From e94db381baa469d73806e4921a87daab260f5060 Mon Sep 17 00:00:00 2001
From: dzhwinter <dongzhihong01@baidu.com>
Date: Mon, 8 Jan 2018 14:05:45 +0800
Subject: [PATCH] Feature/add shared layout (#7233)

* "reuse ShareLoD with no regret"

* "removed base class shareLayout"

* "fix CI"
---
 paddle/framework/data_transform.cc |  5 ++---
 paddle/framework/op_desc.cc        |  1 +
 paddle/framework/operator.cc       | 19 +++++++++++++++++++
 3 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/paddle/framework/data_transform.cc b/paddle/framework/data_transform.cc
index 55825c5b7d2..fed958db158 100644
--- a/paddle/framework/data_transform.cc
+++ b/paddle/framework/data_transform.cc
@@ -146,9 +146,6 @@ void TransDataLayout(const std::vector<int>& axis,
   auto* dst = out->GetMutable<Tensor>();
   PADDLE_ENFORCE(arity(src.dims()) == 4, "Input Arity Only Suppport 4!");
 
-  auto place = kernel_pair.second.place_;
-  CopyFrom(src, place, *ctx, dst);
-
   auto src_dim = src.dims();
   std::vector<int64_t> dst_dim;
 
@@ -158,6 +155,8 @@ void TransDataLayout(const std::vector<int>& axis,
   }
 
   dst->Resize(make_ddim(dst_dim));
+  auto place = kernel_pair.second.place_;
+  dst->mutable_data(place, src.type());
 
   auto src_type = kernel_pair.first.data_type_;
   framework::VisitDataType(src_type, CastDataLayout(ctx, axis, src, dst));
diff --git a/paddle/framework/op_desc.cc b/paddle/framework/op_desc.cc
index e02e572af2c..47c91290e4b 100644
--- a/paddle/framework/op_desc.cc
+++ b/paddle/framework/op_desc.cc
@@ -66,6 +66,7 @@ class CompileTimeInferShapeContext : public InferShapeContext {
                       out);
     out_var->SetLoDLevel(in_var->GetLoDLevel());
   }
+
   bool IsRuntime() const override;
 
  protected:
diff --git a/paddle/framework/operator.cc b/paddle/framework/operator.cc
index adc85b1049f..a1f1be5f342 100644
--- a/paddle/framework/operator.cc
+++ b/paddle/framework/operator.cc
@@ -417,6 +417,25 @@ class RuntimeInferShapeContext : public InferShapeContext {
     auto in_tensor = in_var->Get<LoDTensor>();
     auto* out_tensor = out_var->GetMutable<LoDTensor>();
     out_tensor->set_lod(in_tensor.lod());
+
+    // TODO(dzhwinter) : reuse ShareLoD in most operators.
+    // Need to call ShareLayout explicitly in sequence related ops.
+    // Shall we have a better method to shared info between in/out Tensor?
+    out_tensor->set_layout(in_tensor.layout());
+  }
+
+  void ShareLayout(const std::string& in, const std::string& out, size_t i = 0,
+                   size_t j = 0) const {
+    PADDLE_ENFORCE_LT(i, Inputs(in).size());
+    PADDLE_ENFORCE_LT(j, Outputs(out).size());
+    Variable* in_var = scope_.FindVar(Inputs(in)[i]);
+    Variable* out_var = scope_.FindVar(Outputs(out)[j]);
+    if (!in_var->IsType<LoDTensor>()) return;
+    PADDLE_ENFORCE(out_var->IsType<LoDTensor>(),
+                   "The %d-th output of Output(%s) must be LoDTensor.", j, out);
+    auto in_tensor = in_var->Get<LoDTensor>();
+    auto* out_tensor = out_var->GetMutable<LoDTensor>();
+    out_tensor->set_layout(in_tensor.layout());
   }
 
   bool IsRuntime() const override { return true; }
-- 
GitLab