ShiyuW
/
mindspore2022
forked from huawei/mindspore2022
0
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

!20837 adjust and update irpass 

Merge pull request !20837 from huangbingjian/modify_pass
This commit is contained in:
i-robot 2021-07-27 11:50:44 +00:00 committed by Gitee
parent ce2ebfd371 4515d8dc10
commit cc0e1cd6d2
10 changed files with 44 additions and 156 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
mindspore/ccsrc/frontend/optimizer/irpass.cc
View File

@ -49,7 +49,6 @@
#include "frontend/optimizer/irpass/sparse_tensor_eliminate.h"
#include "frontend/optimizer/irpass/switch_or_switch_layer_defer_inline.h"
#include "frontend/optimizer/irpass/call_graph_tuple_transform.h"
#include "frontend/optimizer/irpass/bool_scalar_eliminate.h"
#include "frontend/optimizer/irpass/recompute_prepare.h"
namespace mindspore {
@ -256,8 +255,6 @@ OptimizeIRPassLib::OptimizeIRPassLib() {
switch_layer_defer_inline_ =
MakeSubstitution(std::make_shared<SwitchLayerDeferInline>(), "switch_layer_defer_inline", prim::kPrimSwitchLayer);
bool_scalar_eliminate_ = MakeSubstitution(std::make_shared<BoolScalarEliminate>(), "bool_scalar_eliminate_", IsCNode);
// recompute
set_cell_output_no_recompute_ = MakeSubstitution(std::make_shared<SetCellOutputNoRecompute>(),
"set_cell_output_no_recompute", IsValueNode<FuncGraph>);

3
mindspore/ccsrc/frontend/optimizer/irpass.h
View File

@ -157,9 +157,6 @@ class OptimizeIRPassLib {
// Pynative Eliminate
SubstitutionPtr pynative_eliminate_;
// Eliminate getattr bool scalar
SubstitutionPtr bool_scalar_eliminate_;
// Recompute
SubstitutionPtr set_cell_output_no_recompute_;
};

58
mindspore/ccsrc/frontend/optimizer/irpass/bool_scalar_eliminate.cc
View File

@ -1,58 +0,0 @@
/**
* Copyright 2021 Huawei Technologies Co., Ltd
*
* 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 "frontend/optimizer/irpass/bool_scalar_eliminate.h"
#include "frontend/optimizer/optimizer.h"
namespace mindspore {
namespace opt {
namespace irpass {
AnfNodePtr BoolScalarEliminate::operator()(const OptimizerPtr &optimizer, const AnfNodePtr &node) {
auto cnode = node->cast<CNodePtr>();
if (cnode == nullptr) {
return nullptr;
}
if (!cnode->IsApply(prim::kPrimGetAttr)) {
return nullptr;
}
auto vnode = cnode->input(1)->cast<ValueNodePtr>();
if (vnode == nullptr) {
return nullptr;
}
if (!vnode->value()->isa<BoolImm>()) {
return nullptr;
}
auto res = optimizer->resource();
auto manager = res->manager();
auto &node_users = manager->node_users();
auto iter = node_users.find(node);
if (iter == node_users.end()) {
return nullptr;
}
AnfNodeIndexSet node_idx_set = iter->second;
for (auto &item : node_idx_set) {
(void)manager->Replace(item.first, vnode);
}
return nullptr;
}
} // namespace irpass
} // namespace opt
} // namespace mindspore

36
mindspore/ccsrc/frontend/optimizer/irpass/bool_scalar_eliminate.h
View File

@ -1,36 +0,0 @@
/**
* Copyright 2021 Huawei Technologies Co., Ltd
*
* 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.
*/
#ifndef MINDSPORE_CCSRC_FRONTEND_OPTIMIZER_IRPASS_BOOL_SCALAR_ELIMINATE_H
#define MINDSPORE_CCSRC_FRONTEND_OPTIMIZER_IRPASS_BOOL_SCALAR_ELIMINATE_H
#include "ir/func_graph.h"
#include "frontend/optimizer/optimizer_caller.h"
#include "ir/pattern_matcher.h"
namespace mindspore {
namespace opt {
namespace irpass {
class BoolScalarEliminate : public OptimizerCaller {
public:
AnfNodePtr operator()(const OptimizerPtr &optimizer, const AnfNodePtr &node) override;
};
} // namespace irpass
} // namespace opt
} // namespace mindspore
#endif // MINDSPORE_CCSRC_FRONTEND_OPTIMIZER_IRPASS_BOOL_SCALAR_ELIMINATE_H

43
mindspore/ccsrc/frontend/optimizer/irpass/gradient_eliminate.cc
View File

@ -65,36 +65,27 @@ AnfNodePtr ExpandJ(const ValueNodePtr &vnode, const pipeline::ResourceBasePtr &r
}
} // namespace internal
bool ExpandJPrim::operator()(const FuncGraphPtr &func_graph, const OptimizerPtr &optimizer) {
auto manager = optimizer->manager();
// Search all j nodes.
GetJPrim(manager);
// Get j nodes that don't have embed j nodes.
std::vector<CNodePtr> todo;
// If graph also contains J(FuncGraph) or J(Primitive), then ignore this graph.
// ExpandJ innermost graph or primitive first.
std::copy_if(j_nodes_.begin(), j_nodes_.end(), std::back_inserter(todo),
[](const CNodePtr &j_node) { return !internal::CheckIfEmbedJ(j_node); });
// Expand j nodes that don't have embed j nodes.
bool change = false;
for (auto &j_node : todo) {
auto expanded_j = internal::ExpandJ(j_node->input(1)->cast<ValueNodePtr>(), optimizer->resource());
manager->Replace(j_node, expanded_j);
change = true;
}
return change;
}
bool ExpandJPrim::operator()(const FuncGraphPtr &root, const OptimizerPtr &optimizer) {
AnfNodePtr ret = root->get_return();
MS_EXCEPTION_IF_NULL(ret);
std::vector<AnfNodePtr> all_nodes = DeepScopedGraphSearch(ret);
void ExpandJPrim::GetJPrim(const FuncGraphManagerPtr &manager) {
j_nodes_.clear();
for (auto &fg : manager->func_graphs()) {
std::vector<AnfNodePtr> toposet = TopoSort(fg->get_return());
for (const auto &node : toposet) {
if (IsPrimitiveCNode(node, prim::kPrimJ)) {
j_nodes_.push_back(node->cast<CNodePtr>());
bool change = false;
auto manager = optimizer->manager();
for (auto &node : all_nodes) {
if (IsPrimitiveCNode(node, prim::kPrimJ)) {
auto j_node = node->cast<CNodePtr>();
// If graph also contains J(FuncGraph) or J(Primitive), then ignore this graph.
// ExpandJ innermost graph or primitive first.
if (internal::CheckIfEmbedJ(j_node)) {
continue;
}
auto expanded_j = internal::ExpandJ(j_node->input(1)->cast<ValueNodePtr>(), optimizer->resource());
manager->Replace(j_node, expanded_j);
change = true;
}
}
return change;
}
} // namespace irpass
} // namespace opt

6
mindspore/ccsrc/frontend/optimizer/irpass/gradient_eliminate.h
View File

@ -36,11 +36,7 @@ class ExpandJPrim {
public:
ExpandJPrim() = default;
virtual ~ExpandJPrim() = default;
bool operator()(const FuncGraphPtr &func_graph, const OptimizerPtr &optimizer);
void GetJPrim(const FuncGraphManagerPtr &manager);
private:
std::vector<CNodePtr> j_nodes_;
bool operator()(const FuncGraphPtr &root, const OptimizerPtr &optimizer);
};
} // namespace irpass
} // namespace opt

30
mindspore/ccsrc/frontend/optimizer/irpass/incorporate_getitem.h
View File

@ -1,5 +1,5 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
* Copyright 2020-2021 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
@ -40,7 +40,7 @@ class GetitemTransform {
GetitemTransform() : cache_() {}
~GetitemTransform() = default;
FuncGraphPtr operator()(const FuncGraphPtr &fg, int64_t idx) {
FuncGraphPtr operator()(const AnfNodePtr &node, const FuncGraphPtr &fg, int64_t idx) {
if (cache_.find(fg) == cache_.end()) {
cache_[fg] = {};
}
@ -61,7 +61,11 @@ class GetitemTransform {
}
new_fg->set_output(cnode->input(ids));
} else {
new_fg->set_output(new_fg->NewCNode({NewValueNode(prim::kPrimTupleGetItem), output, NewValueNode(idx)}));
auto idx_node = NewValueNode(idx);
idx_node->set_abstract(std::make_shared<abstract::AbstractScalar>(idx));
auto output_node = new_fg->NewCNode({NewValueNode(prim::kPrimTupleGetItem), output, idx_node});
output_node->set_abstract(node->abstract());
new_fg->set_output(output_node);
}
cache[idx] = new_fg;
@ -78,7 +82,7 @@ class GetItemTransformACrossGraph {
GetItemTransformACrossGraph() : cache_() {}
~GetItemTransformACrossGraph() = default;
FuncGraphPtr operator()(const FuncGraphPtr &fg, int64_t idx) {
FuncGraphPtr operator()(const AnfNodePtr &node, const FuncGraphPtr &fg, int64_t idx) {
if (cache_.find(fg) == cache_.end()) {
cache_[fg] = {};
}
@ -109,7 +113,11 @@ class GetItemTransformACrossGraph {
}
new_fg->set_output(cnode->input(ids));
} else {
new_fg->set_output(new_fg->NewCNode({NewValueNode(prim::kPrimTupleGetItem), output, NewValueNode(idx)}));
auto idx_node = NewValueNode(idx);
idx_node->set_abstract(std::make_shared<abstract::AbstractScalar>(idx));
auto output_node = new_fg->NewCNode({NewValueNode(prim::kPrimTupleGetItem), output, idx_node});
output_node->set_abstract(node->abstract());
new_fg->set_output(output_node);
}
cache[idx] = new_fg_outer;
@ -136,7 +144,7 @@ class IncorporateGetitem : public AnfVisitor {
return nullptr;
}
auto new_fg = getitem_transform_(fg_, idx_);
auto new_fg = getitem_transform_(node, fg_, idx_);
(void)args_.insert(args_.begin(), NewValueNode(new_fg));
auto new_node = node->func_graph()->NewCNode(args_);
// Check if the another only usage of {G, Xs} is UpdateState{s, {G, Xs}}, if yes, replace
@ -213,7 +221,7 @@ class IncorporateGetitemDepend : public AnfVisitor {
return nullptr;
}
auto new_fg = getitem_transform_(fg_, idx_);
auto new_fg = getitem_transform_(node, fg_, idx_);
(void)args_.insert(args_.begin(), NewValueNode(new_fg));
auto new_fg_cnode = node->func_graph()->NewCNode(args_);
AnfNodePtr new_depend_cnode;
@ -333,8 +341,8 @@ class IncorporateGetitemSwitch : public AnfVisitor {
!ExistEnvNodeInTupleItem(g2_) && !has_env_type) {
return nullptr;
}
auto new_g1 = getitem_transform_(g1_, idx_);
auto new_g2 = getitem_transform_(g2_, idx_);
auto new_g1 = getitem_transform_(node, g1_, idx_);
auto new_g2 = getitem_transform_(node, g2_, idx_);
auto sw_node = fg->NewCNode({NewValueNode(prim::kPrimSwitch), x_, NewValueNode(new_g1), NewValueNode(new_g2)});
(void)args_.insert(args_.begin(), sw_node);
@ -463,7 +471,7 @@ class IncorporateGetitemSwitchLayerA : public AnfVisitor {
std::vector<AnfNodePtr> layers;
for (auto &graph : graphs_) {
auto fg_transform = getitem_transform_(graph, idx_);
auto fg_transform = getitem_transform_(node, graph, idx_);
if (fg_transform == nullptr) {
return nullptr;
}
@ -554,7 +562,7 @@ class IncorporateGetitemSwitchLayerB : public AnfVisitor {
std::vector<AnfNodePtr> layers;
for (auto &graph : graphs_) {
auto fg_transform = getitem_transform_(graph, idx_);
auto fg_transform = getitem_transform_(node, graph, idx_);
if (fg_transform == nullptr) {
return nullptr;
}

8
mindspore/ccsrc/frontend/optimizer/irpass/partial_eliminate.h
View File

@ -159,10 +159,10 @@ class ChoicePartialEliminater : public AnfVisitor {
if (fg->func_graph_cnodes_index().size() != 1) {
// If a graph is used by 2 or more partial nodes at the same time, clone the graph.
auto new_fg = BasicClone(fg);
auto new_fg_node = NewValueNode(new_fg);
MS_EXCEPTION_IF_NULL(fg->manager());
fg->manager()->Replace(fg_node, new_fg_node);
fg_list_[i] = new_fg_node;
auto manager = fg->manager();
MS_EXCEPTION_IF_NULL(manager);
manager->AddFuncGraph(new_fg);
fg_node->cast<ValueNodePtr>()->set_value(new_fg);
}
}
return true;

2
mindspore/ccsrc/frontend/optimizer/irpass/specialize_transform.h
View File

@ -60,7 +60,7 @@ class SpecializeTransform {
new_params.push_back(params[i]);
continue;
}
// replace the parameter with arg.
// replace the parameter with arg in new_fg without changing origin func_graph.
mng->Replace(params[i], NewReplaceValueNode(need_eliminate_args[i]));
}
mng->SetParameters(new_fg, new_params);

11
mindspore/ccsrc/pipeline/jit/pass.cc
View File

@ -125,14 +125,8 @@ FuncGraphPtr PrimBpOptPassStep1(const opt::irpass::OptimizeIRPassLib &irpass, co
irpass.inline_,
});
opt::OptPassConfig bool_scalar_eliminate = opt::OptPassConfig({
irpass.bool_scalar_eliminate_,
});
OptPassGroupMap map({{"ad_eliminate", pynative_eliminate},
{"ad_inline", inline_opt},
{"bool_scalar_eliminate", bool_scalar_eliminate},
{"ad_switch_simplify", switch_simplify}});
OptPassGroupMap map(
{{"ad_eliminate", pynative_eliminate}, {"ad_inline", inline_opt}, {"ad_switch_simplify", switch_simplify}});
auto prim_bprop_opt_step_1 = opt::Optimizer::MakeOptimizer("prim_bprop_opt_step_1", res, map);
FuncGraphPtr func_graph = res->func_graph();
@ -320,7 +314,6 @@ OptPassGroupMap GetOptPassesA(const opt::irpass::OptimizeIRPassLib &irpass) {
false, true);
opt::OptPassConfig accelerated_algorithm = opt::OptPassConfig({irpass.less_batch_normalization_});
opt::OptPassConfig virtual_dataset = opt::OptPassConfig({irpass.virtual_dataset_eliminate_});
opt::OptPassConfig after_resolve_pass =
opt::OptPassConfig({irpass.get_make_ref_eliminate_, irpass.replace_old_param_});