Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix precision of momentum ops #36452

Merged
merged 1 commit into from
Oct 15, 2021
Merged

Fix precision of momentum ops #36452

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
67 changes: 35 additions & 32 deletions paddle/fluid/operators/optimizers/momentum_op.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,14 +173,15 @@ class CPUDenseMomentumFunctor {
}
};

template <typename T, typename MT, typename UpdateMethod>
template <typename T, typename MT, RegularizationType kRegType,
typename UpdateMethod>
class DenseMomentumFunctor;

// NOTE(dzh) for performance.
// avoid if/else in inside kernel, implement GPU UseNesterov/NoNesterov as two
// functor.
template <typename T, typename MT>
class DenseMomentumFunctor<T, MT, UseNesterov> {
template <typename T, typename MT, RegularizationType kRegType>
class DenseMomentumFunctor<T, MT, kRegType, UseNesterov> {
private:
const T* param_;
const T* grad_;
Expand All @@ -193,15 +194,13 @@ class DenseMomentumFunctor<T, MT, UseNesterov> {
T* param_out_;
MT* velocity_out_;
MT* master_param_out_;
const RegularizationType regularization_flag_;
const MT regularization_coeff_;

public:
DenseMomentumFunctor(const T* param, const T* grad, const MT* velocity,
const MultiPrecisionType<MT>* learning_rate,
const MT* master_param, const MT mu,
const MT rescale_grad, const int64_t num,
const RegularizationType regularization_flag,
const MT regularization_coeff, T* param_out,
MT* velocity_out, MT* master_param_out)
: param_(param),
Expand All @@ -215,7 +214,6 @@ class DenseMomentumFunctor<T, MT, UseNesterov> {
param_out_(param_out),
velocity_out_(velocity_out),
master_param_out_(master_param_out),
regularization_flag_(regularization_flag),
regularization_coeff_(regularization_coeff) {}
inline HOSTDEVICE void operator()(size_t i) const {
// put memory access in register
Expand All @@ -225,9 +223,9 @@ class DenseMomentumFunctor<T, MT, UseNesterov> {
const MT lr = static_cast<MT>(lr_[0]);
const MT velocity = velocity_[i];

grad = regularization_flag_ == RegularizationType::kL2DECAY
? grad + regularization_coeff_ * param
: grad;
if (kRegType == RegularizationType::kL2DECAY) {
grad += regularization_coeff_ * param;
}

MT velocity_out = velocity * mu_ + grad;
MT param_out = param - (grad + velocity_out * mu_) * lr;
Expand All @@ -240,8 +238,8 @@ class DenseMomentumFunctor<T, MT, UseNesterov> {
}
};

template <typename T, typename MT>
class DenseMomentumFunctor<T, MT, NoNesterov> {
template <typename T, typename MT, RegularizationType kRegType>
class DenseMomentumFunctor<T, MT, kRegType, NoNesterov> {
private:
const T* param_;
const T* grad_;
Expand All @@ -254,15 +252,13 @@ class DenseMomentumFunctor<T, MT, NoNesterov> {
T* param_out_;
MT* velocity_out_;
MT* master_param_out_;
const RegularizationType regularization_flag_;
const MT regularization_coeff_;

public:
DenseMomentumFunctor(const T* param, const T* grad, const MT* velocity,
const MultiPrecisionType<MT>* learning_rate,
const MT* master_param, const MT mu,
const MT rescale_grad, const int64_t num,
const RegularizationType regularization_flag,
const MT regularization_coeff, T* param_out,
MT* velocity_out, MT* master_param_out)
: param_(param),
Expand All @@ -276,7 +272,6 @@ class DenseMomentumFunctor<T, MT, NoNesterov> {
param_out_(param_out),
velocity_out_(velocity_out),
master_param_out_(master_param_out),
regularization_flag_(regularization_flag),
regularization_coeff_(regularization_coeff) {}
inline HOSTDEVICE void operator()(size_t i) const {
// put memory access in register
Expand All @@ -286,9 +281,9 @@ class DenseMomentumFunctor<T, MT, NoNesterov> {
const MT lr = static_cast<MT>(lr_[0]);
const MT velocity = velocity_[i];

grad = regularization_flag_ == RegularizationType::kL2DECAY
? grad + regularization_coeff_ * param
: grad;
if (kRegType == RegularizationType::kL2DECAY) {
grad += regularization_coeff_ * param;
}

MT velocity_out = velocity * mu_ + grad;
MT param_out = param - lr * velocity_out;
Expand Down Expand Up @@ -522,23 +517,31 @@ class MomentumOpKernel : public framework::OpKernel<T> {
platform::ForRange<DeviceContext> for_range(
static_cast<const DeviceContext&>(ctx.device_context()),
param->numel());
if (use_nesterov) {
DenseMomentumFunctor<T, MT, UseNesterov> functor(
param->data<T>(), grad->data<T>(), velocity->data<MT>(),
learning_rate->data<MPDType>(), master_in_data, mu, rescale_grad,
param->numel(), regularization_flag, regularization_coeff,
param_out->mutable_data<T>(ctx.GetPlace()),
velocity_out->mutable_data<MT>(ctx.GetPlace()), master_out_data);
for_range(functor);
#define PADDLE_LAUNCH_DENSE_MOMENTUM_KERNEL(__nesterov, __reg_type) \
DenseMomentumFunctor<T, MT, __reg_type, __nesterov> functor( \
param->data<T>(), grad->data<T>(), velocity->data<MT>(), \
learning_rate->data<MPDType>(), master_in_data, mu, rescale_grad, \
param->numel(), regularization_coeff, \
param_out->mutable_data<T>(ctx.GetPlace()), \
velocity_out->mutable_data<MT>(ctx.GetPlace()), master_out_data); \
for_range(functor);

if (use_nesterov) {
if (regularization_flag == RegularizationType::kL2DECAY) {
PADDLE_LAUNCH_DENSE_MOMENTUM_KERNEL(UseNesterov,
RegularizationType::kL2DECAY);
} else {
PADDLE_LAUNCH_DENSE_MOMENTUM_KERNEL(UseNesterov,
RegularizationType::kNONE);
}
} else {
DenseMomentumFunctor<T, MT, NoNesterov> functor(
param->data<T>(), grad->data<T>(), velocity->data<MT>(),
learning_rate->data<MPDType>(), master_in_data, mu, rescale_grad,
param->numel(), regularization_flag, regularization_coeff,
param_out->mutable_data<T>(ctx.GetPlace()),
velocity_out->mutable_data<MT>(ctx.GetPlace()), master_out_data);
for_range(functor);
if (regularization_flag == RegularizationType::kL2DECAY) {
PADDLE_LAUNCH_DENSE_MOMENTUM_KERNEL(NoNesterov,
RegularizationType::kL2DECAY);
} else {
PADDLE_LAUNCH_DENSE_MOMENTUM_KERNEL(NoNesterov,
RegularizationType::kNONE);
}
}
}

Expand Down
9 changes: 6 additions & 3 deletions python/paddle/fluid/tests/unittests/test_merged_momentum_op.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,7 +102,7 @@ def run_momentum_op(params,
'Param': p,
'Grad': g,
'Velocity': v,
'LearningRate': lr_var
'LearningRate': lr_var,
}
outputs = {'ParamOut': p, 'VelocityOut': v}
if multi_precision:
Expand All @@ -115,7 +115,7 @@ def run_momentum_op(params,
'Param': param_vars,
'Grad': grad_vars,
'Velocity': velocity_vars,
'LearningRate': lr_var
'LearningRate': lr_var,
}
outputs = {'ParamOut': param_vars, 'VelocityOut': velocity_vars}
if multi_precision:
Expand Down Expand Up @@ -176,7 +176,10 @@ def run_op(use_merged):
outs2 = run_op(False)
self.assertEqual(len(outs1), len(outs2))
for i, (out1, out2) in enumerate(zip(outs1, outs2)):
self.assertTrue(np.allclose(out1, out2, atol=1e-7))
if isinstance(place, paddle.CUDAPlace):
self.assertTrue(np.array_equal(out1, out2))
else:
self.assertTrue(np.allclose(out1, out2, atol=1e-7))

def get_places(self):
places = [paddle.CPUPlace()]
Expand Down