p60975123
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!15209 fix l2normalize ops parallel problem 

From: @yuanwei66
Reviewed-by: @wuxuejian,@c_34
Signed-off-by: @wuxuejian,@c_34
This commit is contained in:
mindspore-ci-bot 2021-04-15 15:13:53 +08:00 committed by Gitee
parent a492a1cd52 ee9b452b57
commit 6d2c357db8
3 changed files with 8 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
mindspore/ccsrc/backend/kernel_compiler/cpu/l2_normalize_cpu_kernel.cc
View File

@ -22,7 +22,7 @@ namespace kernel {
template <typename T> template <typename T>
void L2NormalizeCPUKernel<T>::InitKernel(const CNodePtr &kernel_node) { void L2NormalizeCPUKernel<T>::InitKernel(const CNodePtr &kernel_node) {
MS_EXCEPTION_IF_NULL(kernel_node); MS_EXCEPTION_IF_NULL(kernel_node);
epsilon_ = AnfAlgo::GetNodeAttr<float>(kernel_node, "epsilon"); epsilon_ = static_cast<T>(AnfAlgo::GetNodeAttr<float>(kernel_node, "epsilon"));
axis_ = LongToInt(AnfAlgo::GetNodeAttr<int64_t>(kernel_node, "axis")); axis_ = LongToInt(AnfAlgo::GetNodeAttr<int64_t>(kernel_node, "axis"));
input_shape_ = AnfAlgo::GetInputDeviceShape(kernel_node, 0); input_shape_ = AnfAlgo::GetInputDeviceShape(kernel_node, 0);
output_shape_ = AnfAlgo::GetOutputInferShape(kernel_node, 0); output_shape_ = AnfAlgo::GetOutputInferShape(kernel_node, 0);
@ -35,9 +35,6 @@ void L2NormalizeCPUKernel<T>::InitKernel(const CNodePtr &kernel_node) {
template <typename T> template <typename T>
void L2NormalizeCPUKernel<T>::CalcDenominator(const T *input_addr, const size_t reduce_size, const int dims, void L2NormalizeCPUKernel<T>::CalcDenominator(const T *input_addr, const size_t reduce_size, const int dims,
std::unique_ptr<T[]> *denominator_addr) { std::unique_ptr<T[]> *denominator_addr) {
T temp = (T)0.0;
T epsilon = (T)epsilon_;
T denominator = (T)0.0;
// Calculate transpose axes and stride // Calculate transpose axes and stride
size_t stride = 1; size_t stride = 1;
std::vector<size_t> axes(input_shape_.size()); std::vector<size_t> axes(input_shape_.size());
@ -60,6 +57,8 @@ void L2NormalizeCPUKernel<T>::CalcDenominator(const T *input_addr, const size_t
TransposeIterator tran_base_iter(std::move(transpose_shape), std::move(axes), input_shape_); TransposeIterator tran_base_iter(std::move(transpose_shape), std::move(axes), input_shape_);
auto task = [&](size_t start, size_t end) { auto task = [&](size_t start, size_t end) {
T temp = (T)0.0;
T denominator = (T)0.0;
auto iter = tran_base_iter; auto iter = tran_base_iter;
iter.SetPos(start * stride); iter.SetPos(start * stride);
for (size_t i = start; i < end; ++i) { for (size_t i = start; i < end; ++i) {
@ -71,7 +70,7 @@ void L2NormalizeCPUKernel<T>::CalcDenominator(const T *input_addr, const size_t
denominator += temp * temp; denominator += temp * temp;
iter.GenNextPos(); iter.GenNextPos();
} }
denominator = (denominator > epsilon) ? denominator : epsilon; denominator = (denominator > epsilon_) ? denominator : epsilon_;
(*denominator_addr)[i] = sqrt(denominator); (*denominator_addr)[i] = sqrt(denominator);
} }
}; };
@ -146,7 +145,7 @@ void L2NormalizeCPUKernel<T>::CheckParam(const CNodePtr &kernel_node) {
if (axis_ < -dims || axis_ >= dims) { if (axis_ < -dims || axis_ >= dims) {
MS_LOG(EXCEPTION) << "Attr axis_ " << axis_ << " must be in " << -dims << "~" << dims; MS_LOG(EXCEPTION) << "Attr axis_ " << axis_ << " must be in " << -dims << "~" << dims;
} }
if (epsilon_ == 0.0) { if (epsilon_ == (T)0.0) {
MS_LOG(EXCEPTION) << "Attr epsilon can not be zero."; MS_LOG(EXCEPTION) << "Attr epsilon can not be zero.";
} }
} }

2
mindspore/ccsrc/backend/kernel_compiler/cpu/l2_normalize_cpu_kernel.h
View File

@ -45,7 +45,7 @@ class L2NormalizeCPUKernel : public CPUKernel {
private: private:
std::vector<size_t> input_shape_; std::vector<size_t> input_shape_;
std::vector<size_t> output_shape_; std::vector<size_t> output_shape_;
float epsilon_; T epsilon_;
int axis_; int axis_;
void CheckParam(const CNodePtr &kernel_node); void CheckParam(const CNodePtr &kernel_node);
}; };

4
tests/st/ops/cpu/test_l2normalize_op.py
View File

@ -37,10 +37,10 @@ class Net(Cell):
@pytest.mark.platform_x86_cpu_training @pytest.mark.platform_x86_cpu_training
@pytest.mark.env_onecard @pytest.mark.env_onecard
def test_l2normalize_float32(): def test_l2normalize_float32():
x = np.arange(96).astype(np.float32).reshape(2, 3, 4, 4) x = np.arange(20*20*20*20).astype(np.float32).reshape(20, 20, 20, 20)
expect = x / np.sqrt(np.sum(x**2, axis=0, keepdims=True)) expect = x / np.sqrt(np.sum(x**2, axis=0, keepdims=True))
x = Tensor(x) x = Tensor(x)
error = np.ones(shape=[2, 3, 4, 4]) * 1.0e-5 error = np.ones(shape=[20, 20, 20, 20]) * 1.0e-5
norm_op = Net(axis=0) norm_op = Net(axis=0)
output = norm_op(x) output = norm_op(x)