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

Remove adding unnecessary shape and slice ops in fused_multi_transformer_pass #48112

Merged
merged 1 commit into from
Nov 21, 2022
Merged

Remove adding unnecessary shape and slice ops in fused_multi_transformer_pass #48112

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
45 changes: 1 addition & 44 deletions paddle/fluid/framework/ir/fuse_multi_transformer_layer_pass.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,34 +62,7 @@ MultiTransformerLayerPattern::operator()(bool enable_int8,
fused_multi_transformer_name, "Out");

if (is_decoder) {
auto shape_repr =
PDNodeName(name_scope_, repr_, id_, "shape_" + std::to_string(i));
node_reprs["shape_" + std::to_string(i)] = shape_repr;
auto* shape = pattern->NewNode(shape_repr)->assert_is_op("shape");

auto shape_out_repr =
PDNodeName(name_scope_, repr_, id_, "shape_out_" + std::to_string(i));
node_reprs["shape_out_" + std::to_string(i)] = shape_out_repr;
auto* shape_out =
pattern->NewNode(shape_out_repr)->assert_is_op_output("shape", "Out");

shape->LinksFrom({src_mask}).LinksTo({shape_out});

auto slice_repr =
PDNodeName(name_scope_, repr_, id_, "slice_" + std::to_string(i));
node_reprs["slice_" + std::to_string(i)] = slice_repr;
auto* slice = pattern->NewNode(slice_repr)->assert_is_op("slice");

auto slice_out_repr =
PDNodeName(name_scope_, repr_, id_, "slice_out_" + std::to_string(i));
node_reprs["slice_out_" + std::to_string(i)] = slice_out_repr;
auto* slice_out =
pattern->NewNode(slice_out_repr)->assert_is_op_output("slice", "Out");

slice->LinksFrom({shape_out}).LinksTo({slice_out});

fused_multi_transformer->LinksFrom({x0, src_mask, slice_out})
.LinksTo({out});
fused_multi_transformer->LinksFrom({x0, src_mask}).LinksTo({out});
} else {
auto cache_kv_repr =
PDNodeName(name_scope_, repr_, id_, "cache_kv_" + std::to_string(i));
Expand Down Expand Up @@ -187,10 +160,6 @@ int FuseMultiTransformerLayerPass::BuildFusion(Graph* graph,
std::vector<Node*> fuse_op_nodes;
std::vector<Node*> out_nodes;

std::vector<std::string> unused_node_prefixes = {
"shape_", "shape_out_", "slice_", "slice_out_"};
std::vector<Node*> unused_nodes;

std::vector<OpDesc*> fuse_op_descs;
std::vector<VariableNameMap> fuse_op_input_var_name_maps;
std::vector<VariableNameMap> fuse_op_output_var_name_maps;
Expand Down Expand Up @@ -219,14 +188,6 @@ int FuseMultiTransformerLayerPass::BuildFusion(Graph* graph,
fill_op_node->Op()->SetInput("Input", {x0->Name()});
IR_NODE_UNLINK(out_nodes[i - 1], fill_op_node);
IR_NODE_LINK_TO(x0, fill_op_node);
} else if (is_decoder && i != 0) {
for (const auto& unused_node_prefix : unused_node_prefixes) {
PDNode* unused_pdnode =
multi_layer_pattern.PatternBase::pattern->RetrieveNode(
node_reprs[unused_node_prefix + std::to_string(i)]);
Node* unused_node = subgraph.at(unused_pdnode);
unused_nodes.push_back(unused_node);
}
}
}

Expand Down Expand Up @@ -293,10 +254,6 @@ int FuseMultiTransformerLayerPass::BuildFusion(Graph* graph,
std::unordered_set<const Node*> marked_fuse_op_nodes(
fuse_op_nodes.begin() + 1, fuse_op_nodes.end());

if (is_decoder) {
marked_fuse_op_nodes.insert(unused_nodes.begin(), unused_nodes.end());
}

GraphSafeRemoveNodes(graph, marked_fuse_op_nodes);
++fusion_count;
};
Expand Down
216 changes: 111 additions & 105 deletions paddle/fluid/framework/ir/fused_multi_transformer_decoder_pass.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1146,35 +1146,7 @@ int FusedMultiTransformerDecoderPass::BuildFusion(Graph* graph,
auto cache_kv_name = "cache_kv" + std::to_string(layer_idx);
fused_multi_transformer_op_desc.SetInput("CacheKV", {cache_kv_name});

VarDesc shape_out_desc("shape_out." + std::to_string(layer_idx));
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out." + std::to_string(layer_idx));
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

fused_multi_transformer_op_desc.SetInput("TimeStep", {slice_out->Name()});
fused_multi_transformer_op_desc.SetInput("TimeStep", {"slice_out.0"});

// Out Linear input
fused_multi_transformer_op_desc.SetInput("OutLinearW",
Expand Down Expand Up @@ -1219,12 +1191,42 @@ int FusedMultiTransformerDecoderPass::BuildFusion(Graph* graph,
IR_NODE_LINK_TO(eltadd0_b, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_qk_b, fused_multi_transformer);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
if (layer_idx == 0) {
VarDesc shape_out_desc("shape_out.0");
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out.0");
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
}

IR_NODE_LINK_TO(matmul_linear_w, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_linear_b, fused_multi_transformer);
Expand Down Expand Up @@ -1789,35 +1791,7 @@ int FusedMultiTransformerDecoderFuseQKVPass::BuildFusion(
auto cache_kv_name = "cache_kv" + std::to_string(layer_idx);
fused_multi_transformer_op_desc.SetInput("CacheKV", {cache_kv_name});

VarDesc shape_out_desc("shape_out." + std::to_string(layer_idx));
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out." + std::to_string(layer_idx));
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

fused_multi_transformer_op_desc.SetInput("TimeStep", {slice_out->Name()});
fused_multi_transformer_op_desc.SetInput("TimeStep", {"slice_out.0"});

// Out Linear input
fused_multi_transformer_op_desc.SetInput("OutLinearW",
Expand Down Expand Up @@ -1862,12 +1836,42 @@ int FusedMultiTransformerDecoderFuseQKVPass::BuildFusion(
IR_NODE_LINK_TO(eltadd0_b, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_qk_b, fused_multi_transformer);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
if (layer_idx == 0) {
VarDesc shape_out_desc("shape_out.0");
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out.0");
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
}

IR_NODE_LINK_TO(matmul_linear_w, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_linear_b, fused_multi_transformer);
Expand Down Expand Up @@ -2405,35 +2409,7 @@ int MultiDevicesFusedMultiTransformerDecoderFuseQKVPass::BuildFusion(
auto cache_kv_name = "cache_kv" + std::to_string(layer_idx);
fused_multi_transformer_op_desc.SetInput("CacheKV", {cache_kv_name});

VarDesc shape_out_desc("shape_out." + std::to_string(layer_idx));
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out." + std::to_string(layer_idx));
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

fused_multi_transformer_op_desc.SetInput("TimeStep", {slice_out->Name()});
fused_multi_transformer_op_desc.SetInput("TimeStep", {"slice_out.0"});

// Out Linear input
fused_multi_transformer_op_desc.SetInput("OutLinearW",
Expand Down Expand Up @@ -2483,12 +2459,42 @@ int MultiDevicesFusedMultiTransformerDecoderFuseQKVPass::BuildFusion(
IR_NODE_LINK_TO(eltadd0_b, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_qk_b, fused_multi_transformer);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
if (layer_idx == 0) {
VarDesc shape_out_desc("shape_out.0");
shape_out_desc.SetDataType(proto::VarType::INT32);
shape_out_desc.SetPersistable(false);
auto* shape_out = graph->CreateVarNode(&shape_out_desc);

OpDesc shape_op_desc(layer_norm->Op()->Block());
shape_op_desc.SetType("shape");
shape_op_desc.SetInput("Input", {eltadd_qk_b->Name()});
shape_op_desc.SetOutput("Out", {shape_out->Name()});
auto* shape_op = graph->CreateOpNode(&shape_op_desc);

VarDesc slice_out_desc("slice_out.0");
slice_out_desc.SetDataType(proto::VarType::INT32);
slice_out_desc.SetPersistable(false);
auto* slice_out = graph->CreateVarNode(&slice_out_desc);

OpDesc slice_op_desc(layer_norm->Op()->Block());
slice_op_desc.SetType("slice");
slice_op_desc.SetInput("Input", {shape_out->Name()});
slice_op_desc.SetOutput("Out", {slice_out->Name()});
std::vector<int> axes = {0};
std::vector<int> starts = {3};
std::vector<int> ends = {4};
slice_op_desc.SetAttr("axes", axes);
slice_op_desc.SetAttr("starts", starts);
slice_op_desc.SetAttr("ends", ends);
auto* slice_op = graph->CreateOpNode(&slice_op_desc);

// TimeStep link
IR_NODE_LINK_TO(eltadd_qk_b, shape_op);
IR_NODE_LINK_TO(shape_op, shape_out);
IR_NODE_LINK_TO(shape_out, slice_op);
IR_NODE_LINK_TO(slice_op, slice_out);
IR_NODE_LINK_TO(slice_out, fused_multi_transformer)
}

IR_NODE_LINK_TO(matmul_linear_w, fused_multi_transformer);
IR_NODE_LINK_TO(eltadd_linear_b, fused_multi_transformer);
Expand Down
1 change: 1 addition & 0 deletions paddle/fluid/inference/api/paddle_pass_builder.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -177,6 +177,7 @@ const std::vector<std::string> kGpuLowerPrecisionPasses{
"fused_multi_transformer_decoder_fuse_qkv_pass",
"multi_devices_fused_multi_transformer_encoder_fuse_qkv_pass",
"multi_devices_fused_multi_transformer_decoder_fuse_qkv_pass",
"fuse_multi_transformer_layer_pass",
"gpu_cpu_map_matmul_v2_to_mul_pass",
"gpu_cpu_map_matmul_v2_to_matmul_pass",
"fc_fuse_pass",
Expand Down