Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 1 年 前同步成功

通知 2299
Star 20931
Fork 5422
P
Paddle
提交 6648995f 编写于 作者: T tensor-tang
浏览文件
操作

fix build

上级 74292f41
隐藏空白更改
内联 并排
Showing with 61 addition and 61 deletion
paddle/fluid/operators/crf_decoding_op.h
浏览文件 @ 6648995f
...@@ -82,8 +82,8 @@ class CRFDecodingOpKernel : public framework::OpKernel<T> { ...@@ -82,8 +82,8 @@ class CRFDecodingOpKernel : public framework::OpKernel<T> {
Tensor track; Tensor track;
int* track_value = int* track_value =
track.mutable_data<int>(emission_dims, platform::CPUPlace()); track.mutable_data<int>(emission_dims, platform::CPUPlace());
auto ker = jit::Get<jit::crfdecoding, jit::CRFDecoding, platform::CPUPlace>( auto ker = jit::Get<jit::crfdecoding, jit::CRFDecodingTuples<T>,
tag_num); platform::CPUPlace>(tag_num);
ker(static_cast<int>(seq_len), x, w, alpha_value, track_value, tag_num); ker(static_cast<int>(seq_len), x, w, alpha_value, track_value, tag_num);
T max_score = -std::numeric_limits<T>::max(); T max_score = -std::numeric_limits<T>::max();
int max_i = 0; int max_i = 0;
......
paddle/fluid/operators/elementwise/elementwise_mul_mkldnn_op.cc
浏览文件 @ 6648995f
...@@ -108,7 +108,7 @@ class ElementwiseMulMKLDNNKernel : public framework::OpKernel<T> { ...@@ -108,7 +108,7 @@ class ElementwiseMulMKLDNNKernel : public framework::OpKernel<T> {
constexpr int simd_width = 16; constexpr int simd_width = 16;
int C = c / simd_width; int C = c / simd_width;
auto multiply = jit::Get<jit::nchw16cmulnc, jit::NCHW16CMulNCTuples, auto multiply = jit::Get<jit::nchw16cmulnc, jit::NCHW16CMulNCTuples<T>,
platform::CPUPlace>(0); platform::CPUPlace>(0);
#pragma omp parallel for collapse(2) #pragma omp parallel for collapse(2)
for (int ni = 0; ni < n; ni++) { for (int ni = 0; ni < n; ni++) {
......
paddle/fluid/operators/fused/fusion_gru_op.cc
浏览文件 @ 6648995f
...@@ -183,29 +183,29 @@ class FusionGRUKernel : public framework::OpKernel<T> { ...@@ -183,29 +183,29 @@ class FusionGRUKernel : public framework::OpKernel<T> {
const int total_T = x_dims[0]; \ const int total_T = x_dims[0]; \
const int D3 = wh_dims[1] const int D3 = wh_dims[1]
#define INIT_OTHER_DEFINES \ #define INIT_OTHER_DEFINES \
auto* h0 = ctx.Input<Tensor>("H0"); \ auto* h0 = ctx.Input<Tensor>("H0"); \
auto* wx = ctx.Input<Tensor>("WeightX"); \ auto* wx = ctx.Input<Tensor>("WeightX"); \
auto* bias = ctx.Input<Tensor>("Bias"); \ auto* bias = ctx.Input<Tensor>("Bias"); \
auto* hidden_out = ctx.Output<LoDTensor>("Hidden"); \ auto* hidden_out = ctx.Output<LoDTensor>("Hidden"); \
bool is_reverse = ctx.Attr<bool>("is_reverse"); \ bool is_reverse = ctx.Attr<bool>("is_reverse"); \
const int M = x_dims[1]; \ const int M = x_dims[1]; \
const int D = wh_dims[0]; \ const int D = wh_dims[0]; \
const int D2 = D * 2; \ const int D2 = D * 2; \
const jit::gru_attr_t attr( \ const jit::gru_attr_t attr( \
D, jit::to_kerneltype(ctx.Attr<std::string>("gate_activation")), \ D, jit::to_kerneltype(ctx.Attr<std::string>("gate_activation")), \
jit::to_kerneltype(ctx.Attr<std::string>("activation"))); \ jit::to_kerneltype(ctx.Attr<std::string>("activation"))); \
jit::gru_t one_step; \ jit::gru_t one_step; \
auto ComputeH1 = \ auto ComputeH1 = \
jit::Get<jit::gruh1, jit::GRUTuples, platform::CPUPlace>(attr); \ jit::Get<jit::gruh1, jit::GRUTuples<T>, platform::CPUPlace>(attr); \
auto ComputeHtPart1 = \ auto ComputeHtPart1 = \
jit::Get<jit::gruhtpart1, jit::GRUTuples, platform::CPUPlace>(attr); \ jit::Get<jit::gruhtpart1, jit::GRUTuples<T>, platform::CPUPlace>(attr); \
auto ComputeHtPart2 = \ auto ComputeHtPart2 = \
jit::Get<jit::gruhtpart2, jit::GRUTuples, platform::CPUPlace>(attr); \ jit::Get<jit::gruhtpart2, jit::GRUTuples<T>, platform::CPUPlace>(attr); \
const T* x_data = x->data<T>(); \ const T* x_data = x->data<T>(); \
const T* wx_data = wx->data<T>(); \ const T* wx_data = wx->data<T>(); \
const T* wh_data = wh->data<T>(); \ const T* wh_data = wh->data<T>(); \
auto place = ctx.GetPlace(); \ auto place = ctx.GetPlace(); \
T* xx_data = xx->mutable_data<T>(place) T* xx_data = xx->mutable_data<T>(place)
void SeqCompute(const framework::ExecutionContext& ctx) const { void SeqCompute(const framework::ExecutionContext& ctx) const {
......
paddle/fluid/operators/fused/fusion_lstm_op.cc
浏览文件 @ 6648995f
...@@ -236,33 +236,32 @@ class FuisonLSTMKernel : public framework::OpKernel<T> { ...@@ -236,33 +236,32 @@ class FuisonLSTMKernel : public framework::OpKernel<T> {
const int D = wh_dims[0]; \ const int D = wh_dims[0]; \
const int D4 = wh_dims[1] const int D4 = wh_dims[1]
#define INIT_OTHER_DEFINES \ #define INIT_OTHER_DEFINES \
const T* x_data = x->data<T>(); \ const T* x_data = x->data<T>(); \
const T* wx_data = wx->data<T>(); \ const T* wx_data = wx->data<T>(); \
const T* wh_data = wh->data<T>(); \ const T* wh_data = wh->data<T>(); \
/* diagonal weight*/ \ /* diagonal weight*/ \
const T* wp_data = bias->data<T>() + D4; \ const T* wp_data = bias->data<T>() + D4; \
/* for peephole only*/ \ /* for peephole only*/ \
T* checked_cell_data = nullptr; \ T* checked_cell_data = nullptr; \
auto place = ctx.GetPlace(); \ auto place = ctx.GetPlace(); \
if (use_peepholes) { \ if (use_peepholes) { \
/* w_ic * Ct-1, w_fc * Ct-1 ; w_oc * Ct => ih*/ \ /* w_ic * Ct-1, w_fc * Ct-1 ; w_oc * Ct => ih*/ \
auto* checked_cell = ctx.Output<Tensor>("CheckedCell"); \ auto* checked_cell = ctx.Output<Tensor>("CheckedCell"); \
checked_cell_data = checked_cell->mutable_data<T>(place); \ checked_cell_data = checked_cell->mutable_data<T>(place); \
} \ } \
const jit \ const jit::lstm_attr_t attr( \
: lstm_attr_t attr( \ D, jit::to_kerneltype(ctx.Attr<std::string>("gate_activation")), \
D, jit::to_kerneltype(ctx.Attr<std::string>("gate_activation")), \ jit::to_kerneltype(ctx.Attr<std::string>("candidate_activation")), \
jit::to_kerneltype(ctx.Attr<std::string>("candidate_activation")), \ jit::to_kerneltype(ctx.Attr<std::string>("cell_activation")), \
jit::to_kerneltype(ctx.Attr<std::string>("cell_activation")), \ use_peepholes); \
use_peepholes); \ jit::lstm_t one_step; \
math::jitkernel::lstm_t one_step; \ one_step.wp = wp_data; \
one_step.wp = wp_data; \ one_step.checked = checked_cell_data; \
one_step.checked = checked_cell_data; \ auto ComputeC1H1 = \
auto ComputeC1H1 = \ jit::Get<jit::lstmc1h1, jit::LSTMTuples<T>, platform::CPUPlace>(attr); \
jit::Get<jit::lstmc1h1, jit::LSTMTuples, platform::CPUPlace>(attr); \ auto ComputeCtHt = \
auto ComputeCtHt = \ jit::Get<jit::lstmctht, jit::LSTMTuples<T>, platform::CPUPlace>(attr)
jit::Get<jit::lstmctht, jit::LSTMTuples, platform::CPUPlace>(attr)
// Wh GEMM // Wh GEMM
#define GEMM_WH_ADDON(bs, prev, out) \ #define GEMM_WH_ADDON(bs, prev, out) \
...@@ -434,7 +433,7 @@ class FuisonLSTMKernel : public framework::OpKernel<T> { ...@@ -434,7 +433,7 @@ class FuisonLSTMKernel : public framework::OpKernel<T> {
one_step.ct_1 = cur_prev_c_data; one_step.ct_1 = cur_prev_c_data;
one_step.ct = cur_c_out_data; one_step.ct = cur_c_out_data;
one_step.ht = cur_h_out_data; one_step.ht = cur_h_out_data;
ComputeC1H1(&one_step, &attr); ComputeCtHt(&one_step, &attr);
// move one batch // move one batch
cur_in_data += D4; cur_in_data += D4;
......
paddle/fluid/operators/jit/helper.h
浏览文件 @ 6648995f
...@@ -32,7 +32,7 @@ inline typename std::enable_if< ...@@ -32,7 +32,7 @@ inline typename std::enable_if<
std::is_same<typename KernelTuples::data_type, float>::value && std::is_same<typename KernelTuples::data_type, float>::value &&
std::is_same<PlaceType, platform::CPUPlace>::value, std::is_same<PlaceType, platform::CPUPlace>::value,
typename KernelTuples::func_type>::type typename KernelTuples::func_type>::type
GetJitCode(typename KernelTuples::attr_type attr) { GetJitCode(const typename KernelTuples::attr_type& attr) {
using Func = typename KernelTuples::func_type; using Func = typename KernelTuples::func_type;
using Attr = typename KernelTuples::attr_type; using Attr = typename KernelTuples::attr_type;
size_t key = JitCodeKey<Attr>(attr); size_t key = JitCodeKey<Attr>(attr);
...@@ -68,7 +68,7 @@ inline typename std::enable_if< ...@@ -68,7 +68,7 @@ inline typename std::enable_if<
!std::is_same<typename KernelTuples::data_type, float>::value || !std::is_same<typename KernelTuples::data_type, float>::value ||
!std::is_same<PlaceType, platform::CPUPlace>::value, !std::is_same<PlaceType, platform::CPUPlace>::value,
typename KernelTuples::func_type>::type typename KernelTuples::func_type>::type
GetJitCode(typename KernelTuples::attr_type attr) { GetJitCode(const typename KernelTuples::attr_type& attr) {
return nullptr; return nullptr;
} }
...@@ -93,8 +93,8 @@ inline typename KernelTuples::func_type GetRefer() { ...@@ -93,8 +93,8 @@ inline typename KernelTuples::func_type GetRefer() {
template <KernelType KT, typename KernelTuples, template <KernelType KT, typename KernelTuples,
typename PlaceType = platform::CPUPlace> typename PlaceType = platform::CPUPlace>
// TODO(TJ): const & attr typename KernelTuples::func_type Get(
typename KernelTuples::func_type Get(typename KernelTuples::attr_type attr) { const typename KernelTuples::attr_type& attr) {
auto jitfunc = GetJitCode<KT, KernelTuples, PlaceType>(attr); auto jitfunc = GetJitCode<KT, KernelTuples, PlaceType>(attr);
if (jitfunc) { if (jitfunc) {
return jitfunc; return jitfunc;
......
paddle/fluid/operators/layer_norm_op.h
浏览文件 @ 6648995f
...@@ -230,7 +230,7 @@ class LayerNormKernel : public framework::OpKernel<T> { ...@@ -230,7 +230,7 @@ class LayerNormKernel : public framework::OpKernel<T> {
PADDLE_ENFORCE_EQ(bias->numel(), right); PADDLE_ENFORCE_EQ(bias->numel(), right);
auto ker = auto ker =
jit::Get<jit::layernorm, jit::LayerNormTuples, platform::CPUPlace>( jit::Get<jit::layernorm, jit::LayerNormTuples<T>, platform::CPUPlace>(
right); right);
ker(x.data<T>(), out.data<T>(), mean->data<T>(), var->data<T>(), ker(x.data<T>(), out.data<T>(), mean->data<T>(), var->data<T>(),
scale->data<T>(), bias->data<T>(), static_cast<int>(left), scale->data<T>(), bias->data<T>(), static_cast<int>(left),
......
paddle/fluid/operators/math/fc_compute.h
浏览文件 @ 6648995f
...@@ -31,13 +31,14 @@ inline void FCCompute(const BlasT<DeviceContext, T>& blas, const int M, ...@@ -31,13 +31,14 @@ inline void FCCompute(const BlasT<DeviceContext, T>& blas, const int M,
} }
if (relu) { if (relu) {
auto compute = auto compute =
jit::Get<jit::vaddrelu, jit::XYZNTuples, platform::CPUPlcace>(N); jit::Get<jit::vaddrelu, jit::XYZNTuples<T>, platform::CPUPlace>(N);
for (int i = 0; i < M; i++) { for (int i = 0; i < M; i++) {
T* dst = Y + i * N; T* dst = Y + i * N;
compute(B, dst, dst, N); compute(B, dst, dst, N);
} }
} else { } else {
auto compute = jit::Get<jit::vadd, jit::XYZNTuples, platform::CPUPlcace>(N); auto compute =
jit::Get<jit::vadd, jit::XYZNTuples<T>, platform::CPUPlace>(N);
#ifdef PADDLE_WITH_MKLML #ifdef PADDLE_WITH_MKLML
#pragma omp parallel for #pragma omp parallel for
#endif #endif
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或