microsoft
/
CNTK
зеркало из
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

replaced 6 non-linearity nodes by a single base class and 6 macros, as those 6 nodes were now structurally identical and only differed in TensorView opcodes

This commit is contained in:
Frank Seide 2016-01-01 16:38:20 -08:00
Родитель 351358eb46
Коммит e09548f391
2 изменённых файлов: 57 добавлений и 294 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

350
Source/ComputationNetworkLib/NonlinearityNodes.h
Просмотреть файл

@ -28,9 +28,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
#ifdef ENABLE_TENSORVIEW
// -----------------------------------------------------------------------
// UnaryElementWiseWithOpCodeNode (input) -- elementwise unary op where forward
// and backward are single ElementWiseOperator opcodes and only inputs (but not
// function values) are used.
// UnaryElementWiseWithOpCodeNodeBase (input) -- base for elementwise unary op
// where forward // and backward are single ElementWiseOperator opcodes and
// only inputs (but not // function values) are used.
// -----------------------------------------------------------------------
template<class ElemType, ElementWiseOperator opForward, ElementWiseOperator opBackward>
@ -76,292 +76,55 @@ namespace Microsoft { namespace MSR { namespace CNTK {
#define UnaryElementWiseWithOpCodeNodeBaseMembers UsingComputationNodeMembersBoilerplate;
// -----------------------------------------------------------------------
// SigmoidNode (input) -- sigmoid non-linearity
// SigmoidNode (input)
// TanhNode (input)
// RectifiedLinearNode (input)
// LogNode (input)
// ExpNode (input)
// CosNode (input)
// These are all implemented by single-opcode functions and can thus be declared by a macro.
// -----------------------------------------------------------------------
#if 1
template<class ElemType>
class SigmoidNode : public UnaryElementWiseWithOpCodeNodeBase<ElemType, opSigmoid, opElementwiseProductWithSigmoidDerivative>
{
typedef UnaryElementWiseWithOpCodeNodeBase<ElemType, opSigmoid, opElementwiseProductWithSigmoidDerivative> Base; UnaryElementWiseWithOpCodeNodeBaseMembers;
static const std::wstring TypeName() { return L"Sigmoid"; }
public:
DeclareConstructorFromConfigWithNumInputs(SigmoidNode);
SigmoidNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
};
#else
template<class ElemType>
class SigmoidNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"Sigmoid"; }
public:
DeclareConstructorFromConfigWithNumInputs(SigmoidNode);
SigmoidNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignSigmoidOf(input);
}
#pragma push_macro("DeclareUnaryTensorOp")
#define DeclareUnaryElementWiseWithOpCodeNode(Name, Forward, Backward) \
template<class ElemType> \
class Name ## Node : public UnaryElementWiseWithOpCodeNodeBase<ElemType, op ## Forward, op ## Backward> \
{ \
typedef UnaryElementWiseWithOpCodeNodeBase<ElemType, op ## Forward, op ## Backward> Base; UnaryElementWiseWithOpCodeNodeBaseMembers; \
static const std::wstring TypeName() { return L#Name; } \
public: \
DeclareConstructorFromConfigWithNumInputs(Name ## Node); \
Name ## Node(DEVICEID_TYPE deviceId, const wstring & Name) : \
Base(deviceId, Name) \
{ } \
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// Name Forward and Backward opcodes
DeclareUnaryElementWiseWithOpCodeNode(Sigmoid, Sigmoid, ElementwiseProductWithSigmoidDerivative);
DeclareUnaryElementWiseWithOpCodeNode(Tanh, Tanh, ElementwiseProductWithTanhDerivative);
DeclareUnaryElementWiseWithOpCodeNode(Log, Log, ElementwiseQuotient);
DeclareUnaryElementWiseWithOpCodeNode(Exp, Exp, ElementwiseProductWithExp);
DeclareUnaryElementWiseWithOpCodeNode(RectifiedLinear, LinearRectifier, ElementwiseProductWithLinearRectifierDerivative);
DeclareUnaryElementWiseWithOpCodeNode(Cosine, Cosine, ElementwiseProductWithCosDerivative);
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseProductWithSigmoidDerivativeOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
#pragma pop_macro("DeclareUnaryTensorOp")
#endif
// -----------------------------------------------------------------------
// TanhNode (input) -- sigmoid non-linearity
// -----------------------------------------------------------------------
template<class ElemType>
class TanhNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"Tanh"; }
public:
DeclareConstructorFromConfigWithNumInputs(TanhNode);
TanhNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignTanhOf(input);
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseProductWithTanhDerivativeOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
// -----------------------------------------------------------------------
// LogNode (input) -- sigmoid non-linearity
// -----------------------------------------------------------------------
template<class ElemType>
class LogNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"Log"; }
public:
DeclareConstructorFromConfigWithNumInputs(LogNode);
LogNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignLogOf(input);
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseQuotientOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
// -----------------------------------------------------------------------
// ExpNode (input) -- sigmoid non-linearity
// -----------------------------------------------------------------------
template<class ElemType>
class ExpNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"Exp"; }
public:
DeclareConstructorFromConfigWithNumInputs(ExpNode);
ExpNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignExpOf(input);
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseProductWithExpOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
// -----------------------------------------------------------------------
// LinearRectifierNode (input) -- sigmoid non-linearity
// -----------------------------------------------------------------------
template<class ElemType>
class RectifiedLinearNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"RectifiedLinear"; }
public:
DeclareConstructorFromConfigWithNumInputs(RectifiedLinearNode);
RectifiedLinearNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignLinearRectifierOf(input);
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseProductWithLinearRectifierDerivativeOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
// -----------------------------------------------------------------------
// CosineNode (input) -- sigmoid non-linearity
// -----------------------------------------------------------------------
template<class ElemType>
class CosineNode : public UnaryElementWiseNode<ElemType>
{
typedef UnaryElementWiseNode<ElemType> Base; UsingUnaryElementwiseNodeBaseMembers;
static const std::wstring TypeName() { return L"Cosine"; }
public:
DeclareConstructorFromConfigWithNumInputs(CosineNode);
CosineNode(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
virtual void /*ComputationNode::*/ForwardProp(const FrameRange & fr) override
{
size_t rank = DetermineElementwiseTensorRank();
auto result = ValueTensorFor(rank, fr);
auto input = Input(0)->ValueTensorFor(rank, fr);
result.AssignCosineOf(input);
}
virtual void /*ComputationNode::*/BackpropTo(const size_t inputIndex, const FrameRange & fr) override
{
assert(inputIndex == 0); inputIndex;
// get the args
size_t rank = DetermineElementwiseTensorRank();
auto sliceOutputGrad = GradientTensorFor(rank, fr); // propagate from this one...
auto sliceInputGrad = Input(0)->GradientTensorFor(rank, fr); // ...to this one
auto sliceInputValue = Input(0)->ValueTensorFor(rank, fr);
//auto sliceInputValue = InputUsedInComputingInputNodesGradients(0) ? Input(0)->ValueTensorFor(rank, fr) : TensorView<ElemType>();
//auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueTensorFor(rank, fr) : TensorView<ElemType>();
// do the actual operation
sliceInputGrad.AddElementwiseProductWithCosDerivativeOf(sliceOutputGrad, sliceInputValue);
}
// We don't need our output values in backprop.
virtual bool OutputUsedInComputingInputNodesGradients() const override { return false; }
};
#endif
// -----------------------------------------------------------------------
// NonlinearityNodeBase (input) -- abstract base class that holds what's shared
// between non-linearity nodes like Sigmoid
// SoftmaxNodeBase (input) -- shared base of Softmax and LogSoftmax
// -----------------------------------------------------------------------
// shared base for all element-wise non-linearities
// What this adds over a ComputationNode<ElemType> is a member m_gradientTemp for temp use by derived classes.
// TODO: This was used more broadly, but no longer, so we may be able to simplify the signatures of the virtual functions.
template<class ElemType>
class NonlinearityNodeBase : public ComputationNode<ElemType>, public NumInputs<1>
class SoftmaxNodeBase : public ComputationNode<ElemType>, public NumInputs<1>
{
typedef ComputationNode<ElemType> Base; UsingComputationNodeMembers;
public:
//virtual ComputationNodeBase * NewThis(DEVICEID_TYPE deviceId, const wstring & name) = 0;
DeclareConstructorFromConfigWithNumInputs(NonlinearityNodeBase);
NonlinearityNodeBase(DEVICEID_TYPE deviceId, const wstring & name) :
DeclareConstructorFromConfigWithNumInputs(SoftmaxNodeBase);
SoftmaxNodeBase(DEVICEID_TYPE deviceId, const wstring & name) :
Base(deviceId, name)
{ }
@ -377,7 +140,6 @@ namespace Microsoft { namespace MSR { namespace CNTK {
auto sliceOutputValue = OutputUsedInComputingInputNodesGradients() ? ValueFor(fr) : Matrix<ElemType>();
// do the actual operation
// TODO: Once all is unified then make the order of arguments more logical (in -> out)
BackpropToV(*m_gradientTemp, sliceInputValue, sliceInputGrad, sliceOutputGrad, sliceOutputValue);
}
@ -403,7 +165,7 @@ namespace Microsoft { namespace MSR { namespace CNTK {
Base::CopyTo(nodeP, newName, flags);
if (flags & CopyNodeFlags::copyNodeValue)
{
auto node = dynamic_pointer_cast<NonlinearityNodeBase<ElemType>>(nodeP);
auto node = dynamic_pointer_cast<SoftmaxNodeBase<ElemType>>(nodeP);
*node->m_gradientTemp = *m_gradientTemp;
}
}
@ -425,7 +187,7 @@ namespace Microsoft { namespace MSR { namespace CNTK {
shared_ptr<Matrix<ElemType>> m_gradientTemp;
};
#define UsingNonlinearityNodeBaseMembers UsingComputationNodeMembersBoilerplate; using Base::m_gradientTemp
#define UsingSoftmaxNodeBaseMembers UsingComputationNodeMembersBoilerplate; using Base::m_gradientTemp
#ifndef ENABLE_TENSORVIEW
// -----------------------------------------------------------------------
@ -433,9 +195,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class RectifiedLinearNode : public NonlinearityNodeBase<ElemType>
class RectifiedLinearNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"RectifiedLinear"; }
public:
DeclareConstructorFromConfigWithNumInputs(RectifiedLinearNode);
@ -479,9 +241,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class SigmoidNode : public NonlinearityNodeBase<ElemType>
class SigmoidNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Sigmoid"; }
public:
DeclareConstructorFromConfigWithNumInputs(SigmoidNode);
@ -517,9 +279,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class TanhNode : public NonlinearityNodeBase<ElemType>
class TanhNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Tanh"; }
public:
DeclareConstructorFromConfigWithNumInputs(TanhNode);
@ -557,9 +319,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class LogNode : public NonlinearityNodeBase<ElemType>
class LogNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Log"; }
public:
DeclareConstructorFromConfigWithNumInputs(LogNode);
@ -596,9 +358,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class ExpNode : public NonlinearityNodeBase<ElemType>
class ExpNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Exp"; }
public:
DeclareConstructorFromConfigWithNumInputs(ExpNode);
@ -644,9 +406,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class CosineNode : public NonlinearityNodeBase<ElemType>
class CosineNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Cosine"; }
public:
DeclareConstructorFromConfigWithNumInputs(CosineNode);
@ -685,9 +447,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
//we assume it's column-wise by default
//the derivative will increase the Matrix<ElemType> size to the power of column size and should not be used.
template<class ElemType>
class SoftmaxNode : public NonlinearityNodeBase<ElemType>
class SoftmaxNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Softmax"; }
public:
DeclareConstructorFromConfigWithNumInputs(SoftmaxNode);
@ -752,9 +514,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// -----------------------------------------------------------------------
template<class ElemType>
class LogSoftmaxNode : public NonlinearityNodeBase<ElemType>
class LogSoftmaxNode : public SoftmaxNodeBase<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"LogSoftmax"; }
public:
DeclareConstructorFromConfigWithNumInputs(LogSoftmaxNode);
@ -1325,9 +1087,9 @@ namespace Microsoft { namespace MSR { namespace CNTK {
// this node is not differentiable and so cannot be used in the backpropagation
// TODO: make function value sparse?
template<class ElemType>
class HardmaxNode : public NonlinearityNodeBase/*ComputationNode*/<ElemType>
class HardmaxNode : public SoftmaxNodeBase/*ComputationNode*/<ElemType>
{
typedef NonlinearityNodeBase<ElemType> Base; UsingNonlinearityNodeBaseMembers;
typedef SoftmaxNodeBase<ElemType> Base; UsingSoftmaxNodeBaseMembers;
static const std::wstring TypeName() { return L"Hardmax"; }
public:

1
Tests/UnitTests/MathTests/ConvolutionEngineTests.cpp
Просмотреть файл

@ -133,6 +133,7 @@ fprintf(stderr, "ConvolutionEngineTests.cpp %d\n", __LINE__);
for (int deviceId : { -1, 0 })
{
fprintf(stderr, "ConvolutionEngineTests.cpp %d\n", __LINE__);
auto fact = ConvFact::Create(deviceId, ConvFact::EngineType::Auto, ImageLayoutKind::CHW);
fprintf(stderr, "ConvolutionEngineTests.cpp %d\n", __LINE__);
auto eng = fact->CreateConvEngine(deviceId, 0);