gistfile1.txt

args: deepcl_unittests.exe --gtest_filter=-SLOW*
Note: Google Test filter = -SLOW*
[==========] Running 160 tests from 29 test cases.
[----------] Global test environment set-up.
[----------] 8 tests from testClBlas
[ RUN      ] testClBlas.basic
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
sgemm_Col_NN_B0_3x1x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.basic (656 ms)
[ RUN      ] testClBlas.basicthrice
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
sgemm_Col_NN_B0_3x1x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
initializing clblas
sgemm_Col_NN_B0_3x1x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
initializing clblas
sgemm_Col_NN_B0_3x1x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.basicthrice (156 ms)
[ RUN      ] testClBlas.transA
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
1 2 9 
3 7 5 
initializing clblas
sgemm_Col_NT_B0_3x1x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.transA (63 ms)
[ RUN      ] testClBlas.transB
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
3 
-1 
initializing clblas
sgemm_Col_TN_B0_3x1x2
makeGemmKernel: "sgemm_Col_TN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.transB (56 ms)
[ RUN      ] testClBlas.colMajor
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
&AWrap->getBuffer()893855619888
&BWrap->getBuffer()893855620800
&CWrap->getBuffer()893855619936
sgemm_Col_NN_B0_3x1x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.colMajor (59 ms)
[ RUN      ] testClBlas.colMajor2
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
sgemm_Col_NN_B0_4x3x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.colMajor2 (57 ms)
[ RUN      ] testClBlas.colMajorTransA
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
sgemm_Col_TN_B0_3x1x2
makeGemmKernel: "sgemm_Col_TN_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.colMajorTransA (54 ms)
[ RUN      ] testClBlas.colMajorTransB
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
sgemm_Col_NT_B0_3x1x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" now built; returning.
clblas teardown
[       OK ] testClBlas.colMajorTransB (56 ms)
[----------] 8 tests from testClBlas (1157 ms total)

[----------] 1 test from testDeepCL
[ RUN      ] testDeepCL.basic
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
expected number of output: 4
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

clblas teardown
[       OK ] testDeepCL.basic (262 ms)
[----------] 1 test from testDeepCL (262 ms total)

[----------] 23 tests from testupdateweights
[ RUN      ] testupdateweights.conv1
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

layer 0:InputLayer{ outputPlanes=2 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=5 numFilters=2 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:SquareLossLayer{}

layer 0:InputLayer{ outputPlanes=2 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=5 numFilters=2 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:SquareLossLayer{}

batchSize: 4
inputtotalsize=200 outputTotalSize=72
layer ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=5 numFilters=2 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
weightsize=36 biassize=0
statefultimer v0.7
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
layer 0:InputLayer{ outputPlanes=2 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=5 numFilters=2 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:SquareLossLayer{}
Parameters overview: (skipping 2 layers with 0 params)
layer 1: params=36	100.0%
TOTAL  : params=36
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
idx=8 predicted losschange=0.000111445 actual=0.000112534
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
idx=13 predicted losschange=-0.000886715 actual=-0.000884056
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 0ms
idx=0 predicted losschange=0.000210491 actual=0.000212669
forward try kernel 5
ForwardAuto: kernel 5: this instance cant be used: For ForwardFc, filtersize and inputimagesize must be identical
   ... not valid
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
idx=22 predicted losschange=-0.000164224 actual=-0.000163078
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 115ms
idx=22 predicted losschange=-0.000164224 actual=-0.000163078
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 0ms
   forward kernel 5: cannot be used
   forward kernel 6 time: 1ms
   forward kernel 7 time: 115ms
   forward layer selected kernel 1
idx=35 predicted losschange=-0.000391028 actual=-0.000391006
idx=26 predicted losschange=2.23142e-05 actual=2.57492e-05
idx=27 predicted losschange=9.38328e-05 actual=9.44138e-05
idx=27 predicted losschange=9.38328e-05 actual=9.44138e-05
idx=10 predicted losschange=0.00186697 actual=0.00187111
clblas teardown
[       OK ] testupdateweights.conv1 (828 ms)
[ RUN      ] testupdateweights.conv1z
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

layer 0:InputLayer{ outputPlanes=2 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=3 numFilters=2 filterSize=3 outputSize=3 padZeros=1 biased=0 skip=0} }
layer 2:SquareLossLayer{}

layer 0:InputLayer{ outputPlanes=2 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=3 numFilters=2 filterSize=3 outputSize=3 padZeros=1 biased=0 skip=0} }
layer 2:SquareLossLayer{}

batchSize: 4
inputtotalsize=72 outputTotalSize=72
layer ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=3 numFilters=2 filterSize=3 outputSize=3 padZeros=1 biased=0 skip=0} }
weightsize=36 biassize=0
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=2 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=3 numFilters=2 filterSize=3 outputSize=3 padZeros=1 biased=0 skip=0} }
layer 2:SquareLossLayer{}
Parameters overview: (skipping 2 layers with 0 params)
layer 1: params=36	100.0%
TOTAL  : params=36
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
idx=8 predicted losschange=0.00039831 actual=0.000397682
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
idx=13 predicted losschange=-0.000426502 actual=-0.000426292
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 0ms
idx=0 predicted losschange=0.000143287 actual=0.000144005
forward try kernel 5
ForwardAuto: kernel 5: this instance cant be used: For ForwardFc, padzeros must be disabled
   ... not valid
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
idx=22 predicted losschange=-1.7916e-06 actual=0
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_9x2x18
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 114ms
idx=22 predicted losschange=-1.7916e-06 actual=0
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 0ms
   forward kernel 5: cannot be used
   forward kernel 6 time: 1ms
   forward kernel 7 time: 114ms
   forward layer selected kernel 4
idx=35 predicted losschange=-2.82565e-05 actual=-2.76566e-05
idx=26 predicted losschange=3.62191e-05 actual=3.71933e-05
idx=27 predicted losschange=-0.000319862 actual=-0.000317574
idx=27 predicted losschange=-0.000319862 actual=-0.000317574
idx=10 predicted losschange=-0.000883857 actual=-0.000883102
clblas teardown
[       OK ] testupdateweights.conv1z (748 ms)
[ RUN      ] testupdateweights.numericallytest
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
layer 0:InputLayer{ outputPlanes=1 outputSize=1 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=1 numFilters=1 filterSize=1 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
layer 0:InputLayer{ outputPlanes=1 outputSize=1 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=1 numFilters=1 filterSize=1 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
 loss 0.0367983 loss2 0.0367913 change: 7.01472e-06
sumweightsdiff -0.000264842
 loss change              7.01472e-06
 estimatedLossChangeFromW 7.01413e-06
[       OK ] testupdateweights.numericallytest (529 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
layer 0:InputLayer{ outputPlanes=1 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=3 numFilters=1 filterSize=1 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=3 numFilters=1 filterSize=1 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
 loss 1.23358 loss2 1.21612 change: 0.0174605
sumweightsdiff -0.0132709
 loss change              0.0174605
 estimatedLossChangeFromW 0.0176118
[       OK ] testupdateweights.numericallytest_imagesize3 (543 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize5
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=1 outputSize=5 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=1 outputSize=5 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
 loss 4.12958 loss2 4.11952 change: 0.0100665
sumweightsdiff -0.0101708
 loss change              0.0100665
 estimatedLossChangeFromW 0.0103444
[       OK ] testupdateweights.numericallytest_imagesize5 (562 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize9
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=1 outputSize=9 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=1 outputSize=9 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=1	100.0%
TOTAL  : params=1
 loss 13.4341 loss2 13.4339 change: 0.000207901
sumweightsdiff 0.00153953
 loss change              0.000207901
 estimatedLossChangeFromW 0.000237015
[       OK ] testupdateweights.numericallytest_imagesize9 (561 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize9_filtersize9
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=9 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=81	100.0%
TOTAL  : params=81
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=9 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=81	100.0%
TOTAL  : params=81
 loss 0.135896 loss2 0.0848782 change: 0.0510182
sumweightsdiff -0.0322406
 loss change              0.0510182
 estimatedLossChangeFromW 0.0555841
[       OK ] testupdateweights.numericallytest_imagesize9_filtersize9 (596 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize9_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=3 outputSize=7 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=9 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=9 numFilters=1 filterSize=3 outputSize=7 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
 loss 7.70633 loss2 7.41581 change: 0.290529
sumweightsdiff -0.0898813
 loss change              0.290529
 estimatedLossChangeFromW 0.316231
[       OK ] testupdateweights.numericallytest_imagesize9_filtersize3 (571 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize3_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=3 numFilters=1 filterSize=3 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
layer 0:InputLayer{ outputPlanes=1 outputSize=3 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=3 numFilters=1 filterSize=3 outputSize=1 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
 loss 0.0719101 loss2 0.0694461 change: 0.00246408
sumweightsdiff -0.0110647
 loss change              0.00246408
 estimatedLossChangeFromW 0.00248372
[       OK ] testupdateweights.numericallytest_imagesize3_filtersize3 (539 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize5_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
 loss 1.20022 loss2 1.17241 change: 0.0278131
sumweightsdiff -0.0203888
 loss change              0.0278131
 estimatedLossChangeFromW 0.0280929
[       OK ] testupdateweights.numericallytest_imagesize5_filtersize3 (564 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize5_filtersize3_batchsize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=1 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=9	100.0%
TOTAL  : params=9
 loss 4.97142 loss2 4.78768 change: 0.183745
sumweightsdiff -0.056004
 loss change              0.183745
 estimatedLossChangeFromW 0.193264
[       OK ] testupdateweights.numericallytest_imagesize5_filtersize3_batchsize3 (569 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize5_filtersize3_planes3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
layer 0:InputLayer{ outputPlanes=3 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=3 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=27	100.0%
TOTAL  : params=27
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
layer 0:InputLayer{ outputPlanes=3 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=3 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=27	100.0%
TOTAL  : params=27
 loss 1.08887 loss2 0.9575 change: 0.13137
sumweightsdiff -0.00764532
 loss change              0.13137
 estimatedLossChangeFromW 0.134379
[       OK ] testupdateweights.numericallytest_imagesize5_filtersize3_planes3 (636 ms)
[ RUN      ] testupdateweights.numericallytest_imagesize5_filtersize3_planes3_batchsize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
layer 0:InputLayer{ outputPlanes=3 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=3 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=27	100.0%
TOTAL  : params=27
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
layer 0:InputLayer{ outputPlanes=3 outputSize=5 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=3 inputSize=5 numFilters=1 filterSize=3 outputSize=3 padZeros=0 biased=0 skip=0} }
layer 2:ActivationLayer{ TANH }
layer 3:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=27	100.0%
TOTAL  : params=27
 loss 4.76631 loss2 4.18154 change: 0.584769
sumweightsdiff 0.029606
 loss change              0.584769
 estimatedLossChangeFromW 0.620442
[       OK ] testupdateweights.numericallytest_imagesize5_filtersize3_planes3_batchsize3 (615 ms)
[ RUN      ] testupdateweights.backprop_weights_2
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2 (52 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize2
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=2 -D gInputSizeSquared=4 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=2 -D gOutputSizeSquared=4 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=2 -DgInputStripeOuterNumRows=2 -DgInputStripeInnerSize=4 -DgInputStripeOuterSize=4 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=2 -DgOutputStripeSize=4
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize2 (58 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize3_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=3 -D gInputSizeSquared=9 -D gNumFilters=1 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=3 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=9 -DgInputStripeOuterSize=21 -DgInputStripeMarginSize=6 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
mismatch for i 1
mismatch for i 2
mismatch for i 3
mismatch for i 4
mismatch for i 5
mismatch for i 6
mismatch for i 7
mismatch for i 8
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize3_filtersize3 (56 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize4_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=4 -D gInputSizeSquared=16 -D gNumFilters=1 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=2 -D gOutputSizeSquared=4 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=4 -DgInputStripeOuterNumRows=8 -DgInputStripeInnerSize=16 -DgInputStripeOuterSize=32 -DgInputStripeMarginSize=8 -DgOutputStripeNumRows=2 -DgOutputStripeSize=4
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
mismatch for i 8
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize4_filtersize3 (79 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize5_filtersize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=5 -D gInputSizeSquared=25 -D gNumFilters=1 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=3 -D gOutputSizeSquared=9 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=5 -DgInputStripeOuterNumRows=9 -DgInputStripeInnerSize=25 -DgInputStripeOuterSize=45 -DgInputStripeMarginSize=10 -DgOutputStripeNumRows=3 -DgOutputStripeSize=9
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
mismatch for i 4
mismatch for i 8
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize5_filtersize3 (73 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize3_filtersize1
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=3 -D gInputSizeSquared=9 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=3 -D gOutputSizeSquared=9 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=3 -DgInputStripeOuterNumRows=3 -DgInputStripeInnerSize=9 -DgInputStripeOuterSize=9 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=3 -DgOutputStripeSize=9
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize3_filtersize1 (65 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize16_filtersize1
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=16 -D gInputSizeSquared=256 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=16 -D gOutputSizeSquared=256 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=16 -DgInputStripeOuterNumRows=16 -DgInputStripeInnerSize=256 -DgInputStripeOuterSize=256 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=16 -DgOutputStripeSize=256
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize16_filtersize1 (108 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize17_filtersize1
LayerDimensions{ inputPlanes=1 inputSize=17 numFilters=1 filterSize=1 outputSize=17 padZeros=0 biased=0 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=17 -D gInputSizeSquared=289 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=17 -D gOutputSizeSquared=289 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=17 -DgInputStripeOuterNumRows=17 -DgInputStripeInnerSize=289 -DgInputStripeOuterSize=289 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=17 -DgOutputStripeSize=289
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize17_filtersize1 (91 ms)
[ RUN      ] testupdateweights.backprop_weights_2_upstreamimagesize17_filtersize1_moredata
expectedresult: -958.715
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=17 -D gInputSizeSquared=289 -D gNumFilters=1 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=17 -D gOutputSizeSquared=289 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=17 -DgInputStripeOuterNumRows=17 -DgInputStripeInnerSize=289 -DgInputStripeOuterSize=289 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=17 -DgOutputStripeSize=289
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

mismatch for i 0
[       OK ] testupdateweights.backprop_weights_2_upstreamimagesize17_filtersize1_moredata (88 ms)
[ RUN      ] testupdateweights.backprop_instance3_smaller2
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
numweights: 36
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=96 -D gInputSizeSquared=9216 -D gNumFilters=1 -D gFilterSize=6 -D gHalfFilterSize=3 -D gFilterSizeSquared=36 -D gNumOutputPlanes=1 -D gOutputPlanes=1 -D gOutputSize=91 -D gOutputSizeSquared=8281 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0 -DgNumStripes=8 -DgInputStripeMarginRows=5 -DgInputStripeInnerNumRows=12 -DgInputStripeOuterNumRows=22 -DgInputStripeInnerSize=1152 -DgInputStripeOuterSize=2112 -DgInputStripeMarginSize=480 -DgOutputStripeNumRows=12 -DgOutputStripeSize=1092
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

138 0 0 0 0 0 
132 0 0 0 0 0 
138 0 0 0 0 0 
138 0 0 0 0 0 
138 0 0 0 0 0 
132 0 0 0 0 0 

138 0 0 0 0 0 
132 0 0 0 0 0 
138 0 0 0 0 0 
138 0 0 0 0 0 
138 0 0 0 0 0 
132 0 0 0 0 0 

......
......
......
......
......
......

0=0 0 0 0 0 0 0 0 
1=0 0 0 0 0 0 0 0 
2=0 0 0 0 0 0 0 0 
3=0 0 0 0 0 0 0 0 
4=0 0 0 0 0 0 0 0 
5=0 0 0 0 0 0 0 0 
6=0 0 0 0 0 0 0 0 
7=0 0 0 0 0 0 0 0 
8=0 0 0 0 0 0 0 0 
9=0 0 0 0 0 0 0 0 
10=0 0 0 0 0 0 0 0 
11=0 0 0 0 0 0 0 0 

0=0 0 0 0 0 0 0 0 
1=0 0 0 0 0 0 0 0 
2=0 0 0 0 0 0 0 0 
3=0 0 0 0 0 0 0 0 
4=0 0 0 0 0 0 0 0 
5=0 0 0 0 0 0 0 0 
6=0 0 0 0 0 0 0 0 
7=0 0 0 0 0 0 0 0 
8=0 0 0 0 0 0 0 0 
9=0 0 0 0 0 0 0 0 
10=0 0 0 0 0 0 0 0 
11=0 0 0 0 0 0 0 0 
12=0 0 0 0 0 0 0 0 
13=0 0 0 0 0 0 0 0 
14=0 0 0 0 0 0 0 0 
15=0 0 0 0 0 0 0 0 
16=0 0 0 0 0 0 0 0 
17=0 0 0 0 0 0 0 0 
18=0 0 0 0 0 0 0 0 
19=0 0 0 0 0 0 0 0 
[       OK ] testupdateweights.backprop_instance3_smaller2 (162 ms)
[----------] 23 tests from testupdateweights (8695 ms total)

[----------] 17 tests from testforward
[ RUN      ] testforward.imagesize2_nopadzeros
expected number of output: 4
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

[       OK ] testforward.imagesize2_nopadzeros (248 ms)
[ RUN      ] testforward.imagesize2_padzeros
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

 checking result[0]=0 expecting: 0
 checking result[1]=0 expecting: 0
 checking result[2]=0 expecting: 0
 checking result[3]=0.2 expecting: 0.2
 checking result[4]=-0.13 expecting: -0.13
 checking result[5]=-0.15 expecting: -0.15
 checking result[6]=0 expecting: 0
 checking result[7]=0 expecting: 0
 checking result[8]=0 expecting: 0
 checking result[9]=0 expecting: 0
 checking result[10]=0 expecting: 0
 checking result[11]=0 expecting: 0
 checking result[12]=-0.55 expecting: -0.55
 checking result[13]=0.02 expecting: 0.02
 checking result[14]=0.21 expecting: 0.21
 checking result[27]=-14.3 expecting: -14.3
 checking result[28]=-9.6 expecting: -9.6
 checking result[29]=11.9 expecting: 11.9
 checking result[35]=0.46 expecting: 0.46
[       OK ] testforward.imagesize2_padzeros (96 ms)
[ RUN      ] testforward.imagesize3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

test1 ok
[       OK ] testforward.imagesize3 (94 ms)
[ RUN      ] testforward.test2
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

[       OK ] testforward.test2 (87 ms)
[ RUN      ] testforward.test3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

[       OK ] testforward.test3 (90 ms)
[ RUN      ] testforward.compare_0_1_biased_nopad
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_0_1_biased_nopad (161 ms)
[ RUN      ] testforward.compare_0_1_biased_pad
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_0_1_biased_pad (172 ms)
[ RUN      ] testforward.compare_1_n_biased_nopad
instance: 2
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
instance: 3
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
instance: 4
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
instance: 6
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
instance: 7
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_225x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" now built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" now built; returning.
sgemm_Col_NN_B0_225x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
sgemm_Col_NN_B0_225x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
sgemm_Col_NN_B0_225x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=15 padZeros=0 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_1_n_biased_nopad (1383 ms)
[ RUN      ] testforward.compare_1_n_biased_pad
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
instance: 2
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
instance: 3
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
instance: 4
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
instance: 6
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
instance: 7
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_361x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" now built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" now built; returning.
sgemm_Col_NN_B0_361x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
sgemm_Col_NN_B0_361x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
sgemm_Col_NN_B0_361x8x200
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX08" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX08" already built; returning.
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=5 outputSize=19 padZeros=1 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_1_n_biased_pad (1438 ms)
[ RUN      ] testforward.compare_1_5_biased_nopad
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=19 outputSize=1 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=8 inputSize=19 numFilters=8 filterSize=19 outputSize=1 padZeros=0 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_1_5_biased_nopad (198 ms)
[ RUN      ] testforward.compare_1_4_fcscenario
LayerDimensions{ inputPlanes=10 inputSize=24 numFilters=10 filterSize=24 outputSize=1 padZeros=0 biased=1 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=10 inputSize=24 numFilters=10 filterSize=24 outputSize=1 padZeros=0 biased=1 skip=0}
clblas teardown
[       OK ] testforward.compare_1_4_fcscenario (240 ms)
[ RUN      ] testforward.compare_break1_0_1
LayerDimensions{ inputPlanes=1 inputSize=33 numFilters=1 filterSize=1 outputSize=33 padZeros=0 biased=0 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 1
dump enabled=0
batch 0 batchsize 1
dump enabled=0
LayerDimensions{ inputPlanes=1 inputSize=33 numFilters=1 filterSize=1 outputSize=33 padZeros=0 biased=0 skip=0}
clblas teardown
[       OK ] testforward.compare_break1_0_1 (87 ms)
[ RUN      ] testforward.compare_break1_0_4
LayerDimensions{ inputPlanes=1 inputSize=33 numFilters=1 filterSize=1 outputSize=33 padZeros=0 biased=0 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 1
dump enabled=0
batch 0 batchsize 1
dump enabled=0
LayerDimensions{ inputPlanes=1 inputSize=33 numFilters=1 filterSize=1 outputSize=33 padZeros=0 biased=0 skip=0}
clblas teardown
[       OK ] testforward.compare_break1_0_4 (94 ms)
[ RUN      ] testforward.comparespecific_break2
LayerDimensions{ inputPlanes=64 inputSize=19 numFilters=64 filterSize=19 outputSize=1 padZeros=0 biased=0 skip=0}
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

batch 0 batchsize 4
dump enabled=0
batch 0 batchsize 4
dump enabled=0
LayerDimensions{ inputPlanes=64 inputSize=19 numFilters=64 filterSize=19 outputSize=1 padZeros=0 biased=0 skip=0}
clblas teardown
[       OK ] testforward.comparespecific_break2 (274 ms)
[ RUN      ] testforward.softmax
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
output[0]=0.0320586
output[1]=0.0871443
output[2]=0.643914
output[3]=0.236883
loss 0.44019
loss 3.44019
loss 2.44019
loss 1.44019
[       OK ] testforward.softmax (6 ms)
[ RUN      ] testforward.softmax_byplane
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
output[0]=0.0320586
output[1]=0.0871443
output[2]=0.643914
output[3]=0.236883
loss 0.44019
loss 3.44019
loss 2.44019
loss 1.44019
[       OK ] testforward.softmax_byplane (5 ms)
[ RUN      ] testforward.crash_from_jm
 -D gNumInputPlanes=32 -D gInputPlanes=32 -D gInputSize=28 -D gInputSizeSquared=784 -D gNumFilters=20 -D gFilterSize=28 -D gHalfFilterSize=14 -D gFilterSizeSquared=784 -D gNumOutputPlanes=20 -D gOutputPlanes=20 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

dump enabled=0
[       OK ] testforward.crash_from_jm (241 ms)
[----------] 17 tests from testforward (4915 ms total)

[----------] 2 tests from testfilehelper
[ RUN      ] testfilehelper.testfilehelper
[       OK ] testfilehelper.testfilehelper (19 ms)
[ RUN      ] testfilehelper.testreadchunk
[       OK ] testfilehelper.testreadchunk (5 ms)
[----------] 2 tests from testfilehelper (24 ms total)

[----------] 12 tests from testsimpleconvolvenet
[ RUN      ] testsimpleconvolvenet.imagesize1_planes2_filters2_unbiased_tanh
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 0.141046
 accuracy: 2/2 100%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 132ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 1ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 132ms
   calcGradWeights layer selected kernel 1
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 97ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 97ms
   forward layer selected kernel 1
loss, E, 0.0733092
 accuracy: 2/2 100%
loss, E, 0.0426809
 accuracy: 2/2 100%
loss, E, 0.0262453
 accuracy: 2/2 100%
loss, E, 0.0164245
 accuracy: 2/2 100%
loss, E, 0.0107573
 accuracy: 2/2 100%
accuracy: 2/2
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize1_planes2_filters2_unbiased_tanh (1218 ms)
[ RUN      ] testsimpleconvolvenet.imagesize1_planes2_filters2_tanh
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 0.964924
 accuracy: 1/2 50%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 2ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 0ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 223ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 2ms
   calcGradWeights kernel 3 time: 0ms
   calcGradWeights kernel 4 time: 223ms
   calcGradWeights layer selected kernel 1
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 99ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 0ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 99ms
   forward layer selected kernel 3
loss, E, 0.0057046
 accuracy: 2/2 100%
loss, E, 1.34826e-05
 accuracy: 2/2 100%
loss, E, 3.61967e-08
 accuracy: 2/2 100%
accuracy: 2/2
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize1_planes2_filters2_tanh (1301 ms)
[ RUN      ] testsimpleconvolvenet.imagesize3_n4_filtersize3_tanh
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 1.13283
 accuracy: 3/4 75%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=3 -D gInputSizeSquared=9 -D gNumFilters=2 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=3 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=9 -DgInputStripeOuterSize=21 -DgInputStripeMarginSize=6 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 0ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 224ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 0ms
   calcGradWeights kernel 4 time: 224ms
   calcGradWeights layer selected kernel 1
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
loss, E, 0.00996344
 accuracy: 4/4 100%
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 106ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 106ms
   forward layer selected kernel 1
loss, E, 4.70675e-05
 accuracy: 4/4 100%
loss, E, 4.09645e-07
 accuracy: 4/4 100%
accuracy: 4/4
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize3_n4_filtersize3_tanh (1307 ms)
[ RUN      ] testsimpleconvolvenet.imagesize1_2planes_filtersize1
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 0.751601
 accuracy: 2/2 100%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 0ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 226ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 2ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 0ms
   calcGradWeights kernel 4 time: 226ms
   calcGradWeights layer selected kernel 1
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
loss, E, 0.195916
 accuracy: 2/2 100%
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 96ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 0ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 2ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 96ms
   forward layer selected kernel 1
loss, E, 0.0679117
 accuracy: 2/2 100%
loss, E, 0.023677
 accuracy: 2/2 100%
loss, E, 0.00825563
 accuracy: 2/2 100%
loss, E, 0.00287856
 accuracy: 2/2 100%
loss, E, 0.00100369
 accuracy: 2/2 100%
loss, E, 0.000349964
 accuracy: 2/2 100%
 accuracy: 2/2 100%
accuracy: 2/2
loss, E, 0.000150648
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize1_2planes_filtersize1 (1231 ms)
[ RUN      ] testsimpleconvolvenet.imagesize3_n4_filtersize3_relu
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 1.48951
 accuracy: 2/4 50%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=3 -D gInputSizeSquared=9 -D gNumFilters=2 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=3 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=9 -DgInputStripeOuterSize=21 -DgInputStripeMarginSize=6 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 2ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 254ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 1ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 254ms
   calcGradWeights layer selected kernel 1
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
loss, E, 1.12957
 accuracy: 2/4 50%
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 100ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 0ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 2ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 100ms
   forward layer selected kernel 1
loss, E, 0.070782
 accuracy: 4/4 100%
loss, E, 0.003026
 accuracy: 4/4 100%
loss, E, 0.00021158
 accuracy: 4/4 100%
loss, E, 1.96858e-05
 accuracy: 4/4 100%
loss, E, 2.03002e-06
 accuracy: 4/4 100%
loss, E, 2.15572e-07
 accuracy: 4/4 100%
loss, E, 2.3083e-08
 accuracy: 4/4 100%
loss, E, 2.48239e-09
 accuracy: 4/4 100%
loss, E, 4.14442e-10
 accuracy: 4/4 100%
accuracy: 4/4
loss, E, 4.14442e-10
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize3_n4_filtersize3_relu (1548 ms)
[ RUN      ] testsimpleconvolvenet.imagesize3_n4_filtersize3_linear
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 0.50604
 accuracy: 4/4 100%
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 0ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=3 -D gInputSizeSquared=9 -D gNumFilters=2 -D gFilterSize=3 -D gHalfFilterSize=1 -D gFilterSizeSquared=9 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=2 -DgInputStripeInnerNumRows=3 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=9 -DgInputStripeOuterSize=21 -DgInputStripeMarginSize=6 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 0ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_9x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 336ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 1ms
   calcGradWeights kernel 2 time: 0ms
   calcGradWeights kernel 3 time: 0ms
   calcGradWeights kernel 4 time: 336ms
   calcGradWeights layer selected kernel 2
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
loss, E, 0.0565529
 accuracy: 4/4 100%
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x9
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 135ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 135ms
   forward layer selected kernel 1
loss, E, 0.00777245
 accuracy: 4/4 100%
loss, E, 0.00106831
 accuracy: 4/4 100%
loss, E, 0.000218376
 accuracy: 4/4 100%
accuracy: 4/4
loss, E, 0.000218376
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize3_n4_filtersize3_linear (1662 ms)
[ RUN      ] testsimpleconvolvenet.imagesize1_n2_2layers_unbiased
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
backward try kernel 0
  ... not plausibly optimal, skipping
backward try kernel 1
cl/backward.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

epoch 0 loss, E, 0.0559531
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
backward try kernel 2
cl/backward_cached.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 2 0ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 12ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
epoch 1 loss, E, 0.0254554
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
backward try kernel 3
   ... seems valid
sgemm_Col_NT_B0_1x2x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" now built; returning.
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NT_B0_1x2x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" already built; returning.
BackwardAuto: kernel 3 151ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=2 -D gInputPlanes=2 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
epoch 2 loss, E, 0.0172943
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 2ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 5ms
   backward kernel 0: cannot be used
   backward kernel 1 time: 0ms
   backward kernel 2 time: 0ms
   backward kernel 3 time: 151ms
   backward layer selected kernel 1
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_2x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_2x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 309ms
calcGradWeights try kernel 4
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 75ms
epoch 3 loss, E, 0.0138013
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 2ms
forward try kernel 5
cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 4ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 12ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 309ms
   calcGradWeights layer selected kernel 1
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 1ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 75ms
   calcGradWeights layer selected kernel 1
epoch 4 loss, E, 0.0115848
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 8ms
epoch 5 loss, E, 0.00987036
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 146ms
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 63ms
epoch 6 loss, E, 0.00844797
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 2ms
   forward kernel 5 time: 2ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 146ms
   forward layer selected kernel 1
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 0ms
   forward kernel 3 time: 0ms
   forward kernel 4 time: 5ms
   forward kernel 5 time: 4ms
   forward kernel 6 time: 8ms
   forward kernel 7 time: 63ms
   forward layer selected kernel 2
epoch 7 loss, E, 0.00724182
epoch 8 loss, E, 0.00621212
epoch 9 loss, E, 0.00533106
epoch 10 loss, E, 0.00457645
epoch 11 loss, E, 0.00392979
epoch 12 loss, E, 0.00337539
epoch 13 loss, E, 0.00289992
epoch 14 loss, E, 0.002492
epoch 15 loss, E, 0.00214191
epoch 16 loss, E, 0.00184138
epoch 17 loss, E, 0.00158331
epoch 18 loss, E, 0.00136164
epoch 19 loss, E, 0.0011712
epoch 20 loss, E, 0.00100754
epoch 21 loss, E, 0.000866877
epoch 22 loss, E, 0.000745946
epoch 23 loss, E, 0.000641966
epoch 24 loss, E, 0.000552543
epoch 25 loss, E, 0.000475625
epoch 26 loss, E, 0.000409454
epoch 27 loss, E, 0.000352522
epoch 28 loss, E, 0.000303531
epoch 29 loss, E, 0.00026137
epoch 30 loss, E, 0.000225082
epoch 31 loss, E, 0.000193845
epoch 32 loss, E, 0.000166954
epoch 33 loss, E, 0.000143801
epoch 34 loss, E, 0.000123866
epoch 35 loss, E, 0.000106699
epoch 36 loss, E, 9.19176e-05
epoch 37 loss, E, 7.91864e-05
epoch 38 loss, E, 6.82211e-05
epoch 39 loss, E, 5.87767e-05
layer 0:InputLayer{ outputPlanes=1 outputSize=1 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=1 numFilters=2 filterSize=1 outputSize=1 padZeros=0 biased=1 skip=0} }
layer 2:ActivationLayer{ RELU }
layer 3:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=1 numFilters=2 filterSize=1 outputSize=1 padZeros=0 biased=1 skip=0} }
layer 4:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=4	40.0%
layer 3: params=6	60.0%
TOTAL  : params=10
loss, E, 5.87767e-05
 accuracy: 2/2 100%
accuracy: 2/2
loss, E, 5.87767e-05
loss, E, 5.87767e-05
layer 0:InputLayer{ outputPlanes=1 outputSize=1 }
layer 1:ConvolutionalLayer{ LayerDimensions{ inputPlanes=1 inputSize=1 numFilters=2 filterSize=1 outputSize=1 padZeros=0 biased=1 skip=0} }
layer 2:ActivationLayer{ RELU }
layer 3:ConvolutionalLayer{ LayerDimensions{ inputPlanes=2 inputSize=1 numFilters=2 filterSize=1 outputSize=1 padZeros=0 biased=1 skip=0} }
layer 4:SquareLossLayer{}
Parameters overview: (skipping 3 layers with 0 params)
layer 1: params=4	40.0%
layer 3: params=6	60.0%
TOTAL  : params=10
float weights1[] = {-0.303866f, -1.59823f};
float weights3[] = {0.426358f, -0.719592f, -0.420361f, 0.719566f};
float bias1[] = {-0.324465f, 0.60279f};
float bias3[] = {0.506862f, -0.506837f};
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize1_n2_2layers_unbiased (3275 ms)
[ RUN      ] testsimpleconvolvenet.imagesize1_n2_2layers_biased
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 3ms
backward try kernel 0
  ... not plausibly optimal, skipping
backward try kernel 1
cl/backward.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 1.19067
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
backward try kernel 2
cl/backward_cached.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 2 0ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 7ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 0ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 9ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 2ms
backward try kernel 3
   ... seems valid
sgemm_Col_NT_B0_1x2x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" now built; returning.
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NT_B0_1x2x2
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NL016_KX01" already built; returning.
BackwardAuto: kernel 3 144ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=2 -D gInputPlanes=2 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=1 -D gInputSizeSquared=1 -D gNumFilters=2 -D gFilterSize=1 -D gHalfFilterSize=0 -D gFilterSizeSquared=1 -D gNumOutputPlanes=2 -D gOutputPlanes=2 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=0 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=0 -DgInputStripeInnerNumRows=1 -DgInputStripeOuterNumRows=1 -DgInputStripeInnerSize=1 -DgInputStripeOuterSize=1 -DgInputStripeMarginSize=0 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 0ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 0ms
   backward kernel 0: cannot be used
   backward kernel 1 time: 1ms
   backward kernel 2 time: 0ms
   backward kernel 3 time: 144ms
   backward layer selected kernel 2
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_2x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_2x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 292ms
calcGradWeights try kernel 4
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_1x2x1
makeGemmKernel: "sgemm_Col_TN_B1_ML016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 74ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 3ms
forward try kernel 5
cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 7ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 292ms
   calcGradWeights layer selected kernel 1
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 1ms
   calcGradWeights kernel 2 time: 0ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 74ms
   calcGradWeights layer selected kernel 2
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 0ms
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
loss, E, 0.0667568
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_1x2x1
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 145ms
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_1x2x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_1x2x2
makeGemmKernel: "sgemm_Col_NN_B0_ML016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 70ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 1ms
   forward kernel 2 time: 0ms
   forward kernel 3 time: 9ms
   forward kernel 4 time: 0ms
   forward kernel 5 time: 3ms
   forward kernel 6 time: 0ms
   forward kernel 7 time: 145ms
   forward layer selected kernel 2
   forward kernel 0: cannot be used
   forward kernel 1 time: 3ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 2ms
   forward kernel 4 time: 0ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 1ms
   forward kernel 7 time: 70ms
   forward layer selected kernel 4
loss, E, 0.00923595
loss, E, 0.00112611
loss, E, 0.000117399
loss, E, 1.15641e-05
dump enabled=0
loss, E, 1.78565e-06
 accuracy: 2/2 100%
accuracy: 2/2
loss, E, 1.78565e-06
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize1_n2_2layers_biased (2961 ms)
[ RUN      ] testsimpleconvolvenet.imagesize_5_4_2layers_filtersize_2_4_biased_n3
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 0ms
backward try kernel 0
  ... not plausibly optimal, skipping
backward try kernel 1
cl/backward.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 0ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 1.33412
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 0ms
backward try kernel 2
cl/backward_cached.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 2 0ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 1ms
backward try kernel 3
   ... seems valid
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" now built; returning.
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
BackwardAuto: kernel 3 105ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=3 -D gInputPlanes=3 -D gInputSize=4 -D gInputSizeSquared=16 -D gNumFilters=3 -D gFilterSize=4 -D gHalfFilterSize=2 -D gFilterSizeSquared=16 -D gNumOutputPlanes=3 -D gOutputPlanes=3 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=3 -DgInputStripeInnerNumRows=4 -DgInputStripeOuterNumRows=10 -DgInputStripeInnerSize=16 -DgInputStripeOuterSize=40 -DgInputStripeMarginSize=12 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=5 -D gInputSizeSquared=25 -D gNumFilters=3 -D gFilterSize=2 -D gHalfFilterSize=1 -D gFilterSizeSquared=4 -D gNumOutputPlanes=3 -D gOutputPlanes=3 -D gOutputSize=4 -D gOutputSizeSquared=16 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=1 -DgInputStripeInnerNumRows=5 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=25 -DgInputStripeOuterSize=35 -DgInputStripeMarginSize=5 -DgOutputStripeNumRows=4 -DgOutputStripeSize=16
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
   backward kernel 0: cannot be used
   backward kernel 1 time: 0ms
   backward kernel 2 time: 0ms
   backward kernel 3 time: 105ms
   backward layer selected kernel 1
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 231ms
calcGradWeights try kernel 4
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

kernel 140712488066176 binary 0 size 0
makeGemmKernel: "sgemm_Col_TN_B1_MX032_NX032_KX16_BRANCH_src" now built; returning.
kernel 140712488066176 binary 0 size 0
makeGemmKernel: "sgemm_Col_TN_B1_MX032_NX032_KX16_BRANCH_src" already built; returning.
kernel 140712488066176 binary 0 size 0
makeGemmKernel: "sgemm_Col_TN_B1_MX032_NX032_KX16_BRANCH_src" already built; returning.
BackpropWeightsAuto: kernel 4 259ms
forward try kernel 5
ForwardAuto: kernel 5: this instance cant be used: For ForwardFc, filtersize and inputimagesize must be identical
   ... not valid
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 1ms
forward try kernel 5
cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward_fc_wgperrow.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 5 1ms
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 231ms
   calcGradWeights layer selected kernel 1
   calcGradWeights kernel 0: cannot be used
   calcGradWeights kernel 1 time: 0ms
   calcGradWeights kernel 2 time: 1ms
   calcGradWeights kernel 3 time: 1ms
   calcGradWeights kernel 4 time: 259ms
   calcGradWeights layer selected kernel 1
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NN_B0_16x3x4
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX01" now built; returning.
sgemm_Col_NN_B0_16x3x4
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX01" already built; returning.
sgemm_Col_NN_B0_16x3x4
makeGemmKernel: "sgemm_Col_NN_B0_MX016_NL016_KX01" already built; returning.
ForwardAuto: kernel 7 104ms
forward try kernel 6
cl/forward_byinputplane.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/reduce_segments.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 6 2ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 1ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5: cannot be used
   forward kernel 6 time: 1ms
   forward kernel 7 time: 104ms
   forward layer selected kernel 1
forward try kernel 7
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

makeGemmKernel: "sgemm_Col_NN_B1_MX032_NX032_KX16_BRANCH" now built; returning.
makeGemmKernel: "sgemm_Col_NN_B1_MX032_NX032_KX16_BRANCH" already built; returning.
makeGemmKernel: "sgemm_Col_NN_B1_MX032_NX032_KX16_BRANCH" already built; returning.
ForwardAuto: kernel 7 155ms
   forward kernel 0: cannot be used
   forward kernel 1 time: 0ms
   forward kernel 2 time: 0ms
   forward kernel 3 time: 1ms
   forward kernel 4 time: 1ms
   forward kernel 5 time: 1ms
   forward kernel 6 time: 2ms
   forward kernel 7 time: 155ms
   forward layer selected kernel 1
loss, E, 0.0064664
loss, E, 1.10358e-05
loss, E, 3.56415e-08
loss, E, 1.29538e-10
loss, E, 9.03277e-13
loss, E, 2.17604e-13
loss, E, 1.48326e-13
loss, E, 1.16351e-13
loss, E, 1.09246e-13
loss, E, 1.09246e-13
 accuracy: 3/3 100%
accuracy: 3/3
loss, E, 1.09246e-13
clblas teardown
[       OK ] testsimpleconvolvenet.imagesize_5_4_2layers_filtersize_2_4_biased_n3 (7463 ms)
[ RUN      ] testsimpleconvolvenet.imagesize_5_4_2layers_filtersize_2_4_biased_n6
Using Intel(R) Corporation , OpenCL platform: Intel(R) OpenCL
Using OpenCL device: Intel(R) HD Graphics 4000
initializing clblas
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/copy.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/activate.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/applyActivationDeriv.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/per_element_add.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
forward try kernel 0
  ... not plausibly optimal, skipping
forward try kernel 1
cl/forward1.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 1 1ms
backward try kernel 0
  ... not plausibly optimal, skipping
backward try kernel 1
cl/backward.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
calcGradWeights try kernel 0
  ... not plausibly optimal, skipping
calcGradWeights try kernel 1
cl/backpropweights.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 1 1ms
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

cl/per_element_op2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

loss, E, 3.64011
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
forward try kernel 2
cl/forward2.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 2 1ms
backward try kernel 2
cl/backward_cached.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackwardAuto: kernel 2 1ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 2ms
calcGradWeights try kernel 2
cl/BackpropWeightsScratch.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 2 1ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
forward try kernel 3
cl/forward3.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 3 0ms
backward try kernel 3
   ... seems valid
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" now built; returning.
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
sgemm_Col_NT_B0_1x48x3
makeGemmKernel: "sgemm_Col_NT_B0_ML016_NX016_KX01" already built; returning.
BackwardAuto: kernel 3 100ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=3 -D gInputPlanes=3 -D gInputSize=4 -D gInputSizeSquared=16 -D gNumFilters=3 -D gFilterSize=4 -D gHalfFilterSize=2 -D gFilterSizeSquared=16 -D gNumOutputPlanes=3 -D gOutputPlanes=3 -D gOutputSize=1 -D gOutputSizeSquared=1 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=3 -DgInputStripeInnerNumRows=4 -DgInputStripeOuterNumRows=10 -DgInputStripeInnerSize=16 -DgInputStripeOuterSize=40 -DgInputStripeMarginSize=12 -DgOutputStripeNumRows=1 -DgOutputStripeSize=1
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 2ms
calcGradWeights try kernel 3
options:  -D BIASED -D gNumInputPlanes=1 -D gInputPlanes=1 -D gInputSize=5 -D gInputSizeSquared=25 -D gNumFilters=3 -D gFilterSize=2 -D gHalfFilterSize=1 -D gFilterSizeSquared=4 -D gNumOutputPlanes=3 -D gOutputPlanes=3 -D gOutputSize=4 -D gOutputSizeSquared=16 -D gPadZeros=0 -D gMargin=0 -D gEven=1 -D gSkip=0 -DgNumStripes=1 -DgInputStripeMarginRows=1 -DgInputStripeInnerNumRows=5 -DgInputStripeOuterNumRows=7 -DgInputStripeInnerSize=25 -DgInputStripeOuterSize=35 -DgInputStripeMarginSize=5 -DgOutputStripeNumRows=4 -DgOutputStripeSize=16
cl/BackpropWeightsScratchLarge.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
BackpropWeightsAuto: kernel 3 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 1ms
forward try kernel 4
cl/forward4.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

   ... seems valid
ForwardAuto: kernel 4 2ms
   backward kernel 0: cannot be used
   backward kernel 1 time: 1ms
   backward kernel 2 time: 1ms
   backward kernel 3 time: 100ms
   backward layer selected kernel 1
calcGradWeights try kernel 4
   ... seems valid
cl/per_element_op2_scalar.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" now built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
sgemm_Col_TN_B1_48x3x1
makeGemmKernel: "sgemm_Col_TN_B1_MX016_NL016_KX01" already built; returning.
BackpropWeightsAuto: kernel 4 239ms
calcGradWeights try kernel 4
   ... seems valid
ForwardIm2Col.cl build log: 
fcl build 1 succeeded.
fcl build 2 succeeded.
bcl build succeeded.

kernel 140712488066176 binary 0 size 0
makeGemmKernel: "sgemm_Col_TN_B1_MX032_NX032_KX16_BRANCH_src" already built; returning.
OpenCL error -38 on line 674
Assertion failed: false, file D:\DeepCL\clMathLibraries\clBLAS\src\library\blas\specialCases\GemmSpecialCases.cpp, line 674