211 lines
7.6 KiB
C++
211 lines
7.6 KiB
C++
/* Copyright (c) 2021 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 the License for the specific language governing permissions and
|
|
limitations under the License. */
|
|
#include "paddle/phi/api/lib/kernel_dispatch.h"
|
|
#include <glog/logging.h>
|
|
#include "paddle/phi/core/tensor_array.h"
|
|
#ifdef _MSC_VER
|
|
#include <intrin.h>
|
|
#endif
|
|
|
|
#include "paddle/phi/api/include/context_pool.h"
|
|
#include "paddle/phi/core/compat/convert_utils.h"
|
|
#include "paddle/phi/core/distributed/auto_parallel/dist_tensor.h"
|
|
#include "paddle/phi/core/string_tensor_utils.h"
|
|
#include "paddle/phi/core/tensor_utils.h"
|
|
#ifdef PADDLE_WITH_CUSTOM_DEVICE
|
|
#include "paddle/phi/backends/device_manager.h"
|
|
#endif
|
|
|
|
namespace paddle::experimental::detail {
|
|
|
|
// We need judge whether the allocation is nullptr,
|
|
// whether the allocation is initialized, so we need GetHolder method
|
|
bool HasAllocation(const phi::TensorBase& t) {
|
|
if (phi::DenseTensor::classof(&t)) {
|
|
return phi::DenseTensorUtils::GetHolder(
|
|
static_cast<const phi::DenseTensor&>(t)) != nullptr;
|
|
} else if (phi::SelectedRows::classof(&t)) {
|
|
return phi::DenseTensorUtils::GetHolder(
|
|
static_cast<const phi::SelectedRows&>(t).value()) != nullptr;
|
|
} else if (phi::SparseCsrTensor::classof(&t)) {
|
|
return phi::DenseTensorUtils::GetHolder(
|
|
static_cast<const phi::SparseCsrTensor&>(t)
|
|
.non_zero_elements()) != nullptr;
|
|
} else if (phi::SparseCooTensor::classof(&t)) {
|
|
return phi::DenseTensorUtils::GetHolder(
|
|
static_cast<const phi::SparseCooTensor&>(t)
|
|
.non_zero_elements()) != nullptr;
|
|
} else if (phi::StringTensor::classof(&t)) {
|
|
return phi::StringTensorUtils::GetHolder(
|
|
static_cast<const phi::StringTensor&>(t)) != nullptr;
|
|
} else if (phi::distributed::DistTensor::classof(&t)) {
|
|
return static_cast<const phi::distributed::DistTensor&>(t).defined();
|
|
} else if (phi::TensorArray::classof(&t)) {
|
|
return t.has_allocation();
|
|
} else {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
BackendSet GetTensorBackendSet(const phi::TensorBase& t) {
|
|
if (HasAllocation(t) && t.place().GetType() != AllocationType::UNDEFINED) {
|
|
#ifdef PADDLE_WITH_CUSTOM_DEVICE
|
|
// See Note [ Why `SetDevice` when parsing custom place? ]
|
|
if (t.place().GetType() == AllocationType::CUSTOM) {
|
|
phi::DeviceManager::SetDevice(t.place());
|
|
}
|
|
#endif
|
|
phi::Backend backend_key = phi::TransToPhiBackend(t.place());
|
|
BackendSet backend_set(backend_key);
|
|
VLOG(10) << "update BackendSet by tensor: add [" << backend_key << "]";
|
|
#ifdef PADDLE_WITH_CUSTOM_DEVICE
|
|
if ((backend_key != Backend::CPU) && phi::DenseTensor::classof(&t) &&
|
|
static_cast<const phi::DenseTensor&>(t).meta().use_gpudnn) {
|
|
backend_set = backend_set | BackendSet(Backend::GPUDNN);
|
|
} else if ((backend_key != Backend::CPU) &&
|
|
phi::distributed::DistTensor::classof(&t) &&
|
|
static_cast<const phi::distributed::DistTensor&>(t)
|
|
.value()
|
|
.meta()
|
|
.use_gpudnn) {
|
|
backend_set = backend_set | BackendSet(Backend::GPUDNN);
|
|
}
|
|
#else
|
|
if ((backend_key == Backend::GPU) && phi::DenseTensor::classof(&t) &&
|
|
static_cast<const phi::DenseTensor&>(t).meta().use_gpudnn) {
|
|
backend_set = backend_set | BackendSet(Backend::GPUDNN);
|
|
} else if ((backend_key == Backend::GPU) &&
|
|
phi::distributed::DistTensor::classof(&t) &&
|
|
static_cast<const phi::distributed::DistTensor&>(t)
|
|
.value()
|
|
.meta()
|
|
.use_gpudnn) {
|
|
backend_set = backend_set | BackendSet(Backend::GPUDNN);
|
|
}
|
|
#endif
|
|
return backend_set;
|
|
}
|
|
return BackendSet(Backend::UNDEFINED);
|
|
}
|
|
|
|
std::size_t CountLeadingZeros(uint32_t val) {
|
|
if (val == 0) {
|
|
return 32;
|
|
}
|
|
#if defined(__clang__) || defined(__GNUC__)
|
|
return __builtin_clz(val);
|
|
#elif defined(_MSC_VER)
|
|
// windows don't have built-in clz/ctz function
|
|
DWORD Index;
|
|
_BitScanReverse(&Index, val);
|
|
return (uint32_t)Index ^ 31;
|
|
#else
|
|
std::size_t zero_bits = 0;
|
|
for (std::size_t shift = 32 >> 1; shift; shift >>= 1) {
|
|
uint32_t tmp = val >> shift;
|
|
if (tmp) {
|
|
val = tmp;
|
|
} else {
|
|
zero_bits |= shift;
|
|
}
|
|
}
|
|
return zero_bits;
|
|
#endif
|
|
}
|
|
|
|
} // namespace paddle::experimental::detail
|
|
namespace paddle::experimental {
|
|
|
|
phi::DeviceContext* GetDeviceContextByBackend(phi::Backend backend) {
|
|
auto& pool = paddle::experimental::DeviceContextPool::Instance();
|
|
return pool.GetMutable(phi::TransToPhiPlace(backend));
|
|
}
|
|
|
|
DataType ParseDataType(DataType dtype) { return dtype; }
|
|
DataType ParseDataType(const Tensor& tensor) { return tensor.type(); }
|
|
DataType ParseDataType(const std::vector<Tensor>& tensors) {
|
|
if (tensors.empty()) {
|
|
return DataType::UNDEFINED;
|
|
}
|
|
DataType dtype = tensors[0].type();
|
|
auto n = tensors.size();
|
|
for (size_t i = 1; i < n; ++i) {
|
|
if (tensors[i].type() != dtype) {
|
|
PADDLE_THROW(common::errors::InvalidArgument(
|
|
"The data_type of input tensor in list isn't consistent, "
|
|
"the first tensor is %s, but %dth tensor is %s.",
|
|
dtype,
|
|
i,
|
|
tensors[i].type()));
|
|
}
|
|
}
|
|
return dtype;
|
|
}
|
|
|
|
DataType ParseDataTypeWithInputOrder(DataType dtype, const Tensor& tensor) {
|
|
return dtype != DataType::UNDEFINED ? dtype : ParseDataType(tensor);
|
|
}
|
|
|
|
Backend ParseBackend(const Place& place) {
|
|
#ifdef PADDLE_WITH_CUSTOM_DEVICE
|
|
/**
|
|
* [ Why `SetDevice` when parsing custom place? ]
|
|
* Users are able to call C++ APIs under customOP + customDevice scenario. To
|
|
* make sure `GetDevice` function outputs the accurate place when executing
|
|
* `GetDeviceContextByBackend` function in C++ API, we need to call
|
|
* `SetDevice` first. However, in dygraph mode, `SetDevice` is called at
|
|
* CPython level and calling C++ API directly in customOP cannot reach
|
|
* CPython. Hence, we need to manually set the device here.
|
|
*/
|
|
if (place.GetType() == AllocationType::CUSTOM) {
|
|
phi::DeviceManager::SetDevice(place);
|
|
}
|
|
#endif
|
|
return phi::TransToPhiBackend(place);
|
|
}
|
|
Backend ParseBackend(const Tensor& tensor) {
|
|
Backend backend_key = phi::TransToPhiBackend(tensor.place());
|
|
#ifdef PADDLE_WITH_CUSTOM_DEVICE
|
|
if ((backend_key != Backend::CPU) &&
|
|
phi::DenseTensor::classof(tensor.impl().get()) &&
|
|
static_cast<phi::DenseTensor*>(tensor.impl().get())->meta().use_gpudnn) {
|
|
return Backend::GPUDNN;
|
|
}
|
|
#else
|
|
if ((backend_key == Backend::GPU) &&
|
|
phi::DenseTensor::classof(tensor.impl().get()) &&
|
|
static_cast<phi::DenseTensor*>(tensor.impl().get())->meta().use_gpudnn) {
|
|
return Backend::GPUDNN;
|
|
}
|
|
#endif
|
|
VLOG(8) << "[ParseBackend] Select Backend::" << backend_key;
|
|
return backend_key;
|
|
}
|
|
|
|
Backend ParseBackendWithInputOrder(const Place& place, const Tensor& tensor) {
|
|
return place.GetType() != phi::AllocationType::UNDEFINED
|
|
? ParseBackend(place)
|
|
: ParseBackend(tensor);
|
|
}
|
|
|
|
phi::DataLayout ParseLayout(phi::DataLayout layout) { return layout; }
|
|
phi::DataLayout ParseLayout(const Tensor& tensor) { return tensor.layout(); }
|
|
|
|
phi::DataLayout ParseLayoutWithInputOrder(phi::DataLayout layout,
|
|
const Tensor& tensor) {
|
|
return layout != phi::DataLayout::UNDEFINED ? layout : ParseLayout(tensor);
|
|
}
|
|
|
|
} // namespace paddle::experimental
|