built with CUDA (GPU) support.""" return _test_util.IsGoogleCudaEnabled()
