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

add type promotion logic for eager between tensor and tensor #59518

Merged
merged 23 commits into from
Dec 7, 2023
Merged

add type promotion logic for eager between tensor and tensor #59518

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
23 commits
Select commit Hold shift + click to select a range
3cafffa
add eager T+T logic.
zxcd Nov 29, 2023
a1c649a
remove useless file.
zxcd Nov 29, 2023
4ce9034
remove useless line.
zxcd Nov 29, 2023
359a689
fix
zxcd Nov 29, 2023
a92bb8f
fix
zxcd Nov 29, 2023
83ec3e0
update
zxcd Nov 29, 2023
2636832
fix note.
zxcd Nov 29, 2023
7c4b08e
mv common logic to common dir.
zxcd Nov 30, 2023
3a996f8
fix
zxcd Nov 30, 2023
6784ab7
remove deal for int.
zxcd Nov 30, 2023
defb035
remove int.
zxcd Nov 30, 2023
49b4cf4
only for complie
zxcd Nov 30, 2023
25502e6
ignore other type promotion for now.
zxcd Nov 30, 2023
2e47906
new eager logic.
zxcd Nov 30, 2023
a22d435
fix bug, add where.
zxcd Dec 1, 2023
1b95f1f
fix
zxcd Dec 1, 2023
fe3553c
add dtype check, warnning, rename .h
zxcd Dec 4, 2023
dd7ce58
add warnning
zxcd Dec 4, 2023
e6d8630
add more log.
zxcd Dec 4, 2023
7c87518
bug fix
zxcd Dec 4, 2023
acd341a
Merge branch 'develop' into type_promotion_stage1_eager_T_T
zxcd Dec 4, 2023
77597ee
fix by comment, make logic of eager_gen more readable.
zxcd Dec 6, 2023
5f87014
merge to develop
zxcd Dec 6, 2023
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
30 changes: 30 additions & 0 deletions paddle/fluid/eager/api/manual/eager_manual/forwards/multiply_fwd_func.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,9 +19,11 @@
#include "paddle/fluid/eager/eager_amp_auto_cast.h"
#include "paddle/fluid/eager/eager_layout_auto_tune.h"
#include "paddle/fluid/eager/nan_inf_utils.h"
#include "paddle/fluid/eager/type_promotion_utils.h"
#include "paddle/fluid/platform/place.h"
#include "paddle/fluid/platform/profiler/event_tracing.h"
#include "paddle/phi/api/include/sparse_api.h"
#include "paddle/phi/common/type_promotion.h"
#include "paddle/phi/core/flags.h"

PHI_DECLARE_bool(check_nan_inf);
Expand Down Expand Up @@ -56,6 +58,20 @@ paddle::Tensor multiply_ad_func(const paddle::Tensor& x,
}
}

// Type promotion Logic
if (phi::NeedTypePromotion(x.dtype(), y.dtype())) {
VLOG(5) << "got different data type, run type protmotion automatically.";
LOG(WARNING) << "got different data type, run type protmotion "
"automatically, this may cause data type been changed.";
auto op_name = phi::TransToFluidOpName("multiply");
auto promotion_type = phi::GetPromoteDtype(op_name, x.dtype(), y.dtype());

auto new_x = egr::PromoteCast("x", x, promotion_type);
auto new_y = egr::PromoteCast("y", y, promotion_type);

return multiply_ad_func(new_x, new_y);
}

// Layout autotune

if (egr::Controller::Instance().UseLayoutAutoTune()) {
Expand Down Expand Up @@ -388,6 +404,20 @@ paddle::Tensor multiply_ad_func(const paddle::Tensor& x,
}
}

// Type promotion Logic
if (phi::NeedTypePromotion(x.dtype(), y.dtype())) {
VLOG(5) << "got different data type, run type protmotion automatically.";
LOG(WARNING) << "got different data type, run type protmotion "
"automatically, this may cause data type been changed.";
auto op_name = phi::TransToFluidOpName("multiply");
auto promotion_type = phi::GetPromoteDtype(op_name, x.dtype(), y.dtype());

auto new_x = egr::PromoteCast("x", x, promotion_type);
auto new_y = egr::PromoteCast("y", y, promotion_type);

return multiply_ad_func(new_x, new_y);
}

// Layout autotune

if (egr::Controller::Instance().UseLayoutAutoTune()) {
Expand Down
59 changes: 58 additions & 1 deletion paddle/fluid/eager/auto_code_generator/generator/eager_gen.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,6 +75,14 @@
"tanh_triple_grad",
]

# white ops list whose kernel can automaically do type promotion.
# future will get this list from same place with static graph.
type_promote_white_list = {
"add": ["x", "y"],
"subtract": ["x", "y"],
"where": ["x", "y"],
}

# dict of special api that forward api's output will affect bacward api's output
# bacward api's output usually affected by backward api's input
special_prune_dict = {
Expand Down Expand Up @@ -247,6 +255,8 @@ class {} : public egr::GradNodeBase {{
// Dygraph Record Event
{}
// AMP Logic
{}
// Type promotion Logic
{}
// Layout autotune
{}
Expand Down Expand Up @@ -315,6 +325,8 @@ class {} : public egr::GradNodeBase {{
// Dygraph Record Event
{}
// AMP Logic
{}
// Type promotion Logic
{}
// Layout autotune
{}
Expand Down Expand Up @@ -447,7 +459,8 @@ class {} : public egr::GradNodeBase {{
#include "paddle/fluid/eager/api/manual/eager_manual/dygraph_forward_api.h"
#include "paddle/phi/core/flags.h"
#include "paddle/phi/api/lib/data_transform.h"

#include "paddle/fluid/eager/type_promotion_utils.h"
#include "paddle/phi/common/type_promotion.h"
PHI_DECLARE_bool(check_nan_inf);
PHI_DECLARE_string(tensor_operants_mode);
{}
Expand Down Expand Up @@ -512,6 +525,21 @@ class {} : public egr::GradNodeBase {{
}}
}}
"""

TYPE_PROMOTION_LOGIC_TEMPLATE = """ if (phi::NeedTypePromotion({x}.dtype(), {y}.dtype())) {{
VLOG(5) << "got different data type, run type protmotion automatically.";
LOG(WARNING) << "got different data type, run type protmotion automatically, this may cause data type been changed.";
{op_name}
auto promotion_type = phi::GetPromoteDtype(op_name, {x}.dtype(), {y}.dtype());

auto new_{x} = egr::PromoteCast("{x}", {x}, promotion_type);
auto new_{y} = egr::PromoteCast("{y}", {y}, promotion_type);

{return_value}
}}
"""


LAYOUT_LOGIC_TEMPLATE = """
if (egr::Controller::Instance().UseLayoutAutoTune()) {{
paddle::small_vector<std::vector<paddle::Tensor>, egr::kSlotSmallVectorSize> tensors_vector = {};
Expand Down Expand Up @@ -1459,6 +1487,7 @@ def GenerateForwardDefinitionAndDeclaration(self, is_inplaced):
inputs_call_list = ["" for i in range(num_inputs)]

amp_inputs_call_list = ["" for i in range(num_inputs)]
type_promote_inputs_call_list = ["" for i in range(num_inputs)]
amp_tensors_vector_list = []
amp_tensors_vector_optional_list = []
amp_autocast_list = []
Expand All @@ -1470,6 +1499,11 @@ def GenerateForwardDefinitionAndDeclaration(self, is_inplaced):
inputs_call_list[pos] = f"{name}"
amp_inputs_call_list[pos] = f"new_{name}"
is_optional = name in optional_inputs
if forward_api_name in type_promote_white_list:
if name in type_promote_white_list[forward_api_name]:
type_promote_inputs_call_list[pos] = f"new_{name}"
else:
type_promote_inputs_call_list[pos] = f"{name}"
if IsPlainTensorType(ttype):
if is_optional:
if (
Expand Down Expand Up @@ -1804,7 +1838,28 @@ def GenerateForwardDefinitionAndDeclaration(self, is_inplaced):
amp_autocast_list_str,
amp_call_str,
)
# Forward type promotion logic
if forward_api_name in type_promote_white_list:
# only support two inputs
x = type_promote_white_list[forward_api_name][0]
y = type_promote_white_list[forward_api_name][1]
type_promote_inputs_call_args_str = ", ".join(
type_promote_inputs_call_list
)
type_promote_call_list = f"return {forward_ad_function_name}({type_promote_inputs_call_args_str});"

type_promotion_logic_str = TYPE_PROMOTION_LOGIC_TEMPLATE.format(
x=x,
y=y,
op_name=kernel_trans2_op_name_str,
return_value=type_promote_call_list,
)
else:
type_promotion_logic_str = (
"\n VLOG(5) << \" No Type Promotion for {} api. \"; ".format(
forward_ad_function_name
)
)
# Forward layout autotune
layout_autotune_list_str = " ".join(
layout_autotune_list
Expand Down Expand Up @@ -1849,6 +1904,7 @@ def GenerateForwardDefinitionAndDeclaration(self, is_inplaced):
forward_api_name,
dygraph_event_str,
amp_logic_str,
type_promotion_logic_str,
layout_logic_str,
forward_api_name,
before_log_str,
Expand All @@ -1871,6 +1927,7 @@ def GenerateForwardDefinitionAndDeclaration(self, is_inplaced):
forward_api_name,
dygraph_event_str,
amp_logic_str,
type_promotion_logic_str,
layout_logic_str,
inputs_autograd_meta_str,
forward_api_name,
Expand Down
87 changes: 9 additions & 78 deletions paddle/fluid/eager/type_promotion_utils.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,90 +12,21 @@
// See the License for the specific language governing permissions and
// limitations under the License.
#pragma once
#include "paddle/fluid/eager/api/generated/eager_generated/forwards/dygraph_functions.h"

#include "paddle/fluid/eager/api/utils/global_utils.h"
#include "paddle/phi/common/type_promotion.h"

namespace egr {

inline int DataTypeToNum(const phi::DataType& dtype) {
switch (dtype) {
case phi::DataType::UINT8:
return 0;
case phi::DataType::INT8:
return 1;
case phi::DataType::INT16:
return 2;
case phi::DataType::INT32:
return 3;
case phi::DataType::INT64:
return 4;
case phi::DataType::FLOAT16:
return 5;
case phi::DataType::FLOAT32:
return 6;
case phi::DataType::FLOAT64:
return 7;
case phi::DataType::COMPLEX64:
return 8;
case phi::DataType::COMPLEX128:
return 9;
case phi::DataType::BOOL:
return 10;
case phi::DataType::BFLOAT16:
return 11;
default:
PD_THROW("Invalid enum data type for type promote `", dtype, "`.");
}
}

static inline bool is_support_float(phi::DataType dtype) {
if (dtype == phi::DataType::FLOAT16 || dtype == phi::DataType::FLOAT32 ||
dtype == phi::DataType::FLOAT64 || dtype == phi::DataType::BFLOAT16) {
return true;
} else {
return false;
}
}

static inline bool is_support_int(phi::DataType dtype) {
if (dtype == phi::DataType::INT32 || dtype == phi::DataType::INT64) {
return true;
inline paddle::Tensor PromoteCast(const std::string& input_name,
const paddle::Tensor& input,
const phi::DataType& dst_dtype,
bool trace_backward = true) {
if (input.dtype() != dst_dtype) {
return Cast(input, dst_dtype, trace_backward);
} else {
return false;
return input;
}
}

inline static phi::DataType promoteTypes(phi::DataType a, phi::DataType b) {
constexpr auto u1 = phi::DataType::UINT8;
constexpr auto i1 = phi::DataType::INT8;
constexpr auto i2 = phi::DataType::INT16;
constexpr auto i4 = phi::DataType::INT32;
constexpr auto i8 = phi::DataType::INT64;
constexpr auto f2 = phi::DataType::FLOAT16;
constexpr auto f4 = phi::DataType::FLOAT32;
constexpr auto f8 = phi::DataType::FLOAT64;
constexpr auto c4 = phi::DataType::COMPLEX64;
constexpr auto c8 = phi::DataType::COMPLEX128;
constexpr auto b1 = phi::DataType::BOOL;
constexpr auto bf = phi::DataType::BFLOAT16;

static constexpr phi::DataType _promoteTypesLookup[12][12] = {
/* u1 i1 i2 i4 i8 f2 f4 f8 c4 c8 b1 bf*/
/* u1 */ {u1, i2, i2, i4, i8, f2, f4, f8, c4, c8, u1, bf},
/* i1 */ {i2, i1, i2, i4, i8, f2, f4, f8, c4, c8, i1, bf},
/* i2 */ {i2, i2, i2, i4, i8, f2, f4, f8, c4, c8, i2, bf},
/* i4 */ {i4, i4, i4, i4, i8, f2, f4, f8, c4, c8, i4, bf},
/* i8 */ {i8, i8, i8, i8, i8, f2, f4, f8, c4, c8, i8, bf},
/* f2 */ {f2, f2, f2, f2, f2, f2, f4, f8, c4, c8, f2, f4},
/* f4 */ {f4, f4, f4, f4, f4, f4, f4, f8, c4, c8, f4, f4},
/* f8 */ {f8, f8, f8, f8, f8, f8, f8, f8, c8, c8, f8, f8},
/* c4 */ {c4, c4, c4, c4, c4, c4, c4, c8, c4, c8, c4, c4},
/* c8 */ {c8, c8, c8, c8, c8, c8, c8, c8, c8, c8, c8, c8},
/* b1 */ {u1, i1, i2, i4, i8, f2, f4, f8, c4, c8, b1, bf},
/* bf */ {bf, bf, bf, bf, bf, f4, f4, f8, c4, c8, bf, bf},
};

return _promoteTypesLookup[DataTypeToNum(a)][DataTypeToNum(b)];
}

} // namespace egr
22 changes: 14 additions & 8 deletions paddle/fluid/pybind/eager_math_op_patch.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,7 @@ typedef SSIZE_T ssize_t;
#include "paddle/fluid/memory/allocation/mmap_allocator.h"
#include "paddle/fluid/pybind/op_function_common.h"
#include "paddle/fluid/pybind/tensor_py.h"
#include "paddle/phi/common/type_promotion.h"
#include "paddle/phi/kernels/funcs/math_function.h"

namespace paddle {
Expand Down Expand Up @@ -252,10 +253,11 @@ static PyObject* tensor__add__method(TensorObject* self,
ConvertAllInputsToDistTensor(mesh, self_tensor, other_tensor);
}

// 3. promote types or unify right var type to left var
// 3. promote types or unify right var type to left var, float type promotion
// mv to add_ad_func
phi::DataType lhs_dtype = self_tensor.dtype();
phi::DataType rhs_dtype = other_tensor.dtype();
if (lhs_dtype != rhs_dtype) {
if (lhs_dtype != rhs_dtype && !phi::NeedTypePromotion(lhs_dtype, rhs_dtype)) {
// note: only op_type in _supported_promote_complex_types_ should promote
// dtype
if (_complex_dtypes.find(lhs_dtype) != _complex_dtypes.end() ||
Expand Down Expand Up @@ -358,10 +360,11 @@ static PyObject* tensor__sub__method(TensorObject* self,
ConvertAllInputsToDistTensor(mesh, self_tensor, other_tensor);
}

// 3. promote types or unify right var type to left var
// 3. promote types or unify right var type to left var, float type promotion
// mv to subtract_ad_func
phi::DataType lhs_dtype = self_tensor.dtype();
phi::DataType rhs_dtype = other_tensor.dtype();
if (lhs_dtype != rhs_dtype) {
if (lhs_dtype != rhs_dtype && !phi::NeedTypePromotion(lhs_dtype, rhs_dtype)) {
if (_complex_dtypes.find(lhs_dtype) != _complex_dtypes.end() ||
_complex_dtypes.find(rhs_dtype) != _complex_dtypes.end()) {
phi::DataType promote_dtype =
Expand All @@ -386,6 +389,7 @@ static PyObject* tensor__sub__method(TensorObject* self,
other_tensor = cast_ad_func(other_tensor, lhs_dtype);
}
}

// 4. calculation
VLOG(6) << "Calling subtract_ad_func in tensor__sub__method";
{
Expand Down Expand Up @@ -460,10 +464,11 @@ static PyObject* tensor__rsub__method(TensorObject* self,
ConvertAllInputsToDistTensor(mesh, self_tensor, other_tensor);
}

// 3. promote types or unify right var type to left var
// 3. promote types or unify right var type to left var, float type promotion
// mv to subtract_ad_func
phi::DataType lhs_dtype = self_tensor.dtype();
phi::DataType rhs_dtype = other_tensor.dtype();
if (lhs_dtype != rhs_dtype) {
if (lhs_dtype != rhs_dtype && !phi::NeedTypePromotion(lhs_dtype, rhs_dtype)) {
if (_complex_dtypes.find(lhs_dtype) != _complex_dtypes.end() ||
_complex_dtypes.find(rhs_dtype) != _complex_dtypes.end()) {
phi::DataType promote_dtype =
Expand Down Expand Up @@ -568,10 +573,11 @@ static PyObject* tensor__mul__method(TensorObject* self,
ConvertAllInputsToDistTensor(mesh, self_tensor, other_tensor);
}

// 3. promote types or unify right var type to left var
// 3. promote types or unify right var type to left var, float type promotion
// mv to multiply_ad_func
phi::DataType lhs_dtype = self_tensor.dtype();
phi::DataType rhs_dtype = other_tensor.dtype();
if (lhs_dtype != rhs_dtype) {
if (lhs_dtype != rhs_dtype && !phi::NeedTypePromotion(lhs_dtype, rhs_dtype)) {
// note: only op_type in _supported_promote_complex_types_ should promote
// dtype
if (_complex_dtypes.find(lhs_dtype) != _complex_dtypes.end() ||
Expand Down
Loading