Skip to content

P
pytensor
提交 31a6c527 authored 作者: carriepl's avatar carriepl
浏览文件
操作

Merge pull request #2137 from aalmah/ticket_1356

fixing #1356
上级 9da67d20 8e8f4bbc
隐藏空白字符变更
内嵌 并排
正在显示 包含 180 行增加 和 227 行删除
doc/tutorial/using_gpu_solution_1.py
浏览文件 @ 31a6c527
...@@ -4,12 +4,7 @@ ...@@ -4,12 +4,7 @@
# 1. Raw results # 1. Raw results
#
# same code as in mode_solution_1 but run with following command lines:
# THEANO_FLAGS=mode=FAST_RUN,device=gpu time python program_name.py
# THEANO_FLAGS=mode=FAST_RUN,device=cpu time python program_name.py
# for GPU and CPU respectively
# typical time: 20 sec (CPU), 10 sec (GPU)
import numpy import numpy
import theano import theano
...@@ -28,8 +23,8 @@ rng.randint(size=N, low=0, high=2).astype(theano.config.floatX)) ...@@ -28,8 +23,8 @@ rng.randint(size=N, low=0, high=2).astype(theano.config.floatX))
training_steps = 10000 training_steps = 10000
# Declare Theano symbolic variables # Declare Theano symbolic variables
x = tt.matrix("x") x = theano.shared(D[0], name="x")
y = tt.vector("y") y = theano.shared(D[1], name="y")
w = theano.shared(rng.randn(feats).astype(theano.config.floatX), name="w") w = theano.shared(rng.randn(feats).astype(theano.config.floatX), name="w")
b = theano.shared(numpy.asarray(0., dtype=theano.config.floatX), name="b") b = theano.shared(numpy.asarray(0., dtype=theano.config.floatX), name="b")
x.tag.test_value = D[0] x.tag.test_value = D[0]
...@@ -58,11 +53,11 @@ predict = theano.function(inputs=[x], outputs=Out(theano.sandbox.cuda.basic_ops. ...@@ -58,11 +53,11 @@ predict = theano.function(inputs=[x], outputs=Out(theano.sandbox.cuda.basic_ops.
# Compile expressions to functions # Compile expressions to functions
train = theano.function( train = theano.function(
inputs=[x, y], inputs=[],
outputs=[prediction, xent], outputs=[prediction, xent],
updates={w: w - 0.01 * gw, b: b - 0.01 * gb}, updates={w: w - 0.01 * gw, b: b - 0.01 * gb},
name="train") name="train")
predict = theano.function(inputs=[x], outputs=prediction, predict = theano.function(inputs=[], outputs=prediction,
name="predict") name="predict")
if any([x.op.__class__.__name__ in ['Gemv', 'CGemv', 'Gemm', 'CGemm'] for x in if any([x.op.__class__.__name__ in ['Gemv', 'CGemv', 'Gemm', 'CGemm'] for x in
...@@ -76,7 +71,7 @@ else: ...@@ -76,7 +71,7 @@ else:
print train.maker.fgraph.toposort() print train.maker.fgraph.toposort()
for i in range(training_steps): for i in range(training_steps):
pred, err = train(D[0], D[1]) pred, err = train()
#print "Final model:" #print "Final model:"
#print w.get_value(), b.get_value() #print w.get_value(), b.get_value()
...@@ -84,241 +79,199 @@ print "target values for D" ...@@ -84,241 +79,199 @@ print "target values for D"
print D[1] print D[1]
print "prediction on D" print "prediction on D"
print predict(D[0]) print predict()
""" """
# 2. Profiling # 2. Profiling
#
# same code as above but run with following command lines:
# THEANO_FLAGS=mode=ProfileMode,device=gpu python program_name.py
# THEANO_FLAGS=mode=ProfileMode,device=cpu python program_name.py
# for GPU and CPU
# 2.1 Profiling output for CPU computations # 2.1 Profiling for CPU computations
# In your terminal, type:
$ THEANO_FLAGS=profile=True,device=cpu python using_gpu_solution_1.py
$ THEANO_FLAGS=mode=ProfileMode,device=cpu python program_name.py # You'll see first the output of the script:
Used the cpu
target values for D
prediction on D
Used the cpu Used the cpu
target values for D target values for D
prediction on D prediction on D
ProfileMode.print_summary() # Followed by the output of profiling.. You'll see profiling results for each function
--------------------------- # in the script, followed by a summary for all functions.
# We'll show here only the summary:
Time since import 12.586s
Theano compile time: 0.000s (0.0% since import) Results were produced using an Intel(R) Core(TM) i7-4820K CPU @ 3.70GHz
Optimization time: 0.000s
Linker time: 0.000s Function profiling
Theano fct call 5.147s (40.9% since import) ==================
Theano Op time 3.595s 28.6%(since import) 69.8%(of fct call) Message: Sum of all(3) printed profiles at exit excluding Scan op profile.
Theano function overhead in ProfileMode 1.552s 12.3%(since import) 30.2%(of fct call) Time in 10002 calls to Function.__call__: 1.590916e+00s
20002 Theano fct call, 0.000s per call Time in Function.fn.__call__: 1.492365e+00s (93.805%)
Rest of the time since import 7.440s 59.1% Time in thunks: 1.408159e+00s (88.512%)
Total compile time: 6.309664e+00s
Theano fct summary: Number of Apply nodes: 25
<% total fct time> <total time> <time per call> <nb call> <fct name> Theano Optimizer time: 4.848340e-01s
49.9% 2.567s 2.57e-04s 10000 train Theano validate time: 5.454302e-03s
0.0% 0.000s 1.24e-04s 1 predict Theano Linker time (includes C, CUDA code generation/compiling): 5.691789e+00s
0.0% 0.000s 1.26e-04s 1 predict
50.1% 2.579s 2.58e-04s 10000 train Class
---
Single Op-wise summary: <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Class name>
<% of local_time spent on this kind of Op> <cumulative %> <self seconds> <cumulative seconds> <time per call> [*] <nb_call> <nb_op> <nb_apply> <Op name> 59.6% 59.6% 0.839s 4.19e-05s C 20001 3 theano.tensor.blas_c.CGemv
59.3% 59.3% 2.133s 2.133s 5.33e-05s * 40002 1 6 <class 'theano.tensor.blas_c.CGemv'> 30.1% 89.7% 0.424s 4.71e-06s C 90001 10 theano.tensor.elemwise.Elemwise
34.4% 93.8% 1.238s 3.371s 6.19e-06s * 200002 11 22 <class 'theano.tensor.elemwise.Elemwise'> 5.5% 95.2% 0.078s 7.79e-02s Py 1 1 theano.tensor.blas.Gemv
2.8% 96.6% 0.100s 3.471s 2.51e-06s * 40002 1 6 <class 'theano.tensor.basic.Alloc'> 1.9% 97.1% 0.026s 1.30e-06s C 20001 3 theano.tensor.basic.Alloc
2.1% 98.7% 0.075s 3.546s 1.26e-06s * 60002 2 8 <class 'theano.tensor.elemwise.DimShuffle'> 1.3% 98.4% 0.018s 1.85e-06s C 10000 1 theano.tensor.elemwise.Sum
0.7% 99.3% 0.024s 3.571s 6.11e-07s * 40002 1 6 <class 'theano.tensor.opt.Shape_i'> 1.0% 99.4% 0.014s 4.78e-07s C 30001 4 theano.tensor.elemwise.DimShuffle
0.7% 100.0% 0.024s 3.595s 1.18e-06s * 20000 1 2 <class 'theano.tensor.elemwise.Sum'> 0.6% 100.0% 0.008s 4.23e-07s C 20001 3 theano.compile.ops.Shape_i
... (remaining 0 single Op account for 0.00%(0.00s) of the runtime) ... (remaining 0 Classes account for 0.00%(0.00s) of the runtime)
(*) Op is running a c implementation
Ops
Op-wise summary: ---
<% of local_time spent on this kind of Op> <cumulative %> <self seconds> <cumulative seconds> <time per call> [*] <nb_call> <nb apply> <Op name> <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Op name>
59.3% 59.3% 2.133s 2.133s 5.33e-05s * 40002 6 CGemv{inplace} 59.6% 59.6% 0.839s 4.19e-05s C 20001 3 CGemv{inplace}
18.1% 77.4% 0.650s 2.783s 3.25e-05s * 20000 2 Elemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]}} 15.8% 75.4% 0.223s 2.23e-05s C 10000 1 Elemwise{Composite{[sub(mul(i0, scalar_softplus(i1)), mul(i2, i3, scalar_softplus(i4)))]}}[(0, 4)]
6.4% 83.9% 0.231s 3.014s 1.16e-05s * 20000 2 Elemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)] 7.7% 83.1% 0.109s 1.09e-05s C 10000 1 Elemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(scalar_sigmoid(neg(i0)), i4), i5))]}}[(0, 0)]
4.0% 87.8% 0.142s 3.157s 7.11e-06s * 20000 2 Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)] 5.5% 88.7% 0.078s 7.79e-02s Py 1 1 Gemv{no_inplace}
2.8% 90.6% 0.100s 3.257s 2.51e-06s * 40002 6 Alloc 4.3% 92.9% 0.060s 6.00e-06s C 10000 1 Elemwise{Composite{[GT(scalar_sigmoid(i0), i1)]}}
1.4% 92.1% 0.052s 3.309s 1.30e-06s * 40002 6 InplaceDimShuffle{x} 1.9% 94.8% 0.026s 1.30e-06s C 20001 3 Alloc
1.1% 93.1% 0.038s 3.347s 1.92e-06s * 20000 2 Elemwise{Cast{float32}} 1.3% 96.1% 0.018s 1.85e-06s C 10000 1 Sum{acc_dtype=float64}
1.1% 94.2% 0.038s 3.386s 1.91e-06s * 20000 2 Elemwise{sub,no_inplace} 0.7% 96.8% 0.009s 4.73e-07s C 20001 3 InplaceDimShuffle{x}
1.0% 95.2% 0.036s 3.421s 1.79e-06s * 20000 2 Elemwise{gt,no_inplace} 0.6% 97.4% 0.009s 8.52e-07s C 10000 1 Elemwise{sub,no_inplace}
0.8% 96.0% 0.029s 3.450s 1.44e-06s * 20000 2 Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)] 0.6% 98.0% 0.008s 4.23e-07s C 20001 3 Shape_i{0}
0.8% 96.8% 0.028s 3.479s 1.42e-06s * 20000 2 Elemwise{neg,no_inplace} 0.5% 98.5% 0.007s 7.06e-07s C 10000 1 Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)]
0.7% 97.5% 0.024s 3.503s 6.11e-07s * 40002 6 Shape_i{0} 0.5% 98.9% 0.007s 6.57e-07s C 10000 1 Elemwise{neg,no_inplace}
0.7% 98.1% 0.024s 3.527s 1.18e-06s * 20000 2 Sum 0.3% 99.3% 0.005s 4.88e-07s C 10000 1 InplaceDimShuffle{1,0}
0.6% 98.8% 0.023s 3.550s 1.16e-06s * 20000 2 InplaceDimShuffle{1,0} 0.3% 99.5% 0.004s 3.78e-07s C 10000 1 Elemwise{inv,no_inplace}
0.6% 99.4% 0.023s 3.573s 1.15e-06s * 20000 2 Elemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)] 0.2% 99.8% 0.003s 3.44e-07s C 10000 1 Elemwise{Cast{float32}}
0.6% 100.0% 0.022s 3.595s 1.08e-06s * 20000 2 Elemwise{inv,no_inplace} 0.2% 100.0% 0.003s 3.01e-07s C 10000 1 Elemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)]
0.0% 100.0% 0.000s 3.595s 1.19e-05s * 2 2 Elemwise{Composite{[Composite{[Composite{[Composite{[GT(scalar_sigmoid(i0), i1)]}(neg(i0), i1)]}(sub(i0, i1), i2)]}(neg(i0), i1, i2)]}} 0.0% 100.0% 0.000s 8.11e-06s C 1 1 Elemwise{Composite{[GT(scalar_sigmoid(neg(sub(neg(i0), i1))), i2)]}}
... (remaining 0 Op account for 0.00%(0.00s) of the runtime) ... (remaining 0 Ops account for 0.00%(0.00s) of the runtime)
(*) Op is running a c implementation
Apply
Apply-wise summary: ------
<% of local_time spent at this position> <cumulative %%> <apply time> <cumulative seconds> <time per call> [*] <nb_call> <Apply position> <Apply Op name> <% time> <sum %> <apply time> <time per call> <#call> <id> <Apply name>
14.9% 14.9% 0.536s 0.536s 5.36e-05s * 10000 7 CGemv{inplace}(Alloc.0, TensorConstant{1.0}, x, w, TensorConstant{1.0}) 31.6% 31.6% 0.445s 4.45e-05s 10000 7 CGemv{inplace}(Alloc.0, TensorConstant{1.0}, x, w, TensorConstant{0.0})
14.9% 29.8% 0.534s 1.070s 5.34e-05s * 10000 18 CGemv{inplace}(w, TensorConstant{-0.00999999977648}, x.T, Elemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)].0, TensorConstant{0.999800026417}) 27.9% 59.6% 0.393s 3.93e-05s 10000 17 CGemv{inplace}(w, TensorConstant{-0.00999999977648}, x.T, Elemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(scalar_sigmoid(neg(i0)), i4), i5))]}}[(0, 0)].0, TensorConstant{0.999800026417})
14.8% 44.6% 0.532s 1.602s 5.32e-05s * 10000 7 CGemv{inplace}(Alloc.0, TensorConstant{1.0}, x, w, TensorConstant{1.0}) 15.8% 75.4% 0.223s 2.23e-05s 10000 14 Elemwise{Composite{[sub(mul(i0, scalar_softplus(i1)), mul(i2, i3, scalar_softplus(i4)))]}}[(0, 4)](y, Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, TensorConstant{(1,) of -1.0}, Elemwise{sub,no_inplace}.0, Elemwise{neg,no_inplace}.0)
14.7% 59.3% 0.530s 2.132s 5.30e-05s * 10000 18 CGemv{inplace}(w, TensorConstant{-0.00999999977648}, x.T, Elemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)].0, TensorConstant{0.999800026417}) 7.7% 83.1% 0.109s 1.09e-05s 10000 15 Elemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(scalar_sigmoid(neg(i0)), i4), i5))]}}[(0, 0)](Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, TensorConstant{(1,) of -1.0}, Alloc.0, y, Elemwise{sub,no_inplace}.0, Elemwise{Cast{float32}}.0)
9.1% 68.4% 0.327s 2.460s 3.27e-05s * 10000 13 Elemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]}}(y, Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{neg,no_inplace}.0) 5.5% 88.7% 0.078s 7.79e-02s 1 0 Gemv{no_inplace}(aa, TensorConstant{1.0}, xx, yy, TensorConstant{0.0})
9.0% 77.4% 0.323s 2.783s 3.23e-05s * 10000 13 Elemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]}}(y, Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{neg,no_inplace}.0) 4.3% 92.9% 0.060s 6.00e-06s 10000 13 Elemwise{Composite{[GT(scalar_sigmoid(i0), i1)]}}(Elemwise{neg,no_inplace}.0, TensorConstant{(1,) of 0.5})
3.2% 80.6% 0.116s 2.899s 1.16e-05s * 10000 16 Elemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)](Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)].0, Alloc.0, y, Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{Cast{float32}}.0) 1.3% 94.2% 0.018s 1.85e-06s 10000 16 Sum{acc_dtype=float64}(Elemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(scalar_sigmoid(neg(i0)), i4), i5))]}}[(0, 0)].0)
3.2% 83.9% 0.116s 3.014s 1.16e-05s * 10000 16 Elemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)](Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)].0, Alloc.0, y, Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, Elemwise{sub,no_inplace}.0, Elemwise{Cast{float32}}.0) 1.0% 95.2% 0.013s 1.34e-06s 10000 5 Alloc(TensorConstant{0.0}, Shape_i{0}.0)
2.0% 85.8% 0.071s 3.086s 7.12e-06s * 10000 14 Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)](Elemwise{neg,no_inplace}.0) 0.9% 96.1% 0.013s 1.27e-06s 10000 12 Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0)
2.0% 87.8% 0.071s 3.156s 7.09e-06s * 10000 14 Elemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)](Elemwise{neg,no_inplace}.0) 0.6% 96.7% 0.009s 8.52e-07s 10000 4 Elemwise{sub,no_inplace}(TensorConstant{(1,) of 1.0}, y)
0.9% 88.8% 0.034s 3.190s 3.38e-06s * 10000 12 Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0) 0.5% 97.2% 0.007s 7.06e-07s 10000 9 Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)](CGemv{inplace}.0, InplaceDimShuffle{x}.0)
0.9% 89.7% 0.034s 3.224s 3.37e-06s * 10000 12 Alloc(Elemwise{inv,no_inplace}.0, Shape_i{0}.0) 0.5% 97.6% 0.007s 6.57e-07s 10000 11 Elemwise{neg,no_inplace}(Elemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0)
0.5% 90.2% 0.019s 3.243s 1.93e-06s * 10000 8 Elemwise{Cast{float32}}(InplaceDimShuffle{x}.0) 0.4% 98.1% 0.006s 6.27e-07s 10000 0 InplaceDimShuffle{x}(b)
0.5% 90.8% 0.019s 3.262s 1.92e-06s * 10000 4 Elemwise{sub,no_inplace}(TensorConstant{(1,) of 1.0}, y) 0.4% 98.5% 0.006s 5.90e-07s 10000 1 Shape_i{0}(x)
0.5% 91.3% 0.019s 3.282s 1.90e-06s * 10000 4 Elemwise{sub,no_inplace}(TensorConstant{(1,) of 1.0}, y) 0.3% 98.9% 0.005s 4.88e-07s 10000 2 InplaceDimShuffle{1,0}(x)
... (remaining 35 Apply instances account for 8.71%(0.31s) of the runtime) 0.3% 99.1% 0.004s 3.78e-07s 10000 10 Elemwise{inv,no_inplace}(Elemwise{Cast{float32}}.0)
(*) Op is running a c implementation 0.2% 99.4% 0.003s 3.44e-07s 10000 8 Elemwise{Cast{float32}}(InplaceDimShuffle{x}.0)
0.2% 99.6% 0.003s 3.19e-07s 10000 6 InplaceDimShuffle{x}(Shape_i{0}.0)
Profile of Theano functions memory: 0.2% 99.8% 0.003s 3.01e-07s 10000 18 Elemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)](b, TensorConstant{0.00999999977648}, Sum{acc_dtype=float64}.0)
(This check only the output of each apply node. It don't check the temporary memory used by the op in the apply node.) 0.2% 100.0% 0.003s 2.56e-07s 10000 3 Shape_i{0}(y)
We skipped 4 theano function(s). Each of them used less then 1024B(theano flags ProfileMode.min_memory_size) of total intermediate memory size ... (remaining 5 Apply instances account for 0.00%(0.00s) of the runtime)
Here are tips to potentially make your code run faster
(if you think of new ones, suggest them on the mailing list).
Test them first, as they are not guaranteed to always provide a speedup. # 2.2 Profiling for GPU computations
Sorry, no tip for today.
# In your terminal, type:
$ CUDA_LAUNCH_BLOCKING=1 THEANO_FLAGS=profile=True,device=gpu python using_gpu_solution_1.py
# 2.2 Profiling output for GPU computations
# You'll see first the output of the script:
$ THEANO_FLAGS=mode=ProfileMode,device=gpu python program_name.py
Using gpu device 0: GeForce GTX 580
Used the gpu
target values for D
prediction on D
Used the gpu Used the gpu
target values for D target values for D
prediction on D prediction on D
ProfileMode.print_summary() Results were produced using a GeForce GTX TITAN
---------------------------
# Profiling summary for all functions:
Time since import 25.682s
Theano compile time: 0.000s (0.0% since import) Function profiling
Optimization time: 0.000s ==================
Linker time: 0.000s Message: Sum of all(3) printed profiles at exit excluding Scan op profile.
Theano fct call 17.052s (66.4% since import) Time in 10002 calls to Function.__call__: 3.535239e+00s
Theano Op time 14.548s 56.6%(since import) 85.3%(of fct call) Time in Function.fn.__call__: 3.420863e+00s (96.765%)
Theano function overhead in ProfileMode 2.505s 9.8%(since import) 14.7%(of fct call) Time in thunks: 2.865905e+00s (81.067%)
20002 Theano fct call, 0.001s per call Total compile time: 4.728150e-01s
Rest of the time since import 8.630s 33.6% Number of Apply nodes: 36
Theano Optimizer time: 4.283385e-01s
Theano fct summary: Theano validate time: 7.687330e-03s
<% total fct time> <total time> <time per call> <nb call> <fct name> Theano Linker time (includes C, CUDA code generation/compiling): 2.801418e-02s
50.0% 8.526s 8.53e-04s 10000 train
0.0% 0.001s 1.09e-03s 1 predict Class
50.0% 8.524s 8.52e-04s 10000 train ---
0.0% 0.001s 1.10e-03s 1 predict <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Class name>
45.7% 45.7% 1.308s 1.64e-05s C 80001 9 theano.sandbox.cuda.basic_ops.GpuElemwise
Single Op-wise summary: 17.2% 62.8% 0.492s 2.46e-05s C 20002 4 theano.sandbox.cuda.blas.GpuGemv
<% of local_time spent on this kind of Op> <cumulative %> <self seconds> <cumulative seconds> <time per call> [*] <nb_call> <nb_op> <nb_apply> <Op name> 15.1% 77.9% 0.433s 2.17e-05s C 20001 3 theano.sandbox.cuda.basic_ops.GpuAlloc
54.8% 54.8% 7.968s 7.968s 1.33e-04s 60002 1 8 <class 'theano.sandbox.cuda.basic_ops.GpuFromHost'> 8.2% 86.1% 0.234s 1.17e-05s C 20002 4 theano.sandbox.cuda.basic_ops.HostFromGpu
16.2% 71.0% 2.358s 10.325s 1.47e-05s * 160002 9 18 <class 'theano.sandbox.cuda.basic_ops.GpuElemwise'> 7.2% 93.3% 0.207s 2.07e-05s C 10000 1 theano.sandbox.cuda.basic_ops.GpuCAReduce
12.3% 83.3% 1.795s 12.120s 4.49e-05s * 40002 1 6 <class 'theano.sandbox.cuda.blas.GpuGemv'> 4.4% 97.7% 0.127s 1.27e-05s C 10003 4 theano.sandbox.cuda.basic_ops.GpuFromHost
7.0% 90.4% 1.024s 13.144s 2.56e-05s 40002 1 6 <class 'theano.sandbox.cuda.basic_ops.HostFromGpu'> 0.9% 98.6% 0.025s 8.23e-07s C 30001 4 theano.sandbox.cuda.basic_ops.GpuDimShuffle
5.0% 95.4% 0.728s 13.872s 1.82e-05s * 40002 1 6 <class 'theano.sandbox.cuda.basic_ops.GpuAlloc'> 0.7% 99.3% 0.020s 9.88e-07s C 20001 3 theano.tensor.elemwise.Elemwise
2.1% 97.4% 0.300s 14.171s 1.50e-05s * 20000 1 2 <class 'theano.sandbox.cuda.basic_ops.GpuSum'> 0.5% 99.8% 0.014s 7.18e-07s C 20001 3 theano.compile.ops.Shape_i
1.3% 98.7% 0.189s 14.360s 3.15e-06s * 60002 3 8 <class 'theano.sandbox.cuda.basic_ops.GpuDimShuffle'> 0.2% 100.0% 0.006s 5.78e-07s C 10000 1 theano.tensor.elemwise.DimShuffle
0.6% 99.4% 0.094s 14.454s 2.35e-06s * 40002 2 6 <class 'theano.tensor.elemwise.Elemwise'> ... (remaining 0 Classes account for 0.00%(0.00s) of the runtime)
0.3% 99.7% 0.048s 14.503s 1.21e-06s * 40002 1 6 <class 'theano.tensor.opt.Shape_i'>
0.3% 100.0% 0.045s 14.548s 2.25e-06s * 20000 1 2 <class 'theano.tensor.elemwise.DimShuffle'> Ops
... (remaining 0 single Op account for 0.00%(0.00s) of the runtime) ---
(*) Op is running a c implementation <% time> <sum %> <apply time> <time per call> <type> <#call> <#apply> <Op name>
17.2% 17.2% 0.492s 2.46e-05s C 20001 3 GpuGemv{inplace}
Op-wise summary: 8.2% 25.3% 0.234s 1.17e-05s C 20002 4 HostFromGpu
<% of local_time spent on this kind of Op> <cumulative %> <self seconds> <cumulative seconds> <time per call> [*] <nb_call> <nb apply> <Op name> 8.0% 33.3% 0.228s 2.28e-05s C 10001 2 GpuAlloc{memset_0=True}
54.8% 54.8% 7.968s 7.968s 1.33e-04s 60002 8 GpuFromHost 7.4% 40.7% 0.211s 2.11e-05s C 10000 1 GpuElemwise{Composite{[sub(mul(i0, scalar_softplus(i1)), mul(i2, i3, scalar_softplus(i4)))]},no_inplace}
12.3% 67.1% 1.795s 9.763s 4.49e-05s * 40002 6 GpuGemv{inplace} 7.2% 47.9% 0.207s 2.07e-05s C 10000 1 GpuCAReduce{add}{1}
7.0% 74.1% 1.024s 10.786s 2.56e-05s 40002 6 HostFromGpu 7.1% 55.0% 0.205s 2.05e-05s C 10000 1 GpuAlloc
5.0% 79.1% 0.728s 11.514s 1.82e-05s * 40002 6 GpuAlloc 6.9% 62.0% 0.198s 1.98e-05s C 10000 1 GpuElemwise{sub,no_inplace}
2.3% 81.4% 0.334s 11.848s 1.67e-05s * 20000 2 GpuElemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)] 6.9% 68.9% 0.198s 1.98e-05s C 10000 1 GpuElemwise{inv,no_inplace}
2.2% 83.6% 0.319s 12.167s 1.59e-05s * 20000 2 GpuElemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]},no_inplace} 6.2% 75.1% 0.178s 1.78e-05s C 10000 1 GpuElemwise{neg,no_inplace}
2.1% 85.7% 0.301s 12.468s 1.50e-05s * 20000 2 GpuElemwise{neg,no_inplace} 5.6% 80.6% 0.159s 1.59e-05s C 10000 1 GpuElemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(i4, i5), i6))]}}[(0, 0)]
2.1% 87.8% 0.300s 12.768s 1.50e-05s * 20000 2 GpuSum{1} 4.4% 85.1% 0.127s 1.27e-05s C 10003 4 GpuFromHost
2.0% 89.8% 0.292s 13.060s 1.46e-05s * 20000 2 GpuElemwise{inv,no_inplace} 4.3% 89.4% 0.124s 1.24e-05s C 10000 1 GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)]
1.9% 91.7% 0.283s 13.343s 1.42e-05s * 20000 2 GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)] 4.2% 93.6% 0.121s 1.21e-05s C 10000 1 GpuElemwise{ScalarSigmoid}[(0, 0)]
1.9% 93.7% 0.281s 13.625s 1.41e-05s * 20000 2 GpuElemwise{sub,no_inplace} 4.2% 97.7% 0.119s 1.19e-05s C 10000 1 GpuElemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)]
1.9% 95.5% 0.273s 13.898s 1.37e-05s * 20000 2 GpuElemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)] 0.5% 98.2% 0.014s 7.18e-07s C 20001 3 Shape_i{0}
1.9% 97.4% 0.273s 14.171s 1.37e-05s * 20000 2 GpuElemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)] 0.5% 98.7% 0.013s 1.33e-06s C 10001 2 Elemwise{gt,no_inplace}
1.0% 98.4% 0.141s 14.313s 7.06e-06s * 20002 4 GpuDimShuffle{x} 0.3% 99.0% 0.010s 9.81e-07s C 10000 1 GpuDimShuffle{1,0}
0.4% 98.8% 0.057s 14.370s 2.87e-06s * 20002 4 Elemwise{gt,no_inplace} 0.3% 99.3% 0.008s 7.90e-07s C 10000 1 GpuDimShuffle{0}
0.3% 99.1% 0.048s 14.418s 1.21e-06s * 40002 6 Shape_i{0} 0.2% 99.6% 0.007s 6.97e-07s C 10001 2 GpuDimShuffle{x}
0.3% 99.4% 0.045s 14.463s 2.25e-06s * 20000 2 InplaceDimShuffle{x} 0.2% 99.8% 0.006s 6.50e-07s C 10000 1 Elemwise{Cast{float32}}
0.3% 99.7% 0.037s 14.500s 1.83e-06s * 20000 2 Elemwise{Cast{float32}} ... (remaining 3 Ops account for 0.20%(0.01s) of the runtime)
0.2% 99.8% 0.025s 14.525s 1.24e-06s * 20000 2 GpuDimShuffle{0}
0.2% 100.0% 0.023s 14.548s 1.14e-06s * 20000 2 GpuDimShuffle{1,0} Apply
... (remaining 1 Op account for 0.00%(0.00s) of the runtime) ------
(*) Op is running a c implementation <% time> <sum %> <apply time> <time per call> <#call> <id> <Apply name>
8.8% 8.8% 0.251s 2.51e-05s 10000 22 GpuGemv{inplace}(w, TensorConstant{-0.00999999977648}, GpuDimShuffle{1,0}.0, GpuElemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(i4, i5), i6))]}}[(0, 0)].0, TensorConstant{0.999800026417})
Apply-wise summary: 8.4% 17.2% 0.241s 2.41e-05s 10000 7 GpuGemv{inplace}(GpuAlloc{memset_0=True}.0, TensorConstant{1.0}, x, w, TensorConstant{0.0})
<% of local_time spent at this position> <cumulative %%> <apply time> <cumulative seconds> <time per call> [*] <nb_call> <Apply position> <Apply Op name> 8.0% 25.1% 0.228s 2.28e-05s 10000 5 GpuAlloc{memset_0=True}(CudaNdarrayConstant{[ 0.]}, Shape_i{0}.0)
24.0% 24.0% 3.493s 3.493s 3.49e-04s 10000 1 GpuFromHost(x) 7.4% 32.5% 0.211s 2.11e-05s 10000 13 GpuElemwise{Composite{[sub(mul(i0, scalar_softplus(i1)), mul(i2, i3, scalar_softplus(i4)))]},no_inplace}(y, GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, CudaNdarrayConstant{[-1.]}, GpuElemwise{sub,no_inplace}.0, GpuElemwise{neg,no_inplace}.0)
23.9% 47.9% 3.479s 6.972s 3.48e-04s 10000 1 GpuFromHost(x) 7.2% 39.7% 0.207s 2.07e-05s 10000 21 GpuCAReduce{add}{1}(GpuElemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(i4, i5), i6))]}}[(0, 0)].0)
4.3% 52.3% 0.629s 7.602s 6.29e-05s * 10000 24 GpuGemv{inplace}(w, TensorConstant{-0.00999999977648}, GpuDimShuffle{1,0}.0, GpuElemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)].0, TensorConstant{0.999800026417}) 7.1% 46.9% 0.205s 2.05e-05s 10000 17 GpuAlloc(GpuDimShuffle{0}.0, Shape_i{0}.0)
4.3% 56.6% 0.629s 8.231s 6.29e-05s * 10000 24 GpuGemv{inplace}(w, TensorConstant{-0.00999999977648}, GpuDimShuffle{1,0}.0, GpuElemwise{Composite{[Composite{[Composite{[Composite{[mul(i0, add(i1, i2))]}(i0, neg(i1), true_div(i2, i3))]}(i0, mul(i1, i2, i3), i4, i5)]}(i0, i1, i2, exp(i3), i4, i5)]}}[(0, 0)].0, TensorConstant{0.999800026417}) 6.9% 53.8% 0.198s 1.98e-05s 10000 4 GpuElemwise{sub,no_inplace}(CudaNdarrayConstant{[ 1.]}, y)
1.8% 58.4% 0.269s 8.499s 2.69e-05s * 10000 9 GpuGemv{inplace}(GpuAlloc.0, TensorConstant{1.0}, GpuFromHost.0, w, TensorConstant{1.0}) 6.9% 60.7% 0.198s 1.98e-05s 10000 12 GpuElemwise{inv,no_inplace}(GpuFromHost.0)
1.8% 60.3% 0.268s 8.767s 2.68e-05s * 10000 9 GpuGemv{inplace}(GpuAlloc.0, TensorConstant{1.0}, GpuFromHost.0, w, TensorConstant{1.0}) 6.2% 66.9% 0.178s 1.78e-05s 10000 11 GpuElemwise{neg,no_inplace}(GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0)
1.8% 62.1% 0.266s 9.033s 2.66e-05s 10000 18 HostFromGpu(GpuElemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]},no_inplace}.0) 5.6% 72.5% 0.159s 1.59e-05s 10000 19 GpuElemwise{Composite{[add(mul(scalar_sigmoid(i0), i1, i2, i3), true_div(mul(i4, i5), i6))]}}[(0, 0)](GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)].0, CudaNdarrayConstant{[-1.]}, GpuAlloc.0, y, GpuElemwise{ScalarSigmoid}[(0, 0)].0, GpuElemwise{sub,no_inplace}.0, GpuFromHost.0)
1.8% 63.9% 0.262s 9.296s 2.62e-05s 10000 18 HostFromGpu(GpuElemwise{Composite{[Composite{[Composite{[sub(mul(i0, i1), neg(i2))]}(i0, scalar_softplus(i1), mul(i2, i3))]}(i0, i1, i2, scalar_softplus(i3))]},no_inplace}.0) 4.8% 77.3% 0.138s 1.38e-05s 10000 18 HostFromGpu(GpuElemwise{ScalarSigmoid}[(0, 0)].0)
1.8% 65.7% 0.260s 9.555s 2.60e-05s 10000 3 GpuFromHost(y) 4.4% 81.7% 0.126s 1.26e-05s 10000 10 GpuFromHost(Elemwise{Cast{float32}}.0)
1.8% 67.5% 0.258s 9.813s 2.58e-05s 10000 3 GpuFromHost(y) 4.3% 86.0% 0.124s 1.24e-05s 10000 9 GpuElemwise{Composite{[sub(neg(i0), i1)]}}[(0, 0)](GpuGemv{inplace}.0, GpuDimShuffle{x}.0)
1.7% 69.2% 0.248s 10.061s 2.48e-05s 10000 20 HostFromGpu(GpuElemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)].0) 4.2% 90.2% 0.121s 1.21e-05s 10000 15 GpuElemwise{ScalarSigmoid}[(0, 0)](GpuElemwise{neg,no_inplace}.0)
1.7% 70.9% 0.247s 10.309s 2.47e-05s 10000 20 HostFromGpu(GpuElemwise{ScalarSigmoid{output_types_preference=transfer_type{0}}}[(0, 0)].0) 4.2% 94.4% 0.119s 1.19e-05s 10000 23 GpuElemwise{Composite{[sub(i0, mul(i1, i2))]}}[(0, 0)](b, CudaNdarrayConstant{0.00999999977648}, GpuCAReduce{add}{1}.0)
1.6% 72.5% 0.238s 10.547s 2.38e-05s 10000 12 GpuFromHost(Elemwise{Cast{float32}}.0) 3.4% 97.7% 0.096s 9.61e-06s 10000 16 HostFromGpu(GpuElemwise{Composite{[sub(mul(i0, scalar_softplus(i1)), mul(i2, i3, scalar_softplus(i4)))]},no_inplace}.0)
1.6% 74.1% 0.237s 10.785s 2.37e-05s 10000 12 GpuFromHost(Elemwise{Cast{float32}}.0) 0.5% 98.2% 0.013s 1.33e-06s 10000 20 Elemwise{gt,no_inplace}(HostFromGpu.0, TensorConstant{(1,) of 0.5})
1.3% 75.4% 0.185s 10.969s 1.85e-05s * 10000 6 GpuAlloc(CudaNdarrayConstant{[ 1.58212732e-09]}, Shape_i{0}.0) 0.3% 98.5% 0.010s 9.81e-07s 10000 2 GpuDimShuffle{1,0}(x)
... (remaining 53 Apply instances account for 24.60%(3.58s) of the runtime) 0.3% 98.8% 0.008s 8.27e-07s 10000 1 Shape_i{0}(x)
(*) Op is running a c implementation 0.3% 99.1% 0.008s 7.90e-07s 10000 14 GpuDimShuffle{0}(GpuElemwise{inv,no_inplace}.0)
... (remaining 16 Apply instances account for 0.90%(0.03s) of the runtime)
Some info useful for gpu:
Spent 1.211s(8.324%) in cpu Op, 13.337s(91.676%) in gpu Op and 0.000s(0.000%) transfert Op
Theano function input that are float64
<fct name> <input name> <input type> <str input>
List of apply that don't have float64 as input but have float64 in outputs
(Useful to know if we forgot some cast when using floatX=float32 or gpu code)
<Apply> <Apply position> <fct name> <inputs type> <outputs type>
Profile of Theano functions memory:
(This check only the output of each apply node. It don't check the temporary memory used by the op in the apply node.)
We skipped 4 theano function(s). Each of them used less then 1024B(theano flags ProfileMode.min_memory_size) of total intermediate memory size
Here are tips to potentially make your code run faster
(if you think of new ones, suggest them on the mailing list).
Test them first, as they are not guaranteed to always provide a speedup.
Sorry, no tip for today.
# 3. Conclusions # 3. Conclusions
Examine and compare 'Ops' summaries for CPU and GPU. Usually GPU ops 'GpuFromHost' and 'HostFromGpu' by themselves
Facts: consume a large amount of extra time, but by making as few as possible data transfers between GPU and CPU, you can minimize their overhead.
Examine and compare 'Single Op-wise' summaries for CPU and GPU. GPU ops 'GpuFromHost' (and 'HostFromGpu') by themselves Notice that each of the GPU ops consumes more time than its CPU counterpart. This is because the ops operate on small inputs;
consume a large amount of extra time. Furthermore, notice that each of the GPU ops consumes more time than its CPU counterpart. if you increase the input data size (e.g. set N = 4000), you will see a gain from using the GPU.
An additional experiment also confirms that adding an 'out' instance in the GPU version only brings about a minor
improvement in this situation.
Tentative conclusion:
The large number of external training steps (10000) generates disproportionate GPU overhead costs.
Tentative solution:
Include the training steps inside the definition of the Theano function.
Implement this solution and put it to test.
""" """
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