Merge pull request PaddlePaddle#40 from Superjomn/fix/lower

make simple lower test case works

cinn/ir/ir.cc

@@ -294,6 +294,11 @@ std::vector<const Expr *> PolyFor::expr_fields() const { return {&init, &conditi
 bool Var::operator==(const Var &o) const { return o->name == operator->()->name; }
 bool Var::operator!=(const Var &o) const { return !(*this == o); }
 
+Var &Var::operator=(_Var_ *x) {
+  *this = Var(x);
+  return *this;
+}
+
 }  // namespace ir
 
 namespace common {

cinn/ir/ir.h

@@ -278,6 +278,8 @@ struct Var : public IrNodeRef {
   bool operator==(const Var& o) const;
   bool operator!=(const Var& o) const;
 
+  Var& operator=(_Var_* x);
+
   const _Var_* operator->() const { return get(); }
   _Var_* operator->() { return get(); }
   const _Var_* get() const { return static_cast<const _Var_*>(ptr()); }

cinn/ir/ir_operators.h

@@ -1,3 +1,4 @@
+#pragma once
 #include "cinn/ir/ir.h"
 
 namespace cinn {
@@ -80,12 +81,12 @@ Expr operator>=(POD a, Expr b) {
 }
 
 //--
-Expr operator+(Expr a, Expr b) { return Add::Make(a, b); }
-Expr operator-(Expr a, Expr b) { return Sub::Make(a, b); }
-Expr operator*(Expr a, Expr b) { return Mul::Make(a, b); }
-Expr operator/(Expr a, Expr b) { return Div::Make(a, b); }
-Expr operator&&(Expr a, Expr b) { return And::Make(Expr(a), Expr(b)); }
-Expr operator||(Expr a, Expr b) { return Or::Make(Expr(a), Expr(b)); }
+inline Expr operator+(Expr a, Expr b) { return Add::Make(a, b); }
+inline Expr operator-(Expr a, Expr b) { return Sub::Make(a, b); }
+inline Expr operator*(Expr a, Expr b) { return Mul::Make(a, b); }
+inline Expr operator/(Expr a, Expr b) { return Div::Make(a, b); }
+inline Expr operator&&(Expr a, Expr b) { return And::Make(Expr(a), Expr(b)); }
+inline Expr operator||(Expr a, Expr b) { return Or::Make(Expr(a), Expr(b)); }
 
 }  // namespace ir
 }  // namespace cinn

cinn/lang/CMakeLists.txt

@@ -9,3 +9,4 @@ endforeach()
 cc_test(test_compute SRCS compute_test.cc DEPS core)
 cc_test(test_placeholder SRCS placeholder_test.cc DEPS core)
 cc_test(test_tensor SRCS tensor_test.cc DEPS core)
+cc_test(test_lower SRCS lower_test.cc DEPS core)

cinn/lang/compute.h

@@ -5,6 +5,7 @@
 #include <vector>
 
 #include "cinn/ir/ir.h"
+#include "cinn/ir/ir_operators.h"
 #include "cinn/lang/placeholder.h"
 #include "cinn/poly/schedule.h"
 

cinn/lang/lower.cc

@@ -1,6 +1,10 @@
 #include "cinn/lang/lower.h"
 
+#include <map>
+#include <set>
+
 #include "cinn/ir/ir_printer.h"
+#include "cinn/optim/remove_nested_block.h"
 #include "cinn/optim/replace_call_with_expr.h"
 #include "cinn/poly/ast_gen.h"
 
@@ -78,19 +82,28 @@ std::vector<LoweredFunc> Lower(const std::string& name, const std::vector<Tensor
   auto schedule = poly::CreateSchedule(stages);
 
   // generate the expressions for each group.
-  std::vector<Expr> block;
+  std::vector<Expr> exprs;
+  CHECK_GT(schedule->gened_groups().size(), 0) << "no group is generated";
   for (auto& group : schedule->gened_groups()) {
+    CHECK_GT(group.nodes.size(), 0) << "group is empty";
     std::map<std::string, Expr> tuple_to_expr;
     for (auto& node : group.nodes) {
-      auto& tensor                = tensor_dic.at(node->id());
-      tuple_to_expr[tensor->name] = tensor->body();
+      auto& tensor = tensor_dic.at(node->id());
+      // NOTE here just schedule the compute node.
+      if (!tensor->is_compute_node()) continue;
+
+      tuple_to_expr[tensor->name] = tensor->tensor_store_expanded_body();
     }
 
     Expr group_expr = LowerGroup(group, tuple_to_expr);
     VLOG(3) << "group expr: " << group_expr;
-    block.push_back(group_expr);
+    exprs.push_back(group_expr);
   }
 
+  Expr block = ir::Block::Make(exprs);
+  // call passes
+  optim::RemoveNestedBlock(&block);
+
   // prepare arguments
   std::vector<Argument> arguments;
   for (auto& arg : args) {

cinn/lang/lower.h

@@ -3,6 +3,9 @@
  */
 
 #pragma once
+#include <string>
+#include <vector>
+
 #include "cinn/ir/function.h"
 #include "cinn/ir/ir.h"
 #include "cinn/lang/module.h"

cinn/lang/lower_test.cc

@@ -0,0 +1,44 @@
+#include "cinn/lang/lower.h"
+
+#include <gtest/gtest.h>
+
+#include "cinn/lang/compute.h"
+#include "cinn/lang/placeholder.h"
+#include "cinn/utils/string.h"
+
+namespace cinn {
+namespace lang {
+
+TEST(lower, basic) {
+  const int M = 100;
+  const int N = 15;
+
+  Placeholder<float> A("A", {Expr(M), Expr(N)});
+
+  auto B = Compute(
+      {M, N}, [=](Var i, Var j) -> Expr { return A(i, j) + 1.f; }, "B");
+
+  auto lower_funcs = Lower("cal_B", {A, B});
+
+  LOG(INFO) << "lower_size " << lower_funcs.size();
+
+#define TEST_SOUTPUT(x, out)  LOG(INFO) << "\n" << x; \
+  EXPECT_EQ(utils::GetStreamCnt(x), utils::Trim(out));
+
+  auto out = R"ROC(
+{
+  poly_for (0, (c1 <= 99), 1)
+  {
+    poly_for (0, (c3 <= 14), 1)
+    {
+      A(c1, c3)
+      B[((c1 * 15) + c3)] = (A(c1, c3) + 1)
+    }
+  }
+}
+)ROC";
+  TEST_SOUTPUT(lower_funcs.front().body, out);
+}
+
+}  // namespace lang
+}  // namespace cinn

cinn/lang/placeholder.h

@@ -33,6 +33,8 @@ class Placeholder {
   Expr operator()(const std::vector<Expr> &indices) const;
   // @}
 
+  operator ir::Tensor() { return tensor_; }
+
  private:
   ir::Tensor tensor_;
 };

cinn/lang/tensor.cc

@@ -3,19 +3,44 @@
 #include <cstring>
 
 #include "cinn/common/common.h"
+#include "cinn/ir/ir_operators.h"
 #include "cinn/ir/ir_visitor.h"
 #include "cinn/ir/operation.h"
 #include "cinn/poly/stage.h"
 
 namespace cinn {
 namespace ir {
 
+namespace detail {
+
+Expr ExpandTo1DIndice(const std::vector<Expr> &shape, const std::vector<Expr> &indices) {
+  CHECK_EQ(shape.size(), indices.size());
+  Expr res = indices.front() * shape[1];
+  for (int i = 1; i < shape.size() - 1; i++) {
+    res = res + indices[i] * shape[i + 1];
+  }
+  if (shape.size() > 1) res = res + indices.back();
+  return res;
+}
+
+Expr ExpandTo1DIndice(const std::vector<int> &shape, const std::vector<Expr> &indices) {
+  std::vector<Expr> shape_;
+  for (int v : shape) shape_.push_back(Expr(v));
+  return ExpandTo1DIndice(shape, indices);
+}
+
+}  // namespace detail
+
 Tensor _Tensor_::Make(const std::string &name, const std::vector<Expr> &shape, FunctionRef fn) {
+  CHECK(!shape.empty()) << "Tensor shape is set empty";
+  CHECK(!name.empty()) << "Tensor name is set empty";
   auto n      = make_shared<_Tensor_>();
   n->name     = name;
   n->shape    = shape;
   n->operaion = fn;
   n->InitStage();
+  n->InitAxis();
+  n->SetDefaultBindedBuffer();
   return Tensor(n);
 }
 
@@ -26,8 +51,13 @@ Tensor _Tensor_::Make(const std::string &name,
                       Type dtype,
                       const std::map<std::string, IrNodeRef> &attrs,
                       const std::vector<Expr> &body) {
+  CHECK(!shape.empty()) << "Tensor shape is set empty";
+  CHECK(!name.empty()) << "Tensor name is set empty";
+
   auto op          = ComputeOp::Make(name, tag, attrs, axis, body, shape);
   auto *compute_op = const_cast<ComputeOp *>(op->As<ComputeOp>());
+
+  CHECK_EQ(axis.size(), shape.size()) << "axis not match the dimension in shape";
   compute_op->axis = axis;
 
   auto n      = make_shared<_Tensor_>();
@@ -36,6 +66,7 @@ Tensor _Tensor_::Make(const std::string &name,
   n->shape    = shape;
   n->set_type(dtype);
   n->InitStage();
+  n->SetDefaultBindedBuffer();
   return Tensor(n);
 }
 
@@ -76,6 +107,12 @@ void _Tensor_::InitStage() {
   }
 }
 
+void _Tensor_::InitAxis() {
+  CHECK(!shape.empty());
+  CHECK(axis.empty()) << "duplicate init axis";
+  axis = common::GenDefaultAxis(shape.size());
+}
+
 isl::set _Tensor_::GenerateIslDomain() {
   CHECK(!shape.empty()) << "shape should be set";
   std::vector<poly::Dim> dims;
@@ -138,5 +175,12 @@ Expr _Tensor_::body() const {
   NOT_IMPLEMENTED;
 }
 
+Expr _Tensor_::tensor_store_expanded_body() const {
+  CHECK(!is_placeholder_node()) << "placeholder should not expand store";
+  std::vector<Expr> axis_;
+  for (auto &a : axis) axis_.push_back(Expr(a));
+  return ir::Store::Make(buffer_var, body(), detail::ExpandTo1DIndice(shape, axis_));
+}
+
 }  // namespace ir
 }  // namespace cinn

cinn/lang/tensor.h

@@ -25,6 +25,11 @@ namespace ir {
 namespace detail {
 constexpr bool LE(int a, int b) { return a <= b; }
 constexpr bool GE(int a, int b) { return a >= b; }
+
+//! Expand milti-dim indices to 1-dim index.
+Expr ExpandTo1DIndice(const std::vector<int>& shape, const std::vector<Expr>& indices);
+Expr ExpandTo1DIndice(const std::vector<Expr>& shape, const std::vector<Expr>& indices);
+
 }  // namespace detail
 
 class _Tensor_;
@@ -54,11 +59,6 @@ class Tensor : public ir::IrNodeRef {
   //! Get number of dimensions.
   inline size_t ndims() const;
 
-  inline const _Tensor_* operator->() const { return As<_Tensor_>(); }
-  inline _Tensor_* operator->() { return As<_Tensor_>(); }
-
-  inline operator Expr() const { return Expr(get()); }
-
   /**
    * Take elements from the tensor.
    * This take one or multiple expressions as indices.
@@ -82,6 +82,11 @@ class Tensor : public ir::IrNodeRef {
    * @return The result expression representing a tensor read.
    */
   Expr operator()(const std::vector<Expr>& indices) const;
+
+  inline const _Tensor_* operator->() const { return As<_Tensor_>(); }
+  inline _Tensor_* operator->() { return As<_Tensor_>(); }
+
+  inline operator Expr() const { return Expr(get()); }
 };
 
 /**
@@ -99,6 +104,8 @@ class _Tensor_ : public ExprNode<_Tensor_> {
   std::string name;
   //! Polyhedral element for analysis and schedule.
   poly::Stage* stage{};
+  //! The binded buffer, for each tensor if it is not inline.
+  Var buffer_var;
 
   //! Generate a tensor from a computation.
   static Tensor Make(const std::string& name,
@@ -112,12 +119,19 @@ class _Tensor_ : public ExprNode<_Tensor_> {
   //! Generate a tensor from a function.
   static Tensor Make(const std::string& name, const std::vector<Expr>& shape, FunctionRef fn);
 
+  //! Tell the operation type.
+  // @{
   bool is_compute_node() const;
   bool is_placeholder_node() const;
   const char* operation_type() const;
+  // @}
 
   //! The expression generate this tensor, will be empty if it is a PlaceHolder.
   Expr body() const;
+  //! Get the expression with `store(tensor)` inserted into the body.
+  Expr tensor_store_expanded_body() const;
+
+  Expr inline_expanded(const std::vector<Expr>& indices);
 
   std::vector<Expr*> expr_fields() override;
   std::vector<const Expr*> expr_fields() const override;
@@ -133,6 +147,13 @@ class _Tensor_ : public ExprNode<_Tensor_> {
   //! It is based on the shape.
   void InitStage();
 
+  //! Initialize the axis field after the shape field is assigned.
+  void InitAxis();
+
+  //! Bind the tensor to a buffer by default.
+  //! NOTE it should called by all the Make.
+  void SetDefaultBindedBuffer() { buffer_var = ir::_Var_::Make(name, type()).As<_Var_>(); }
+
   isl::set GenerateIslDomain();
 };
 

cinn/poly/graph.cc

@@ -105,11 +105,11 @@ std::vector<Group> PartitionGraphByIterationDomain(common::Graph* graph) {
   std::set<DataFlowGraphNode*> groups_gathered;
   std::vector<DataFlowGraphNode*> groups_in_topo_order;
 
-  std::vector<common::GraphNode*> nodes;
-  std::vector<common::GraphEdge*> edges;
+  std::vector<common::GraphNode*> nodes_in_order;
+  std::vector<common::GraphEdge*> edges_in_order;
   std::map<DataFlowGraphNode*, std::vector<DataFlowGraphNode*>> node_groups;
-  std::tie(nodes, edges) = graph->topological_order();
-  for (auto* n : nodes) {
+  std::tie(nodes_in_order, edges_in_order) = graph->topological_order();
+  for (auto* n : nodes_in_order) {
     auto* node     = n->As<DataFlowGraphNode>();
     auto* ancestor = node->group_ancestor();
     if (!groups_gathered.count(ancestor)) {
@@ -122,14 +122,22 @@ std::vector<Group> PartitionGraphByIterationDomain(common::Graph* graph) {
 
   std::vector<Group> groups;
   // preparing result
-  for (auto* n : groups_in_topo_order) {
+  for (auto* ancestor : groups_in_topo_order) {
     Group group;
-    for (auto* c : node_groups[n]) {
+    for (auto* c : node_groups[ancestor]) {
       group.nodes.push_back(c);
     }
     groups.emplace_back(group);
   }
 
+  // NOTE DEBUG
+  // check there are same count of nodes both in the orginal graph and the groups.
+  // @{
+  int num_node_in_groups = 0;
+  for (auto& group : groups) num_node_in_groups += group.nodes.size();
+  CHECK_EQ(num_node_in_groups, graph->num_nodes());
+  // @}
+
   return groups;
 }
 
@@ -148,9 +156,10 @@ std::unique_ptr<common::Graph> CreateGraph(const std::vector<Stage*>& stages) {
     }
   }
 
-  auto* graph = new common::Graph;
+  std::unique_ptr<common::Graph> graph(new common::Graph);
   for (auto& item : id2stage) graph->RegisterNode(item.first, item.second.get());
-  return std::unique_ptr<common::Graph>(graph);
+  VLOG(3) << "created graph:\n" << graph->Visualize();
+  return graph;
 }
 
 }  // namespace poly