Skip to content

P
Paddle

PaddlePaddle / Paddle
1 年多 前同步成功

通知 2302
Star 20931
Fork 5422
P
Paddle
未验证 提交 fcc8a87b 编写于 作者: Z zhoutianzi666 提交者： GitHub
浏览文件
操作

[inference] add slice trt layer (#43648) 

* add fc, multihead_mul, shape tensor infer, slice
上级 d41a9373
隐藏空白更改
内联 并排
Showing with 500 addition and 219 deletion
paddle/fluid/inference/tensorrt/convert/fc_op.cc
浏览文件 @ fcc8a87b
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
......@@ -31,29 +34,57 @@ namespace tensorrt {
class FcOpConverter : public OpConverter {
public:
nvinfer1::ILayer* reshape_before_fc(nvinfer1::ITensor* before_fc,
nvinfer1::Dims x_dim, int x_num_col_dims,
nvinfer1::Dims x_dim,
int x_num_col_dims,
std::string output_name) {
// add shuffle before fc
nvinfer1::Dims reshape_before_fc_dim;
reshape_before_fc_dim.nbDims = x_num_col_dims + 3;
// padding shape "* x q x 1 x 1"
for (int i = 0; i < reshape_before_fc_dim.nbDims; i++) {
reshape_before_fc_dim.d[i] = 1;
}
for (int i = 0; i < x_dim.nbDims; i++) {
if (i < x_num_col_dims) {
reshape_before_fc_dim.d[i] = 0;
} else {
if (x_dim.d[i] < 0) {
reshape_before_fc_dim.d[x_num_col_dims] = -1;
break;
nvinfer1::ITensor* filal_reshape_before_fc_shape_tensor = nullptr;
if (!engine_->with_dynamic_shape()) {
for (int i = 0; i < reshape_before_fc_dim.nbDims; i++) {
reshape_before_fc_dim.d[i] = 1;
}
for (int i = 0; i < x_dim.nbDims; i++) {
if (i < x_num_col_dims) {
reshape_before_fc_dim.d[i] = 0;
} else {
reshape_before_fc_dim.d[x_num_col_dims] *= x_dim.d[i];
}
reshape_before_fc_dim.d[x_num_col_dims] *= x_dim.d[i];
}
} else {
std::vector<nvinfer1::ITensor*> reshape_before_fc_shape_tensor;
nvinfer1::ITensor* input_shape_tensor = Shape(before_fc);
for (int i = 0; i < reshape_before_fc_dim.nbDims; i++) {
reshape_before_fc_shape_tensor.push_back(Add1DConstantLayer(1));
}
for (int i = 0; i < x_dim.nbDims; i++) {
if (i < x_num_col_dims) {
reshape_before_fc_shape_tensor[i] =
GetEleTensorOfShape(input_shape_tensor, i);
} else {
reshape_before_fc_shape_tensor[x_num_col_dims] =
Prod(GetEleTensorOfShape(input_shape_tensor, i),
reshape_before_fc_shape_tensor[x_num_col_dims]);
}
}
filal_reshape_before_fc_shape_tensor =
Concat(reshape_before_fc_shape_tensor);
}
auto* reshape_before_fc_layer =
TRT_ENGINE_ADD_LAYER(engine_, Shuffle, *before_fc);
reshape_before_fc_layer->setReshapeDimensions(reshape_before_fc_dim);
if (!engine_->with_dynamic_shape()) {
reshape_before_fc_layer->setReshapeDimensions(reshape_before_fc_dim);
} else {
reshape_before_fc_layer->setInput(1,
*filal_reshape_before_fc_shape_tensor);
}
reshape_before_fc_layer->setName(
("fc_op_reshape_before_fc: Shuffle (Output: " + output_name + ")")
.c_str());
......@@ -61,21 +92,39 @@ class FcOpConverter : public OpConverter {
}
nvinfer1::ILayer* reshape_after_fc(nvinfer1::ITensor* after_fc,
nvinfer1::Dims x_dim, int x_num_col_dims) {
nvinfer1::Dims x_dim,
int x_num_col_dims) {
// add shuffle after fc
nvinfer1::Dims reshape_after_fc_dim;
reshape_after_fc_dim.nbDims = x_num_col_dims + 1;
for (int i = 0; i < reshape_after_fc_dim.nbDims; i++) {
reshape_after_fc_dim.d[i] = 0;
nvinfer1::ITensor* filal_reshape_after_fc_shape_tensor = nullptr;
if (!engine_->with_dynamic_shape()) {
for (int i = 0; i < reshape_after_fc_dim.nbDims; i++) {
reshape_after_fc_dim.d[i] = 0;
}
} else {
std::vector<int> gather_indices(x_num_col_dims + 1);
std::iota(gather_indices.begin(), gather_indices.end(), 0);
filal_reshape_after_fc_shape_tensor =
Gather(Shape(after_fc), gather_indices);
}
auto* reshape_after_fc_layer =
TRT_ENGINE_ADD_LAYER(engine_, Shuffle, *after_fc);
reshape_after_fc_layer->setReshapeDimensions(reshape_after_fc_dim);
if (!engine_->with_dynamic_shape()) {
reshape_after_fc_layer->setReshapeDimensions(reshape_after_fc_dim);
} else {
reshape_after_fc_layer->setInput(1, *filal_reshape_after_fc_shape_tensor);
}
return reshape_after_fc_layer;
}
void operator()(const framework::proto::OpDesc& op,
const framework::Scope& scope, bool test_mode) override {
const framework::Scope& scope,
bool test_mode) override {
VLOG(3) << "convert a fluid fc op to tensorrt fc layer without bias";
framework::OpDesc op_desc(op, nullptr);
auto output_name = op_desc.Output("Out").front();
......@@ -93,8 +142,9 @@ class FcOpConverter : public OpConverter {
// Declare weights
auto* Y_v = scope.FindVar(op_desc.Input(w_name).front());
PADDLE_ENFORCE_NOT_NULL(
Y_v, platform::errors::NotFound(
"Can not find %s presistale var of fc in scope.", w_name));
Y_v,
platform::errors::NotFound(
"Can not find %s presistale var of fc in scope.", w_name));
auto* Y_t = Y_v->GetMutable<framework::LoDTensor>();
int x_num_col_dims =
op_desc.HasAttr("x_num_col_dims")
......@@ -125,7 +175,8 @@ class FcOpConverter : public OpConverter {
}
weight_data = engine_->GetWeightCPUData(op_desc.Input(w_name).front(), Y_t);
PADDLE_ENFORCE_EQ(Y_t->dims().size(), 2UL,
PADDLE_ENFORCE_EQ(Y_t->dims().size(),
2UL,
platform::errors::InvalidArgument(
"The fc's weight should be a matrix with 2 dims, but "
"it's %d-dimensional.",
......@@ -140,7 +191,8 @@ class FcOpConverter : public OpConverter {
}
};
auto regist_fc = [&](nvinfer1::ITensor* inputs, int n_output,
auto regist_fc = [&](nvinfer1::ITensor* inputs,
int n_output,
TensorRTEngine::Weight& weight,
TensorRTEngine::Weight& bias) {
if (enable_int8 || support_int8) {
......@@ -148,7 +200,8 @@ class FcOpConverter : public OpConverter {
float out_scale = 0;
if (enable_int8) {
PADDLE_ENFORCE_EQ(
op_desc.HasAttr("out_threshold"), true,
op_desc.HasAttr("out_threshold"),
true,
platform::errors::InvalidArgument(
"must have out threshold in fc layers in int8 mode"));
out_scale = BOOST_GET_CONST(float, op_desc.GetAttr("out_threshold"));
......@@ -156,9 +209,13 @@ class FcOpConverter : public OpConverter {
out_scale = BOOST_GET_CONST(float, op_desc.GetAttr("Out"));
}
nvinfer1::DimsHW nv_ksize(1, 1);
auto* fc_layer_int8 =
TRT_ENGINE_ADD_LAYER(engine_, Convolution, *inputs, n_output,
nv_ksize, weight.get(), bias.get());
auto* fc_layer_int8 = TRT_ENGINE_ADD_LAYER(engine_,
Convolution,
*inputs,
n_output,
nv_ksize,
weight.get(),
bias.get());
fc_layer_int8->setName(
("fc_op_int8_conv1x1: Convolution (Output: " + output_name + ")")
.c_str());
......@@ -171,21 +228,29 @@ class FcOpConverter : public OpConverter {
.c_str());
engine_->SetTensorDynamicRange(fc_after_reshape_int8->getOutput(0),
out_scale);
nvinfer1::IActivationLayer* relu_layer_int8 = TRT_ENGINE_ADD_LAYER(
engine_, Activation, *(fc_after_reshape_int8->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_int8, "relu_after_fc_shuffle",
{output_name}, test_mode);
nvinfer1::IActivationLayer* relu_layer_int8 =
TRT_ENGINE_ADD_LAYER(engine_,
Activation,
*(fc_after_reshape_int8->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_int8,
"relu_after_fc_shuffle",
{output_name},
test_mode);
} else {
RreplenishLayerAndOutput(fc_after_reshape_int8,
"fc_op_int8_reshape_after_fc: Shuffle",
{output_name}, test_mode);
{output_name},
test_mode);
}
} else {
// add fc layer
auto* fc_layer_float =
TRT_ENGINE_ADD_LAYER(engine_, FullyConnected, *inputs, n_output,
weight.get(), bias.get());
auto* fc_layer_float = TRT_ENGINE_ADD_LAYER(engine_,
FullyConnected,
*inputs,
n_output,
weight.get(),
bias.get());
fc_layer_float->setName(
("fc_op_float: FullyConnected (Output: " + output_name + ")")
.c_str());
......@@ -195,14 +260,20 @@ class FcOpConverter : public OpConverter {
fc_after_reshape_float->setName(
("float_reshape_after_fc: Shuffle (Output: " + output_name + ")")
.c_str());
nvinfer1::IActivationLayer* relu_layer_float = TRT_ENGINE_ADD_LAYER(
engine_, Activation, *(fc_after_reshape_float->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_float, "relu_after_fc_shuffle",
{output_name}, test_mode);
nvinfer1::IActivationLayer* relu_layer_float =
TRT_ENGINE_ADD_LAYER(engine_,
Activation,
*(fc_after_reshape_float->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_float,
"relu_after_fc_shuffle",
{output_name},
test_mode);
} else {
RreplenishLayerAndOutput(fc_after_reshape_float, "shuffle_after_fc",
{output_name}, test_mode);
RreplenishLayerAndOutput(fc_after_reshape_float,
"shuffle_after_fc",
{output_name},
test_mode);
}
}
};
......@@ -251,15 +322,20 @@ class FcOpConverter : public OpConverter {
if (enable_int8 || support_int8) {
// add conv1x1 layer
nvinfer1::DimsHW nv_ksize(1, 1);
auto* fc_layer_int8 =
TRT_ENGINE_ADD_LAYER(engine_, Convolution, *X, n_output, nv_ksize,
weight.get(), bias.get());
auto* fc_layer_int8 = TRT_ENGINE_ADD_LAYER(engine_,
Convolution,
*X,
n_output,
nv_ksize,
weight.get(),
bias.get());
if (activation_type == "relu") {
fc_layer_int8->setName(
("ernie_fc_op_int8: Convolution (Output: " + output_name + ")")
.c_str());
PADDLE_ENFORCE_EQ(
op_desc.HasAttr("out_threshold"), true,
op_desc.HasAttr("out_threshold"),
true,
platform::errors::InvalidArgument(
"must have out threshold in fc layers in int8 mode"));
float out_scale = 0;
......@@ -271,15 +347,20 @@ class FcOpConverter : public OpConverter {
}
engine_->SetTensorDynamicRange(fc_layer_int8->getOutput(0),
out_scale);
nvinfer1::IActivationLayer* relu_layer_int8 = TRT_ENGINE_ADD_LAYER(
engine_, Activation, *(fc_layer_int8->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_int8, "relu_after_ernie_fc_int8",
{output_name}, test_mode);
nvinfer1::IActivationLayer* relu_layer_int8 =
TRT_ENGINE_ADD_LAYER(engine_,
Activation,
*(fc_layer_int8->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_int8,
"relu_after_ernie_fc_int8",
{output_name},
test_mode);
} else {
RreplenishLayerAndOutput(fc_layer_int8,
"ernie_fc_op_int8: Convolution",
{output_name}, test_mode);
{output_name},
test_mode);
}
} else {
// add fc layer
......@@ -288,25 +369,30 @@ class FcOpConverter : public OpConverter {
if (activation_type == "relu") {
fc_layer_float->setName(
("ernie_fc_op_float: (Output: " + output_name + ")").c_str());
nvinfer1::IActivationLayer* relu_layer_float = TRT_ENGINE_ADD_LAYER(
engine_, Activation, *(fc_layer_float->getOutput(0)),
nvinfer1::ActivationType::kRELU);
nvinfer1::IActivationLayer* relu_layer_float =
TRT_ENGINE_ADD_LAYER(engine_,
Activation,
*(fc_layer_float->getOutput(0)),
nvinfer1::ActivationType::kRELU);
RreplenishLayerAndOutput(relu_layer_float,
"relu_after_ernie_fc_float", {output_name},
"relu_after_ernie_fc_float",
{output_name},
test_mode);
} else {
RreplenishLayerAndOutput(fc_layer_float, "ernie_fc_op_float",
{output_name}, test_mode);
RreplenishLayerAndOutput(
fc_layer_float, "ernie_fc_op_float", {output_name}, test_mode);
}
}
} else { // need reshape input before and after fc
PADDLE_ENFORCE_GT(
x_dim.nbDims, x_num_col_dims,
x_dim.nbDims,
x_num_col_dims,
platform::errors::InvalidArgument(
"Params and input dims mismatch. Paddle-TRT FC "
"converter expects x_dim.nbDims > x_num_col_dims, but "
"x_dim.nbDims : %d, x_num_col_dims : %d.",
x_dim.nbDims, x_num_col_dims));
x_dim.nbDims,
x_num_col_dims));
auto* reshape_before_fc_layer =
reshape_before_fc(X, x_dim, x_num_col_dims, output_name);
auto* reshape_itensor = reshape_before_fc_layer->getOutput(0);
......
paddle/fluid/inference/tensorrt/convert/multihead_matmul_op.cc
浏览文件 @ fcc8a87b
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See
......@@ -19,7 +22,8 @@ namespace tensorrt {
class MultiheadMatMulOpConverter : public OpConverter {
public:
void operator()(const framework::proto::OpDesc& op,
const framework::Scope& scope, bool test_mode) override {
const framework::Scope& scope,
bool test_mode) override {
VLOG(3) << "convert a fluid multihead_mamul op to a corresponding tensorrt "
"network structure";
framework::OpDesc op_desc(op, nullptr);
......@@ -49,8 +53,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
float* bias_data = engine_->GetWeightCPUData(bias_name, bias_t);
std::vector<float> weight_data_tmp;
weight_data_tmp.reserve(weight_t->numel());
memcpy(weight_data_tmp.data(), weight_data,
weight_t->numel() * sizeof(float));
memcpy(
weight_data_tmp.data(), weight_data, weight_t->numel() * sizeof(float));
// (hidden_in, 3, hidden_out)
const auto& weight_dims = weight_t->dims();
......@@ -98,14 +102,15 @@ class MultiheadMatMulOpConverter : public OpConverter {
nvinfer1::ILayer* fc_layer = nullptr;
float dp_probs = 1.0 / 127.0;
nvinfer1::DimsHW nv_ksize(1, 1);
fc_layer = TRT_ENGINE_ADD_LAYER(engine_, Convolution, *input, n,
nv_ksize, weight, bias);
fc_layer = TRT_ENGINE_ADD_LAYER(
engine_, Convolution, *input, n, nv_ksize, weight, bias);
fc_layer->setName(
("Multihead: Convolution/FullyConnected: (Output: " +
output_name + ")")
.c_str());
PADDLE_ENFORCE_EQ(
op_desc.HasAttr("fc_out_threshold"), true,
op_desc.HasAttr("fc_out_threshold"),
true,
platform::errors::InvalidArgument(
"must have out_threshold in multihead layers in int8 mode"));
float out_scale =
......@@ -119,13 +124,19 @@ class MultiheadMatMulOpConverter : public OpConverter {
"CustomQKVToContextPluginDynamic", "3");
assert(creator != nullptr);
std::vector<nvinfer1::PluginField> fields{
{"hidden_size", &hidden_out, nvinfer1::PluginFieldType::kINT32,
{"hidden_size",
&hidden_out,
nvinfer1::PluginFieldType::kINT32,
1},
{"num_heads", &head_number, nvinfer1::PluginFieldType::kINT32,
{"num_heads",
&head_number,
nvinfer1::PluginFieldType::kINT32,
1}};
if (qkv2context_plugin_int8) {
fields.push_back({"dq_probs", &dp_probs,
nvinfer1::PluginFieldType::kFLOAT32, 1});
fields.push_back({"dq_probs",
&dp_probs,
nvinfer1::PluginFieldType::kFLOAT32,
1});
}
nvinfer1::PluginFieldCollection* plugin_collection =
static_cast<nvinfer1::PluginFieldCollection*>(malloc(
......@@ -154,7 +165,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
engine_->GetITensor(engine_->network()->getInput(3)->getName());
engine_->SetTensorDynamicRange(max_seqlen_tensor, 1.0f);
auto* shuffle_layer = TRT_ENGINE_ADD_LAYER(
engine_, Shuffle,
engine_,
Shuffle,
*const_cast<nvinfer1::ITensor*>(max_seqlen_tensor));
nvinfer1::Dims shape_dim;
shape_dim.nbDims = 1;
......@@ -173,8 +185,11 @@ class MultiheadMatMulOpConverter : public OpConverter {
// [3, head_number, head_size, hidden_in] -> [head_number, 3,
// head_size,
// hidden_in]
auto transpose_weight_v2 = [](const float* src, float* dst, int three,
int head_number, int head_size,
auto transpose_weight_v2 = [](const float* src,
float* dst,
int three,
int head_number,
int head_size,
int hidden_in) {
const int HH = head_size * hidden_in;
for (int i = 0; i < three; ++i) {
......@@ -187,41 +202,47 @@ class MultiheadMatMulOpConverter : public OpConverter {
}
};
// [3, head_number, head_size] -> [head_number, 3, head_size]
auto transpose_bias_v2 = [](const float* src, float* dst, int N,
int H) {
for (int i = 0; i < 3; ++i) {
for (int n = 0; n < N; ++n) {
for (int h = 0; h < H; ++h) {
dst[n * 3 * H + i * H + h] = src[i * N * H + n * H + h];
auto transpose_bias_v2 =
[](const float* src, float* dst, int N, int H) {
for (int i = 0; i < 3; ++i) {
for (int n = 0; n < N; ++n) {
for (int h = 0; h < H; ++h) {
dst[n * 3 * H + i * H + h] = src[i * N * H + n * H + h];
}
}
}
}
}
};
memcpy(weight_data_tmp.data(), weight_data,
};
memcpy(weight_data_tmp.data(),
weight_data,
weight_t->numel() * sizeof(float));
transpose_weight_v2(weight_data_tmp.data(), weight_data, three,
head_number, head_size, hidden_in);
transpose_weight_v2(weight_data_tmp.data(),
weight_data,
three,
head_number,
head_size,
hidden_in);
std::vector<float> bias_data_tmp;
bias_data_tmp.reserve(bias_t->numel());
memcpy(bias_data_tmp.data(), bias_data,
bias_t->numel() * sizeof(float));
transpose_bias_v2(bias_data_tmp.data(), bias_data, head_number,
head_size);
memcpy(
bias_data_tmp.data(), bias_data, bias_t->numel() * sizeof(float));
transpose_bias_v2(
bias_data_tmp.data(), bias_data, head_number, head_size);
nvinfer1::ILayer* fc_layer = nullptr;
float dp_probs = 1.0 / 127.0;
if (op_desc.HasAttr("Input_scale")) {
nvinfer1::DimsHW nv_ksize(1, 1);
fc_layer = TRT_ENGINE_ADD_LAYER(engine_, Convolution, *input, n,
nv_ksize, weight, bias);
fc_layer = TRT_ENGINE_ADD_LAYER(
engine_, Convolution, *input, n, nv_ksize, weight, bias);
} else {
fc_layer = TRT_ENGINE_ADD_LAYER(engine_, FullyConnected, *input, n,
weight, bias);
fc_layer = TRT_ENGINE_ADD_LAYER(
engine_, FullyConnected, *input, n, weight, bias);
}
if (op_desc.HasAttr("fc_out_threshold")) {
PADDLE_ENFORCE_EQ(op_desc.HasAttr("fc_out_threshold"), true,
PADDLE_ENFORCE_EQ(op_desc.HasAttr("fc_out_threshold"),
true,
platform::errors::InvalidArgument(
"must have out threshold in multihead layers "
"in int8 mode"));
......@@ -245,15 +266,21 @@ class MultiheadMatMulOpConverter : public OpConverter {
int var_seqlen = 1;
std::vector<nvinfer1::PluginField> fields{
{"type_id", &type, nvinfer1::PluginFieldType::kINT32, 1},
{"hidden_size", &hidden_out, nvinfer1::PluginFieldType::kINT32,
{"hidden_size",
&hidden_out,
nvinfer1::PluginFieldType::kINT32,
1},
{"num_heads", &head_number, nvinfer1::PluginFieldType::kINT32, 1},
{"has_mask", &has_mask, nvinfer1::PluginFieldType::kINT32, 1},
{"var_seqlen", &var_seqlen, nvinfer1::PluginFieldType::kINT32,
{"var_seqlen",
&var_seqlen,
nvinfer1::PluginFieldType::kINT32,
1}};
if (qkv2context_plugin_int8) {
fields.push_back({"dq_probs", &dp_probs,
nvinfer1::PluginFieldType::kFLOAT32, 1});
fields.push_back({"dq_probs",
&dp_probs,
nvinfer1::PluginFieldType::kFLOAT32,
1});
}
nvinfer1::PluginFieldCollection* plugin_collection =
static_cast<nvinfer1::PluginFieldCollection*>(malloc(
......@@ -274,7 +301,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
auto max_seqlen_tensor = engine_->GetITensor("mask_id");
auto* shuffle_layer = TRT_ENGINE_ADD_LAYER(
engine_, Shuffle,
engine_,
Shuffle,
*const_cast<nvinfer1::ITensor*>(max_seqlen_tensor));
nvinfer1::Dims shape_dim;
shape_dim.nbDims = 1;
......@@ -290,7 +318,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
}
} else {
PADDLE_ENFORCE_EQ(
input->getDimensions().nbDims, 3,
input->getDimensions().nbDims,
3,
platform::errors::InvalidArgument(
"The Input dim of the MultiheadMatMul should be 3, "
"but it's (%d) now.",
......@@ -309,20 +338,24 @@ class MultiheadMatMulOpConverter : public OpConverter {
static_cast<size_t>(bias_t->numel())};
// add shuffle before fc
nvinfer1::Dims reshape_before_fc_dim;
reshape_before_fc_dim.nbDims = 5;
reshape_before_fc_dim.d[0] = 0;
reshape_before_fc_dim.d[1] = 0;
reshape_before_fc_dim.d[2] = 0;
reshape_before_fc_dim.d[3] = 1;
reshape_before_fc_dim.d[4] = 1;
std::vector<nvinfer1::ITensor*> reshape_before_fc_shape_tensor;
nvinfer1::ITensor* input_shape_tensor = Shape(input);
for (int i = 0; i < 5; i++) {
reshape_before_fc_shape_tensor.push_back(Add1DConstantLayer(1));
}
for (int i = 0; i < 3; i++) {
reshape_before_fc_shape_tensor[i] =
GetEleTensorOfShape(input_shape_tensor, i);
}
auto* reshape_before_fc_layer =
TRT_ENGINE_ADD_LAYER(engine_, Shuffle, *input);
if (op_desc.HasAttr("Input_scale")) {
engine_->SetTensorDynamicRange(reshape_before_fc_layer->getOutput(0),
in_scale);
}
reshape_before_fc_layer->setReshapeDimensions(reshape_before_fc_dim);
reshape_before_fc_layer->setInput(
1, *Concat(reshape_before_fc_shape_tensor));
reshape_before_fc_layer->setName(
("shuffle_before_multihead_mamul(Output: " + output_name + ")")
.c_str());
......@@ -331,18 +364,28 @@ class MultiheadMatMulOpConverter : public OpConverter {
nvinfer1::ILayer* fc_layer = nullptr;
if (op_desc.HasAttr("Input_scale")) {
nvinfer1::DimsHW nv_ksize(1, 1);
fc_layer = TRT_ENGINE_ADD_LAYER(
engine_, Convolution, *reshape_before_fc_layer->getOutput(0), n,
nv_ksize, weight.get(), bias.get());
fc_layer =
TRT_ENGINE_ADD_LAYER(engine_,
Convolution,
*reshape_before_fc_layer->getOutput(0),
n,
nv_ksize,
weight.get(),
bias.get());
} else {
fc_layer = TRT_ENGINE_ADD_LAYER(
engine_, FullyConnected, *reshape_before_fc_layer->getOutput(0),
n, weight.get(), bias.get());
fc_layer =
TRT_ENGINE_ADD_LAYER(engine_,
FullyConnected,
*reshape_before_fc_layer->getOutput(0),
n,
weight.get(),
bias.get());
}
if (op_desc.HasAttr("fc_out_threshold")) {
PADDLE_ENFORCE_EQ(
op_desc.HasAttr("fc_out_threshold"), true,
op_desc.HasAttr("fc_out_threshold"),
true,
platform::errors::InvalidArgument(
"must have out threshold in multihead layers in int8 mode"));
float out_scale =
......@@ -369,8 +412,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
with_fp16 = true;
}
plugin::DynamicPluginTensorRT* plugin =
new plugin::QkvToContextPluginDynamic(hidden_in, head_number,
head_size, scale, with_fp16);
new plugin::QkvToContextPluginDynamic(
hidden_in, head_number, head_size, scale, with_fp16);
layer = engine_->AddDynamicPlugin(plugin_inputs.data(), 2, plugin);
}
} else {
......@@ -380,8 +423,8 @@ class MultiheadMatMulOpConverter : public OpConverter {
"You can use the config.SetTRTDynamicShapeInfo(...) interface to set "
"the shape information to run the dynamic shape mode."));
}
RreplenishLayerAndOutput(layer, "multihead_matmul", {output_name},
test_mode);
RreplenishLayerAndOutput(
layer, "multihead_matmul", {output_name}, test_mode);
}
};
......
paddle/fluid/inference/tensorrt/convert/op_converter.h
浏览文件 @ fcc8a87b
......@@ -47,14 +47,16 @@ class OpConverter {
// test_mode: whether the instance executes in an unit test.
void ConvertOp(const framework::proto::OpDesc& op,
const std::unordered_set<std::string>& parameters,
const framework::Scope& scope, TensorRTEngine* engine,
const framework::Scope& scope,
TensorRTEngine* engine,
bool test_mode = false) {
framework::OpDesc op_desc(op, nullptr);
OpConverter* it{nullptr};
if (op_desc.Type() == "mul") {
PADDLE_ENFORCE_EQ(op_desc.Input("Y").size(), 1UL,
PADDLE_ENFORCE_EQ(op_desc.Input("Y").size(),
1UL,
platform::errors::InvalidArgument(
"The input op mul's Input(\"Y\")."
"size() should equal to 1, but reveceid "
......@@ -70,7 +72,8 @@ class OpConverter {
"add", "mul", "sub", "div", "max", "min", "pow"};
static std::unordered_set<std::string> add_weight_op_set{
"add", "mul", "sub", "div", "pow"};
PADDLE_ENFORCE_EQ(op_desc.Input("Y").size(), 1UL,
PADDLE_ENFORCE_EQ(op_desc.Input("Y").size(),
1UL,
platform::errors::InvalidArgument(
"The input op's Input(\"Y\")."
"size() should equal to 1, but reveceid "
......@@ -81,64 +84,74 @@ class OpConverter {
std::string Y = op_desc.Input("Y")[0];
if (parameters.count(Y)) {
PADDLE_ENFORCE_GT(
add_weight_op_set.count(op_type), 0,
add_weight_op_set.count(op_type),
0,
platform::errors::Unimplemented("Unsupported elementwise type %s",
op_type.c_str()));
it = Registry<OpConverter>::Global().Lookup("elementwise_" + op_type +
"_weight");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented(
"no OpConverter for optype [%s]", op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
} else {
PADDLE_ENFORCE_GT(
add_tensor_op_set.count(op_type), 0,
add_tensor_op_set.count(op_type),
0,
platform::errors::Unimplemented("Unsupported elementwise type %s",
op_type.c_str()));
it = Registry<OpConverter>::Global().Lookup("elementwise_" + op_type +
"_tensor");
}
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
if (op_desc.Type() == "depthwise_conv2d") {
it = Registry<OpConverter>::Global().Lookup("conv2d");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
if (op_desc.Type() == "depthwise_conv2d_transpose") {
it = Registry<OpConverter>::Global().Lookup("conv2d_transpose");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
if (op_desc.Type() == "transpose2") {
it = Registry<OpConverter>::Global().Lookup("transpose");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
if (op_desc.Type() == "flatten2") {
it = Registry<OpConverter>::Global().Lookup("flatten");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
// reshape2 == reshape
if (op_desc.Type() == "reshape2") {
it = Registry<OpConverter>::Global().Lookup("reshape");
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
}
if (!it) {
it = Registry<OpConverter>::Global().Lookup(op_desc.Type());
}
PADDLE_ENFORCE_NOT_NULL(
it, platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it,
platform::errors::Unimplemented("no OpConverter for optype [%s]",
op_desc.Type()));
it->SetEngine(engine);
(*it)(op, scope, test_mode);
......@@ -214,7 +227,8 @@ class OpConverter {
// the INetwork's inputs and outputs should specified in some other modules.
void ConvertBlock(const framework::proto::BlockDesc& block,
const std::unordered_set<std::string>& parameters,
const framework::Scope& scope, TensorRTEngine* engine) {
const framework::Scope& scope,
TensorRTEngine* engine) {
std::unique_lock<std::mutex> lk(mut_);
for (int i = 0; i < block.ops_size(); i++) {
const auto& op = block.ops(i);
......@@ -224,20 +238,24 @@ class OpConverter {
// The scope here should be inited with the parameter vars.
void ConvertBlockToTRTEngine(
framework::BlockDesc* block_desc, const framework::Scope& scope,
framework::BlockDesc* block_desc,
const framework::Scope& scope,
const std::vector<std::string>& inputs,
const std::unordered_set<std::string>& parameters,
const std::vector<std::string>& outputs, TensorRTEngine* engine) {
const std::vector<std::string>& outputs,
TensorRTEngine* engine) {
engine->InitNetwork();
bool all_dynamic_shape_set = true;
for (auto& input : inputs) {
if (parameters.count(input)) continue;
auto* var = block_desc->FindVar(input);
PADDLE_ENFORCE_NOT_NULL(
var, platform::errors::NotFound("no variable called %s in block.",
input.c_str()));
var,
platform::errors::NotFound("no variable called %s in block.",
input.c_str()));
PADDLE_ENFORCE_EQ(
var->GetType(), FluidDT::VarType_Type_LOD_TENSOR,
var->GetType(),
FluidDT::VarType_Type_LOD_TENSOR,
platform::errors::InvalidArgument("TensorRT engine only takes "
"LoDTensor as input"));
auto var_shape = var->GetShape();
......@@ -262,7 +280,8 @@ class OpConverter {
} else {
input_shape.push_back(min_input_shape[i]);
// the i dimension should be same.
PADDLE_ENFORCE_EQ(min_input_shape[i], optim_input_shape[i],
PADDLE_ENFORCE_EQ(min_input_shape[i],
optim_input_shape[i],
platform::errors::InvalidArgument(
"The dim (%d) of the min_input_shape and "
"optim_input_shape should be same."));
......@@ -282,7 +301,8 @@ class OpConverter {
Vec2TRT_Dims(var_shape, input));
}
}
PADDLE_ENFORCE_EQ(all_dynamic_shape_set, true,
PADDLE_ENFORCE_EQ(all_dynamic_shape_set,
true,
platform::errors::InvalidArgument(
"some trt inputs dynamic shape info not set, "
"check the INFO log above for more details."));
......@@ -297,7 +317,8 @@ class OpConverter {
// rank(result) = rank(input)
nvinfer1::ITensor* Gather(nvinfer1::ITensor* input,
const std::vector<int32_t> indices, int axis = 0) {
const std::vector<int32_t> indices,
int axis = 0) {
auto* indices_tensor = Add1DConstantLayer(indices, " ");
auto* result =
TRT_ENGINE_ADD_LAYER(engine_, Gather, *input, *indices_tensor, axis)
......@@ -326,8 +347,8 @@ class OpConverter {
// Concat not make rank changed
nvinfer1::ITensor* Concat(const std::vector<nvinfer1::ITensor*>& inputs,
int axis = 0) {
auto* layer = TRT_ENGINE_ADD_LAYER(engine_, Concatenation, inputs.data(),
inputs.size());
auto* layer = TRT_ENGINE_ADD_LAYER(
engine_, Concatenation, inputs.data(), inputs.size());
if (axis != 0) layer->setAxis(axis);
nvinfer1::ITensor* c = layer->getOutput(0);
return c;
......@@ -335,48 +356,48 @@ class OpConverter {
nvinfer1::ITensor* Sum(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kSUM)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kSUM)
->getOutput(0);
return c;
}
nvinfer1::ITensor* Prod(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kPROD)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kPROD)
->getOutput(0);
return c;
}
nvinfer1::ITensor* Min(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kMIN)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kMIN)
->getOutput(0);
return c;
}
nvinfer1::ITensor* Max(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kMAX)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kMAX)
->getOutput(0);
return c;
}
nvinfer1::ITensor* Sub(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kSUB)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kSUB)
->getOutput(0);
return c;
}
nvinfer1::ITensor* Div(nvinfer1::ITensor* a, nvinfer1::ITensor* b) {
nvinfer1::ITensor* c =
TRT_ENGINE_ADD_LAYER(engine_, ElementWise, *a, *b,
nvinfer1::ElementWiseOperation::kDIV)
TRT_ENGINE_ADD_LAYER(
engine_, ElementWise, *a, *b, nvinfer1::ElementWiseOperation::kDIV)
->getOutput(0);
return c;
}
......@@ -390,10 +411,14 @@ class OpConverter {
// Get element tensor of 1D shape tensor
nvinfer1::ITensor* GetEleTensorOfShape(nvinfer1::ITensor* shape_tensor,
int index, bool is_scalar = false) {
int index,
bool is_scalar = false) {
auto* tensor =
TRT_ENGINE_ADD_LAYER(engine_, Gather, *shape_tensor,
*Add1DConstantLayer(index, " ", is_scalar), 0)
TRT_ENGINE_ADD_LAYER(engine_,
Gather,
*shape_tensor,
*Add1DConstantLayer(index, " ", is_scalar),
0)
->getOutput(0);
return tensor;
}
......@@ -403,8 +428,8 @@ class OpConverter {
const std::vector<int32_t>& weight_dims,
const std::string& weight_name) {
std::unique_ptr<framework::Tensor> tmp_tensor(new framework::Tensor());
int data_size = std::accumulate(weight_dims.begin(), weight_dims.end(), 1,
std::multiplies<int>());
int data_size = std::accumulate(
weight_dims.begin(), weight_dims.end(), 1, std::multiplies<int>());
tmp_tensor->Resize({data_size});
auto* tmp_data = tmp_tensor->mutable_data<float>(platform::CPUPlace());
for (int i = 0; i < data_size; i++) {
......@@ -489,7 +514,8 @@ class OpConverter {
}
void RreplenishLayerAndOutput(
nvinfer1::ILayer* layer, const std::string& layer_type,
nvinfer1::ILayer* layer,
const std::string& layer_type,
const std::vector<std::string>& output_tensor_names,
bool test_mode = false) {
size_t num_out = output_tensor_names.size();
......
paddle/fluid/inference/tensorrt/convert/slice_op.cc
浏览文件 @ fcc8a87b
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
......@@ -22,7 +19,8 @@ namespace tensorrt {
class SliceOpConverter : public OpConverter {
public:
void operator()(const framework::proto::OpDesc& op,
const framework::Scope& scope, bool test_mode) override {
const framework::Scope& scope,
bool test_mode) override {
// This OP is implemented by trt dynamic shpae plugin.
// Dynamic shape plugin requires TRT version greater than 6.0.
VLOG(4) << "convert slice op to tensorrt layer";
......@@ -63,28 +61,118 @@ class SliceOpConverter : public OpConverter {
}
ends[i] = std::min(ends[i], input_dims.d[axes[i]]);
PADDLE_ENFORCE_GT(
ends[i], starts[i],
ends[i],
starts[i],
platform::errors::InvalidArgument(
"Attr(ends) should be greater than attr(starts) in "
"slice op. But received ends = %d, starts = %d.",
ends[i], starts[i]));
ends[i],
starts[i]));
}
}
nvinfer1::ILayer* layer = nullptr;
if (engine_->with_dynamic_shape()) {
#if IS_TRT_VERSION_GE(6000)
auto nchw_input_dims = input->getDimensions();
nvinfer1::Dims trt_start_dims;
trt_start_dims.nbDims = nchw_input_dims.nbDims;
memset(trt_start_dims.d, 0, sizeof(int32_t) * nchw_input_dims.nbDims);
nvinfer1::Dims trt_size_dims = trt_start_dims;
nvinfer1::Dims trt_end_dims = trt_start_dims;
nvinfer1::Dims trt_step_dims = trt_start_dims;
for (int i = 0; i < trt_step_dims.nbDims; i++) trt_step_dims.d[i] = 1;
// input : [N,C,H,W]
bool has_neg_indices = false;
for (size_t i = 0; i < axes.size(); i++) {
int trt_axis = axes[i];
trt_start_dims.d[trt_axis] = starts[i];
trt_end_dims.d[trt_axis] = ends[i];
if (starts[i] < 0 || ends[i] < 0) has_neg_indices = true;
}
auto* shape_tensor = Shape(input);
auto* start_tensor = Add1DConstantLayer(trt_start_dims);
if (has_neg_indices) {
start_tensor = FixNegIndices(shape_tensor, start_tensor);
}
std::vector<nvinfer1::ITensor*> end_vec_tensor;
for (int i = 0; i < trt_end_dims.nbDims; i++) {
end_vec_tensor.push_back(GetEleTensorOfShape(shape_tensor, i));
}
for (size_t i = 0; i < axes.size(); i++) {
int trt_axis = axes[i];
if (ends[i] >= 0) {
end_vec_tensor[trt_axis] = Add1DConstantLayer(ends[i]);
} else {
end_vec_tensor[trt_axis] =
Sum(end_vec_tensor[trt_axis], Add1DConstantLayer(ends[i]));
}
}
// CI failed in trt 6015 but success in 7134, may be a trt bug
#if IS_TRT_VERSION_GE(7134)
auto* size_tensor =
Sub(Min(Concat(end_vec_tensor), shape_tensor), start_tensor);
#else
auto* size_tensor = Sub(Concat(end_vec_tensor), start_tensor);
#endif
layer = TRT_ENGINE_ADD_LAYER(
engine_, Slice, *input, trt_start_dims, trt_size_dims, trt_step_dims);
layer->setInput(1, *start_tensor);
layer->setInput(2, *size_tensor);
if (decrease_axises.size() > 0) {
std::vector<int32_t> gather_indices;
for (int i = 0; i < trt_size_dims.nbDims; i++) {
if (decrease_axises.end() !=
std::find(decrease_axises.begin(), decrease_axises.end(), i))
continue;
gather_indices.push_back(i);
}
if (gather_indices.empty())
gather_indices.push_back(decrease_axises[0]);
auto real_size_tensor = Gather(size_tensor, gather_indices);
layer = TRT_ENGINE_ADD_LAYER(engine_, Shuffle, *layer->getOutput(0));
layer->setInput(1, *real_size_tensor);
}
#else
bool with_fp16 =
engine_->WithFp16() && !engine_->disable_trt_plugin_fp16();
int decrease_axis = decrease_axises.size() == 0 ? -1 : decrease_axises[0];
plugin::SlicePluginDynamic* plugin = new plugin::SlicePluginDynamic(
starts, ends, axes, decrease_axis, with_fp16);
layer = engine_->AddDynamicPlugin(&input, 1, plugin);
#endif
} else {
#if IS_TRT_VERSION_GE(6000)
auto chw_input_dims = input->getDimensions();
nvinfer1::Dims trt_start_dims;
trt_start_dims.nbDims = chw_input_dims.nbDims;
memset(trt_start_dims.d, 0, sizeof(int32_t) * chw_input_dims.nbDims);
nvinfer1::Dims trt_size_dims = chw_input_dims;
nvinfer1::Dims trt_step_dims;
trt_step_dims.nbDims = chw_input_dims.nbDims;
for (int i = 0; i < trt_step_dims.nbDims; i++) trt_step_dims.d[i] = 1;
// input : [C,H,W]
for (size_t i = 0; i < axes.size(); i++) {
int trt_axis = axes[i] - 1;
trt_start_dims.d[trt_axis] = starts[i];
trt_size_dims.d[trt_axis] = ends[i] - starts[i];
}
layer = TRT_ENGINE_ADD_LAYER(
engine_, Slice, *input, trt_start_dims, trt_size_dims, trt_step_dims);
#else
bool with_fp16 =
engine_->WithFp16() && !engine_->disable_trt_plugin_fp16();
plugin::SlicePlugin* plugin =
new plugin::SlicePlugin(starts, ends, axes, with_fp16);
layer = engine_->AddPlugin(&input, 1, plugin);
#endif
}
RreplenishLayerAndOutput(layer, "slice", {output_name}, test_mode);
}
......
paddle/fluid/inference/tensorrt/engine.cc
浏览文件 @ fcc8a87b
......@@ -49,7 +49,8 @@ void TensorRTEngine::InitNetwork() {
optim_profiles_[i] = infer_builder_->createOptimizationProfile();
}
void TensorRTEngine::Execute(int batch_size, std::vector<void *> *buffers,
void TensorRTEngine::Execute(int batch_size,
std::vector<void *> *buffers,
cudaStream_t stream) {
freshDeviceId();
auto infer_context = context();
......@@ -129,14 +130,32 @@ void TensorRTEngine::FreezeNetwork() {
}
#if IS_TRT_VERSION_GE(5122)
auto is_layer_int8 = [&](nvinfer1::ILayer *layer) -> bool {
auto layer_int8_fallback = [&](nvinfer1::ILayer *layer) -> bool {
if (layer->getType() == nvinfer1::LayerType::kSHAPE) {
return false;
}
bool all_int = true;
for (int j = 0; j < layer->getNbInputs(); j++) {
auto *temp_in = layer->getInput(j);
if (temp_in->getType() != nvinfer1::DataType::kINT32) {
all_int = false;
}
}
for (int j = 0; j < layer->getNbOutputs(); j++) {
auto *temp_out = layer->getOutput(j);
if (temp_out->getType() != nvinfer1::DataType::kINT32) {
all_int = false;
}
}
if (all_int) return false;
for (int j = 0; j < layer->getNbInputs(); j++) {
auto *temp_in = layer->getInput(j);
if (!temp_in->dynamicRangeIsSet()) {
VLOG(1) << "Layer(Name: " << layer->getName()
<< ") is set to float32 because its input("
<< temp_in->getName() << ") doesn't have dynamic range.";
return false;
return true;
}
}
for (int j = 0; j < layer->getNbOutputs(); j++) {
......@@ -145,10 +164,10 @@ void TensorRTEngine::FreezeNetwork() {
VLOG(1) << "Layer(Name: " << layer->getName()
<< ") is set to float32 because its output("
<< temp_out->getName() << ") doesn't have dynamic range.";
return false;
return true;
}
}
return true;
return false;
};
// If a layer's output is the network's output, or not all of its inputs
// and outputs have scales,
......@@ -157,7 +176,7 @@ void TensorRTEngine::FreezeNetwork() {
int layers_no_int8 = 0;
for (int i = 0; i < network()->getNbLayers(); i++) {
auto layer = network()->getLayer(i);
if (!is_layer_int8(layer)) {
if (layer_int8_fallback(layer)) {
layer->setPrecision(nvinfer1::DataType::kFLOAT);
++layers_no_int8;
}
......@@ -208,7 +227,8 @@ void TensorRTEngine::FreezeNetwork() {
for (auto &input : min_input_shape_) {
#if IS_TRT_VERSION_LT(7000)
// trt6 will check all_of input > 0
if (!(std::all_of(input.second.begin(), input.second.end(),
if (!(std::all_of(input.second.begin(),
input.second.end(),
[](int x) { return x > 0; }) &&
std::all_of(max_input_shape_[input.first].begin(),
max_input_shape_[input.first].end(),
......@@ -225,13 +245,16 @@ void TensorRTEngine::FreezeNetwork() {
<< ", opt: " << Vec2Str(optim_input_shape_[input.first]);
optim_profiles_[i]->setDimensions(
input.first.c_str(), nvinfer1::OptProfileSelector::kMIN,
input.first.c_str(),
nvinfer1::OptProfileSelector::kMIN,
Vec2TRT_Dims(input.second, input.first, true));
optim_profiles_[i]->setDimensions(
input.first.c_str(), nvinfer1::OptProfileSelector::kMAX,
input.first.c_str(),
nvinfer1::OptProfileSelector::kMAX,
Vec2TRT_Dims(max_input_shape_[input.first], input.first, true));
optim_profiles_[i]->setDimensions(
input.first.c_str(), nvinfer1::OptProfileSelector::kOPT,
input.first.c_str(),
nvinfer1::OptProfileSelector::kOPT,
Vec2TRT_Dims(optim_input_shape_[input.first], input.first, true));
}
infer_builder_config_->addOptimizationProfile(optim_profiles_[i]);
......@@ -265,9 +288,10 @@ void TensorRTEngine::FreezeNetwork() {
#endif
PADDLE_ENFORCE_NOT_NULL(
infer_engine_, platform::errors::Fatal(
"Build TensorRT cuda engine failed! Please recheck "
"you configurations related to paddle-TensorRT."));
infer_engine_,
platform::errors::Fatal(
"Build TensorRT cuda engine failed! Please recheck "
"you configurations related to paddle-TensorRT."));
binding_num_ = infer_engine_->getNbBindings();
// reset status for dynamic shape clone
......@@ -282,16 +306,19 @@ void TensorRTEngine::FreezeNetwork() {
nvinfer1::ITensor *TensorRTEngine::DeclareInput(const std::string &name,
nvinfer1::DataType dtype,
const nvinfer1::Dims &dims) {
PADDLE_ENFORCE_EQ(network() != nullptr, true,
PADDLE_ENFORCE_EQ(network() != nullptr,
true,
platform::errors::InvalidArgument(
"The TRT network should be initialized first."));
auto *input = network()->addInput(name.c_str(), dtype, dims);
PADDLE_ENFORCE_NOT_NULL(
input, platform::errors::InvalidArgument("Adding input %s failed in "
"TensorRT inference network. "
"Please recheck your input.",
name));
PADDLE_ENFORCE_EQ(input->isNetworkInput(), true,
input,
platform::errors::InvalidArgument("Adding input %s failed in "
"TensorRT inference network. "
"Please recheck your input.",
name));
PADDLE_ENFORCE_EQ(input->isNetworkInput(),
true,
platform::errors::InvalidArgument(
"Input %s is not the input of TRT inference network. "
"Please recheck your input.",
......@@ -300,22 +327,26 @@ nvinfer1::ITensor *TensorRTEngine::DeclareInput(const std::string &name,
return input;
}
void TensorRTEngine::DeclareOutput(const nvinfer1::ILayer *layer, int offset,
void TensorRTEngine::DeclareOutput(const nvinfer1::ILayer *layer,
int offset,
const std::string &name) {
auto *output = layer->getOutput(offset);
SetITensor(name, output);
PADDLE_ENFORCE_NOT_NULL(
output, platform::errors::InvalidArgument(
"The output %s of TRT engine should not be null.", name));
output,
platform::errors::InvalidArgument(
"The output %s of TRT engine should not be null.", name));
output->setName(name.c_str());
PADDLE_ENFORCE_EQ(output->isNetworkInput(), false,
PADDLE_ENFORCE_EQ(output->isNetworkInput(),
false,
platform::errors::InvalidArgument(
"The output %s of TRT engine should not be the input "
"of the network at the same time.",
name));
network()->markOutput(*output);
PADDLE_ENFORCE_EQ(
output->isNetworkOutput(), true,
output->isNetworkOutput(),
true,
platform::errors::InvalidArgument(
"The output %s of TRT engine should be the output of the network.",
name));
......@@ -324,10 +355,12 @@ void TensorRTEngine::DeclareOutput(const nvinfer1::ILayer *layer, int offset,
void TensorRTEngine::DeclareOutput(const std::string &name) {
auto *output = TensorRTEngine::GetITensor(name);
PADDLE_ENFORCE_NOT_NULL(
output, platform::errors::InvalidArgument(
"The output %s of TRT engine should not be null.", name));
output,
platform::errors::InvalidArgument(
"The output %s of TRT engine should not be null.", name));
output->setName(name.c_str());
PADDLE_ENFORCE_EQ(output->isNetworkInput(), false,
PADDLE_ENFORCE_EQ(output->isNetworkInput(),
false,
platform::errors::InvalidArgument(
"The output %s of TRT engine should not be the input "
"of the network at the same time.",
......@@ -338,17 +371,20 @@ void TensorRTEngine::DeclareOutput(const std::string &name) {
void TensorRTEngine::SetITensor(const std::string &name,
nvinfer1::ITensor *tensor) {
PADDLE_ENFORCE_NOT_NULL(
tensor, platform::errors::InvalidArgument(
"Tensor named %s of TRT engine should not be null.", name));
tensor,
platform::errors::InvalidArgument(
"Tensor named %s of TRT engine should not be null.", name));
PADDLE_ENFORCE_EQ(
0, itensor_map_.count(name),
0,
itensor_map_.count(name),
platform::errors::InvalidArgument(
"Tensor named %s of TRT engine should not be duplicated", name));
itensor_map_[name] = tensor;
}
nvinfer1::ITensor *TensorRTEngine::GetITensor(const std::string &name) {
PADDLE_ENFORCE_EQ(itensor_map_.count(name), true,
PADDLE_ENFORCE_EQ(itensor_map_.count(name),
true,
platform::errors::NotFound(
"Tensor named %s is not found in TRT engine", name));
return itensor_map_[name];
......@@ -365,15 +401,16 @@ float *TensorRTEngine::GetWeightCPUData(const std::string &name,
std::string splitter = "__";
std::string name_with_suffix = name + splitter + name_suffix;
platform::CPUPlace cpu_place;
PADDLE_ENFORCE_EQ(weight_map.count(name_with_suffix), 0,
PADDLE_ENFORCE_EQ(weight_map.count(name_with_suffix),
0,
platform::errors::AlreadyExists(
"The weight named %s is set into the weight map "
"twice in TRT OP converter.",
name_with_suffix));
weight_map[name_with_suffix].reset(new framework::Tensor());
weight_map[name_with_suffix]->Resize(weight_tensor->dims());
paddle::framework::TensorCopySync(*weight_tensor, cpu_place,
weight_map[name_with_suffix].get());
paddle::framework::TensorCopySync(
*weight_tensor, cpu_place, weight_map[name_with_suffix].get());
float *weight_data =
weight_map[name_with_suffix]->mutable_data<float>(cpu_place);
name_suffix_counter += 1;
......@@ -383,21 +420,24 @@ float *TensorRTEngine::GetWeightCPUData(const std::string &name,
int TensorRTEngine::GetRuntimeBatch() { return runtime_batch_; }
nvinfer1::IPluginV2Layer *TensorRTEngine::AddPlugin(
nvinfer1::ITensor *const *inputs, int num_inputs,
nvinfer1::ITensor *const *inputs,
int num_inputs,
plugin::PluginTensorRT *plugin) {
owned_plugin_.emplace_back(plugin);
return network()->addPluginV2(inputs, num_inputs, *plugin);
}
nvinfer1::IPluginV2Layer *TensorRTEngine::AddPluginV2Ext(
nvinfer1::ITensor *const *inputs, int num_inputs,
nvinfer1::ITensor *const *inputs,
int num_inputs,
plugin::PluginTensorRTV2Ext *plugin) {
owned_plugin_v2ext_.emplace_back(plugin);
return network()->addPluginV2(inputs, num_inputs, *plugin);
}
nvinfer1::IPluginV2Layer *TensorRTEngine::AddPluginV2IOExt(
nvinfer1::ITensor *const *inputs, int num_inputs,
nvinfer1::ITensor *const *inputs,
int num_inputs,
nvinfer1::IPluginV2IOExt *plugin) {
owned_plugin_v2ioext_.emplace_back(plugin);
return network()->addPluginV2(inputs, num_inputs, *plugin);
......@@ -406,10 +446,12 @@ nvinfer1::IPluginV2Layer *TensorRTEngine::AddPluginV2IOExt(
void TensorRTEngine::freshDeviceId() {
int count;
cudaGetDeviceCount(&count);
PADDLE_ENFORCE_LT(device_id_, count,
PADDLE_ENFORCE_LT(device_id_,
count,
platform::errors::OutOfRange(
"Device id %d exceeds the current device count: %d.",
device_id_, count));
device_id_,
count));
platform::SetDeviceId(device_id_);
}
......
python/paddle/fluid/tests/unittests/ir/inference/test_trt_convert_slice.py
浏览文件 @ fcc8a87b
......@@ -62,7 +62,7 @@ class TrtConvertSliceTest(TrtLayerAutoScanTest):
for axes in [[0, 1], [1, 3], [2, 3]]:
for starts in [[0, 1]]:
for ends in [[2, 2], [5, 5]]:
for ends in [[2, 2], [5, 5], [1, -1]]:
for decrease_axis in [[], [1], [2], [-1], [-100]]:
for infer_flags in [[-1]]:
dics = [{
......@@ -118,10 +118,6 @@ class TrtConvertSliceTest(TrtLayerAutoScanTest):
return 0, 3
if dynamic_shape == False and len(attrs[0]["decrease_axis"]) != 0:
return 0, 3
if dynamic_shape:
for i in range(len(attrs[0]["starts"])):
if attrs[0]["starts"][i] < 0 or attrs[0]["ends"][i] < 0:
return 0, 3
if not dynamic_shape:
for x in attrs[0]["axes"]:
if x == 0:
......
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