[BYOC] Refine AnnotateTarget and MergeCompilerRegion Passes (#5277)

* add target to region

* refactor annotate_target

* Make all unit test working

* quick fix

* enable BN, unit test failed

* Fix vm test, unit test. Refactor annotate_target a bit.

* quick fix fusion

* revert fusion change

* style fix

* Refactor merge region pass

* format

* minor fix

* Skip e2e test

* lint

* support AnnotateTarget multiple runs

* Add HasAttr and revert DNNL codegen

* address comment

Co-authored-by: Zhi Chen <chzhi@amazon.com>

python/tvm/relay/op/contrib/dnnl.py

@@ -56,10 +56,17 @@ def _register_external_op_helper(op_name, supported=True):
     return _func_wrapper
 
 
-_register_external_op_helper("nn.batch_norm")
 _register_external_op_helper("nn.conv2d")
 _register_external_op_helper("nn.dense")
 _register_external_op_helper("nn.relu")
 _register_external_op_helper("add")
 _register_external_op_helper("subtract")
 _register_external_op_helper("multiply")
+
+
+@reg.register("nn.batch_norm", "target.dnnl")
+def batch_norm(attrs, args):
+    """Check if the external DNNL codegen should be used.
+    FIXME(@zhiics, @comaniac): Turn off due to not support of multiple outputs.
+    """
+    return False

python/tvm/relay/transform/transform.py

@@ -587,14 +587,14 @@ def PartitionGraph():
 
 
 
-def AnnotateTarget(target):
+def AnnotateTarget(targets):
     """Annotate ops in an experession with a provied compiler/target and then
     use it for codegen.
 
     Parameters
     ----------
-    target : String
-        The target compiler used for codegen.
+    targets : str or List[str]
+        The list of target compilers used for codegen.
 
     Returns
     -------
@@ -602,7 +602,9 @@ def AnnotateTarget(target):
         The annotated pass that wrapps ops with subgraph_start and
         subgraph_end.
     """
-    return _ffi_api.AnnotateTarget(target)
+    if isinstance(targets, str):
+        targets = [targets]
+    return _ffi_api.AnnotateTarget([tvm.runtime.container.String(t) for t in targets])
 
 
 def Inline():

src/relay/analysis/annotated_region_set.cc

@@ -21,6 +21,7 @@
 
 #include <tvm/relay/expr.h>
 #include <tvm/ir/error.h>
+#include <tvm/runtime/container.h>
 
 #include <unordered_map>
 #include <vector>
@@ -31,7 +32,7 @@ namespace relay {
 
 AnnotatedRegion AnnotatedRegionSetNode::GetRegion(const Expr& expr) const {
   for (auto candidate : regions_) {
-    if (candidate->nodes.find(expr) != candidate->nodes.end()) {
+    if (candidate->nodes_.find(expr) != candidate->nodes_.end()) {
       return candidate;
     }
   }
@@ -45,26 +46,26 @@ void AnnotatedRegionSetNode::MergeRegions(AnnotatedRegion src,
   }
 
   // Merge src to dest and erase src.
-  dest->nodes.insert(src->nodes.begin(), src->nodes.end());
-  for (const auto& input : src->ins) {
-    dest->ins.push_back(input);
+  dest->nodes_.insert(src->nodes_.begin(), src->nodes_.end());
+  for (const auto& input : src->ins_) {
+    dest->ins_.push_back(input);
   }
-  for (const auto& output : src->outs) {
-    dest->outs.push_back(output);
+  for (const auto& output : src->outs_) {
+    dest->outs_.push_back(output);
   }
   // if any of the outputs of src are inputs of dest, they become internal nodes
   // so remove them from outs
   std::vector<Expr> ins_to_remove;
-  for (const auto& input : dest->ins) {
+  for (const auto& input : dest->ins_) {
     auto call = Downcast<Call>(input);
-    auto it = src->nodes.find(call->args[0]);
-    if (it != src->nodes.end()) {
-      dest->outs.remove(*it);
+    auto it = src->nodes_.find(call->args[0]);
+    if (it != src->nodes_.end()) {
+      dest->outs_.remove(*it);
       ins_to_remove.push_back(input);
     }
   }
   for (const auto& input : ins_to_remove) {
-    dest->ins.remove(input);
+    dest->ins_.remove(input);
   }
   regions_.erase(src);
 }
@@ -74,25 +75,21 @@ void AnnotatedRegionSetNode::AddToRegion(AnnotatedRegion dest, const Expr& expr)
   if (src.defined()) {
     MergeRegions(src, dest);
   } else {
-    dest->nodes.insert(expr);
+    dest->nodes_.insert(expr);
   }
 }
 
-AnnotatedRegion AnnotatedRegionSetNode::MakeRegion() {
+AnnotatedRegion AnnotatedRegionSetNode::MakeRegion(const std::string& target) {
   auto ret = regions_.emplace(AnnotatedRegion());
-  (*ret.first)->id = region_id_++;
+  (*ret.first)->id_ = region_id_++;
+  (*ret.first)->target_ = target;
   return *ret.first;
 }
 
 class AnnotatedRegionSet::Creator : public ExprVisitor {
  public:
-  Creator(const Op& region_begin_op, const Op& region_end_op) :
-    begin_op_(region_begin_op), end_op_(region_end_op) {}
-
-  AnnotatedRegionSet Create(const Expr& expr) {
-    VisitExpr(expr);
-    return std::move(region_set_);
-  }
+  Creator(const Op& region_begin_op, const Op& region_end_op)
+      : begin_op_(region_begin_op), end_op_(region_end_op) {}
 
   void VisitExpr_(const CallNode* call) {
     auto op_node = call->op.as<OpNode>();
@@ -115,24 +112,35 @@ class AnnotatedRegionSet::Creator : public ExprVisitor {
                       << "Cannot find the corresponding region for start annotation:\n"
                       << AsText(GetRef<Call>(call), false));
       }
-      region->ins.push_back(GetRef<Call>(call));
+      region->ins_.push_back(GetRef<Call>(call));
     } else {
       CHECK_EQ(call->op, end_op_);
       // The annotation node is inserted on edge so it must have only one argument.
       CHECK_EQ(call->args.size(), 1U);
+      std::string target = call->attrs.as<CompilerAttrs>()->compiler;
 
       // Check if the argument already belongs to a region
       auto region = region_set_->GetRegion(call->args[0]);
       if (!region.defined()) {
-        region = region_set_->MakeRegion();
-        region->nodes.insert(call->args[0]);
+        // Create a new region if the argument is not belonged to any regions yet.
+        region = region_set_->MakeRegion(target);
+        region->nodes_.insert(call->args[0]);
+      } else {
+        // If the argument is belonged to a region, it must have the same target.
+        // Otherwise we should see a region_begin op.
+        CHECK_EQ(region->GetTarget(), target);
       }
-      region->nodes.insert(GetRef<Call>(call));
-      region->outs.push_back(GetRef<Call>(call));
+      region->nodes_.insert(GetRef<Call>(call));
+      region->outs_.push_back(GetRef<Call>(call));
     }
     ExprVisitor::VisitExpr_(call);
   }
 
+  AnnotatedRegionSet Create(const Expr& expr) {
+    VisitExpr(expr);
+    return std::move(region_set_);
+  }
+
   void VisitExpr_(const TupleNode* op) {
     auto region = region_set_->GetRegion(GetRef<Tuple>(op));
     if (region.defined()) {

src/relay/analysis/annotated_region_set.h

@@ -32,6 +32,7 @@
 #include <tvm/relay/expr.h>
 #include <tvm/ir/error.h>
 #include <tvm/relay/expr_functor.h>
+#include <tvm/runtime/container.h>
 #include <tvm/relay/transform.h>
 
 #include <string>
@@ -49,33 +50,39 @@ class AnnotatedRegionSet;
 class AnnotatedRegionNode : public Object {
  public:
   void VisitAttrs(AttrVisitor* v) {
-    v->Visit("id", &id);
-    Array<Expr> nodes_array(nodes.begin(), nodes.end());
+    v->Visit("id", &id_);
+    v->Visit("target", &target_);
+    Array<Expr> nodes_array(nodes_.begin(), nodes_.end());
     v->Visit("nodes", &nodes_array);
-    Array<Expr> args_array(ins.begin(), ins.end());
+    Array<Expr> args_array(ins_.begin(), ins_.end());
     v->Visit("args", &args_array);
-    Array<Expr> rets_array(outs.begin(), outs.end());
+    Array<Expr> rets_array(outs_.begin(), outs_.end());
     v->Visit("rets", &rets_array);
   }
 
   /*! \brief Get the region ID. */
   int GetID() const {
-    return id;
+    return id_;
+  }
+
+  /*! \brief Get the region target. */
+  std::string GetTarget() const {
+    return target_;
   }
 
   /*! \brief Get the region's inputs. */
   std::list<Expr> GetInputs() const {
-    return ins;
+    return ins_;
   }
 
   /*! \brief Get the region's outputs. */
   std::list<Expr> GetOutputs() const {
-    return outs;
+    return outs_;
   }
 
   /*! \brief Get the region's nodes. */
   std::unordered_set<Expr, ObjectHash, ObjectEqual> GetNodes() const {
-    return nodes;
+    return nodes_;
   }
 
   static constexpr const char* _type_key = "relay.AnnotatedRegion";
@@ -83,13 +90,15 @@ class AnnotatedRegionNode : public Object {
 
  protected:
   /*! \brief The region ID. */
-  int id{-1};
+  int id_{-1};
+  /*! \brief The target for this region. */
+  std::string target_ = "default";
   /*! \brief The inputs to this region. */
-  std::list<Expr> ins;
+  std::list<Expr> ins_;
   /*! \brief The outputs of this region */
-  std::list<Expr> outs;
+  std::list<Expr> outs_;
   /*! \brief Nodes in this region. */
-  std::unordered_set<Expr, ObjectHash, ObjectEqual> nodes;
+  std::unordered_set<Expr, ObjectHash, ObjectEqual> nodes_;
 
   friend class AnnotatedRegionSet;
   friend class AnnotatedRegionSetNode;
@@ -184,11 +193,11 @@ class AnnotatedRegionSetNode : public Object {
   void AddToRegion(AnnotatedRegion dest, const Expr& expr);
 
   /*!
-   * \brief Make a new region.
+   * \brief Make a new region for a target.
    *
    * \return The new region.
    */
-  AnnotatedRegion MakeRegion();
+  AnnotatedRegion MakeRegion(const std::string& target);
 
   std::unordered_set<AnnotatedRegion, ObjectHash, ObjectEqual> regions_;
   /*! \brief The next region ID to assign. */

src/relay/backend/contrib/dnnl/codegen.cc

@@ -53,19 +53,12 @@ class CodegenDNNL : public ExprVisitor, public CodegenCBase {
   }
 
   void VisitExpr_(const TupleGetItemNode* op) final {
-    VisitExpr(op->tuple);
-    CHECK(out_.size() > static_cast<size_t>(op->index));
-
-    // Only keep the item we want for the child node.
-    // FIXME(@comaniac): The other items should still be requried for the primary outputs.
-    auto item = out_[op->index];
-    out_.clear();
-    out_.push_back(item);
+    // Do nothing
   }
 
   void VisitExpr_(const CallNode* call) final {
     std::ostringstream decl_stream;
-
+    std::ostringstream buf_stream;
     // Args: ID
     std::vector<std::string> args;
 
@@ -103,45 +96,20 @@ class CodegenDNNL : public ExprVisitor, public CodegenCBase {
       }
     }
 
-    // Analyze the output buffers
-    std::vector<Type> out_types;
-    if (call->checked_type()->IsInstance<TupleTypeNode>()) {
-      auto type_node = call->checked_type().as<TupleTypeNode>();
-      for (auto field : type_node->fields) {
-        CHECK(field->IsInstance<TensorTypeNode>());
-        out_types.push_back(field);
-      }
-    } else if (call->checked_type()->IsInstance<TensorTypeNode>()) {
-      CHECK(call->checked_type()->IsInstance<TensorTypeNode>());
-      out_types.push_back(call->checked_type());
-    } else {
-      LOG(FATAL) << "Unrecognized type node: " << AsText(call->checked_type(), false);
-    }
-
-    out_.clear();
-    for (auto out_type : out_types) {
-      const auto& dtype = GetDtypeString(out_type.as<TensorTypeNode>());
-
-      std::string out = "buf_" + std::to_string(buf_idx_++);
-      auto out_shape = GetShape(out_type);
-      int out_size = 1;
-      for (size_t i = 0; i < out_shape.size(); ++i) {
-        out_size *= out_shape[i];
-      }
-      this->PrintIndents();
-      std::ostringstream buf_stream;
-      buf_stream << "float* " << out << " = (float*)std::malloc(4 * " << out_size << ");";
-      buf_decl_.push_back(buf_stream.str());
-      decl_stream << ", " << out;
-
-      // Update output buffer
-      Output output;
-      output.name = out;
-      output.dtype = dtype;
-      output.need_copy = true;
-      output.size = out_size;
-      out_.push_back(output);
+    // Analyze the output buffer
+    auto type_node = call->checked_type().as<TensorTypeNode>();
+    CHECK(type_node);
+    const auto& dtype = GetDtypeString(type_node);
+    std::string out = "buf_" + std::to_string(buf_idx_++);
+    auto out_shape = GetShape(call->checked_type());
+    int out_size = 1;
+    for (size_t i = 0; i < out_shape.size(); ++i) {
+      out_size *= out_shape[i];
     }
+    this->PrintIndents();
+    buf_stream << "float* " << out << " = (float*)std::malloc(4 * " << out_size << ");";
+    buf_decl_.push_back(buf_stream.str());
+    decl_stream << ", " << out;
 
     // Attach attribute arguments
     for (size_t i = 0; i < args.size(); ++i) {
@@ -149,6 +117,15 @@ class CodegenDNNL : public ExprVisitor, public CodegenCBase {
     }
     decl_stream << ");";
     ext_func_body.push_back(decl_stream.str());
+
+    // Update output buffer
+    out_.clear();
+    Output output;
+    output.name = out;
+    output.dtype = dtype;
+    output.need_copy = true;
+    output.size = out_size;
+    out_.push_back(output);
   }
 
   std::string JIT(void) {

src/relay/backend/vm/compiler.cc

@@ -924,13 +924,6 @@ IRModule VMCompiler::OptimizeModule(const IRModule& mod, const TargetsMap& targe
   pass_seqs.push_back(transform::LambdaLift());
   pass_seqs.push_back(transform::InlinePrimitives());
 
-  // Manifest the allocations.
-  pass_seqs.push_back(transform::ManifestAlloc(this->target_host_));
-  // Compute away possibly introduced constant computation.
-  pass_seqs.push_back(transform::FoldConstant());
-  // Fuse the shape functions.
-  pass_seqs.push_back(transform::FuseOps());
-
   // Inline the functions that are lifted to the module scope. We perform this
   // pass after all other optimization passes but before the memory allocation
   // pass. This is because memory allocation pass will insert `invoke_tvm_op`
@@ -938,6 +931,12 @@ IRModule VMCompiler::OptimizeModule(const IRModule& mod, const TargetsMap& targe
   // external codegen.
   pass_seqs.push_back(transform::Inline());
 
+  // Manifest the allocations.
+  pass_seqs.push_back(transform::ManifestAlloc(this->target_host_));
+  // Compute away possibly introduced constant computation.
+  pass_seqs.push_back(transform::FoldConstant());
+  // Fuse the shape functions.
+  pass_seqs.push_back(transform::FuseOps());
   // Manifest the allocations needed for the shape functions.
   pass_seqs.push_back(transform::ManifestAlloc(this->target_host_));
 

src/relay/transforms/partition_graph.cc

@@ -477,13 +477,48 @@ class Partitioner : public ExprMutator {
   IRModule module_;
 };
 
+class DefaultRemover : public ExprMutator {
+ public:
+  explicit DefaultRemover(const IRModule& module) : module_(module) {}
+
+  IRModule Remove() {
+    auto glob_funcs = module_->functions;
+    for (const auto& pair : glob_funcs) {
+      if (auto* fn = pair.second.as<FunctionNode>()) {
+        auto func = GetRef<Function>(fn);
+        func = Function(func->params, VisitExpr(func->body), func->ret_type, func->type_params,
+                        func->attrs);
+        module_->Update(pair.first, func);
+      }
+    }
+    return module_;
+  }
+
+  Expr VisitExpr_(const CallNode* call) final {
+    auto attrs = call->attrs.as<CompilerAttrs>();
+    if (attrs != nullptr && attrs->compiler == "default") {
+      return VisitExpr(call->args[0]);
+    }
+    return ExprMutator::VisitExpr_(call);
+  }
+
+ private:
+  IRModule module_;
+};
+
 }  // namespace partitioning
 
 namespace transform {
 
 Pass PartitionGraph() {
   runtime::TypedPackedFunc<IRModule(IRModule, PassContext)> part_func =
-      [=](IRModule m, PassContext pc) { return partitioning::Partitioner(m).Partition(); };
+      [=](IRModule m, PassContext pc) {
+        // TODO(@comaniac, @zhiics): We should also handle the annotation with "default" attribute
+        // by treating them as un-annotated, but we don't have it yet. This workaround pass removes
+        // all "default" annotations and should be deleted in the future.
+        auto new_m = partitioning::DefaultRemover(m).Remove();
+        return partitioning::Partitioner(new_m).Partition();
+  };
   auto partitioned = CreateModulePass(part_func, 0, "PartitionGraph", {});
   return Sequential({partitioned, InferType()});
 }

src/runtime/contrib/dnnl/dnnl.cc

@@ -169,11 +169,9 @@ extern "C" void dnnl_relu(float* data, float* out, int p_N_, int p_C_, int p_H_,
   read_from_dnnl_memory(out, dst_memory);
 }
 
-extern "C" void dnnl_bn(float* data, float* gamma, float* beta, float* mean, float* variance,
-                        float* out, float* new_mean, float* new_variance, int p_N_, int p_C_,
+extern "C" void dnnl_bn(float* data, float* gamma, float* beta, float* mean,
+                        float* variance, float* out, int p_N_, int p_C_,
                         int p_H_, int p_W_, int p_E_) {
-  // FIXME(@comaniac): BN has 3 outputs: out, new_mean and new_variance, but we do not update
-  // the rest two because no one cares about them for now. Should update it in the future.
   using tag = memory::format_tag;
   using dt = memory::data_type;
 

src/runtime/contrib/dnnl/dnnl_kernel.h

@@ -44,8 +44,8 @@ extern "C" TVM_DLL void dnnl_dense(float* data, float* weight, float* out, int p
 extern "C" TVM_DLL void dnnl_relu(float* data, float* out, int p_N_, int p_C_, int p_H_, int p_W_);
 
 extern "C" TVM_DLL void dnnl_bn(float* data, float* gamma, float* beta, float* mean,
-                                float* variance, float* out, float* new_mean, float* new_variance,
-                                int p_n_, int p_c_, int p_h_, int p_w_, int p_e_);
+                                float* variance, float* out, int p_n_, int p_c_, int p_h_, int p_w_,
+                                int p_e_);
 
 extern "C" TVM_DLL void dnnl_add(float* data, float* weight, float* out, int p_n_, int p_c_,
                                  int p_h_, int p_w_);

tests/python/relay/test_annotated_regions.py

@@ -15,13 +15,15 @@
 # specific language governing permissions and limitations
 # under the License.
 # pylint: disable=no-else-return, unidiomatic-typecheck, invalid-name
+import tvm
 from tvm import relay
 from tvm.relay.op.annotation import compiler_begin, compiler_end
 
 
-def check_region(region_set, args, nodes, rets):
+def check_region(region_set, target, args, nodes, rets):
     region = region_set.get_region(args[0])
     assert region
+    assert target == region.target
     assert set(args) == set(region.args)
     assert set(nodes) == set(region.nodes)
     assert set(rets) == set(region.rets)
@@ -51,24 +53,28 @@ def test_region_set_creator_diamond():
     assert len(region_set) == 4
     check_region(
         region_set,
+        'test_target',
         [cb_1],
         [cb_1, O_1, ce_1, ce_2],
         [ce_1, ce_2],
     )
     check_region(
         region_set,
+        'test_target',
         [cb_2],
         [cb_2, O_2, ce_3],
         [ce_3],
     )
     check_region(
         region_set,
+        'default',
         [cb_d],
         [cb_d, X, ce_d],
         [ce_d],
     )
     check_region(
         region_set,
+        'test_target',
         [cb_3, cb_4],
         [cb_3, cb_4, O_3, ce_4],
         [ce_4],
@@ -88,7 +94,9 @@ def test_region_set_creator_merged():
     cb_3 = compiler_begin(ce_3, 'test_target')
     cb_4 = compiler_begin(ce_d, 'test_target')
     O_3 = relay.add(cb_3, cb_4)
-    ce_4 = compiler_end(O_3, 'test_target')
+    O_4 = relay.add(cb_3, cb_4)
+    O_5 = relay.Tuple([O_3, O_4])
+    ce_4 = compiler_end(O_5, 'test_target')
     merged = relay.Function([data], ce_4)
 
     region_set = relay.analysis.AnnotatedRegionSet(merged,
@@ -97,20 +105,23 @@ def test_region_set_creator_merged():
     assert len(region_set) == 3
     check_region(
         region_set,
+        'test_target',
         [cb_1],
         [cb_1, O_1, O_2, ce_2, ce_3],
         [ce_2, ce_3],
     )
     check_region(
         region_set,
+        'default',
         [cb_d],
         [cb_d, X, ce_d],
         [ce_d],
     )
     check_region(
         region_set,
+        'test_target',
         [cb_3, cb_4],
-        [cb_3, cb_4, O_3, ce_4],
+        [cb_3, cb_4, O_3, O_4, O_5, ce_4],
         [ce_4],
     )
 
@@ -118,4 +129,3 @@ def test_region_set_creator_merged():
 if __name__ == "__main__":
     test_region_set_creator_diamond()
     test_region_set_creator_merged()
-

tests/python/relay/test_annotate_target.py → tests/python/relay/test_pass_annotate_target.py

@@ -186,12 +186,11 @@ def test_extern_dnnl_mobilenet():
                  (1, 1000), ref_res.asnumpy(), tol=1e-5, params=params)
 
 
-@reg.register("nn.relu", "target.test")
-def relu(attrs, args):
-    return True
-
-
 def test_multiple_ends():
+    @reg.register("nn.relu", "target.test")
+    def relu(attrs, args):  # pylint: disable=unused-variable
+        return True
+
     def before():
         x = relay.var("x", shape=(10, 10))
         r = relay.nn.relu(x)
@@ -208,10 +207,17 @@ def test_multiple_ends():
         r = relay.nn.relu(cb_1)
         ce_1 = relay.annotation.compiler_end(r, "test")
         ce_2 = relay.annotation.compiler_end(r, "test")
-        a_1 = relay.abs(ce_1)
-        a_2 = relay.abs(ce_2)
-        out = relay.add(a_1, a_2)
-        f = relay.Function([x], out)
+        cb_2 = relay.annotation.compiler_begin(ce_1, "default")
+        cb_3 = relay.annotation.compiler_begin(ce_2, "default")
+        a_1 = relay.abs(cb_2)
+        a_2 = relay.abs(cb_3)
+        ce_3 = relay.annotation.compiler_end(a_1, "default")
+        ce_4 = relay.annotation.compiler_end(a_2, "default")
+        cb_4 = relay.annotation.compiler_begin(ce_3, "default")
+        cb_5 = relay.annotation.compiler_begin(ce_4, "default")
+        out = relay.add(cb_4, cb_5)
+        ce_6 = relay.annotation.compiler_end(out, "default")
+        f = relay.Function([x], ce_6)
         mod = tvm.IRModule.from_expr(f)
         return mod
 
@@ -220,6 +226,72 @@ def test_multiple_ends():
     assert tvm.ir.structural_equal(expected, result)
 
 
+def test_type_propagation():
+    target = "test_type_propagation"
+
+    @reg.register("nn.relu", "target." + target)
+    def relu(attrs, args): # pylint: disable=unused-variable
+        return args[0].checked_type.dtype == "float32"
+
+    def before():
+        x = relay.var("x", shape=(10, 10))
+        r = relay.nn.relu(x)
+        out = relay.nn.relu(r)
+        f = relay.Function([x], out)
+        mod = tvm.IRModule.from_expr(f)
+        return mod
+
+    # If the type isn't propogated, then the relu checker function will fail to get the dtype.
+    assert transform.AnnotateTarget(target)(before())
+
+
+def test_tuple():
+    target = "test_tuple"
+
+    @reg.register("nn.relu", "target." + target)
+    def relu(attrs, args): # pylint: disable=unused-variable
+        return True
+
+    @reg.register("concatenate", "target." + target)
+    def concatenate(attrs, args):  # pylint: disable=unused-variable
+        return True
+
+    """Test that TupleNode is included in annotation when surrounded by supported nodes."""
+    def before():
+        x = relay.var("x", shape=(10, 5))
+        y = relay.var("y", shape=(10, 5))
+        a_1 = relay.nn.relu(x)
+        a_2 = relay.nn.relu(y)
+        out = relay.concatenate((a_1, a_2), axis=1)
+        f = relay.Function([x, y], out)
+        mod = tvm.IRModule.from_expr(f)
+        return mod
+
+    def after():
+        x = relay.var("x", shape=(10, 5))
+        y = relay.var("y", shape=(10, 5))
+        cb_1 = relay.annotation.compiler_begin(x, target)
+        cb_2 = relay.annotation.compiler_begin(y, target)
+        a_1 = relay.nn.relu(cb_1)
+        a_2 = relay.nn.relu(cb_2)
+        ce_1 = relay.annotation.compiler_end(a_1, target)
+        ce_2 = relay.annotation.compiler_end(a_2, target)
+        cb_3 = relay.annotation.compiler_begin(ce_1, target)
+        cb_4 = relay.annotation.compiler_begin(ce_2, target)
+        tup = relay.Tuple([cb_3, cb_4])
+        ce_3 = relay.annotation.compiler_end(tup, target)
+        cb_3 = relay.annotation.compiler_begin(ce_3, target)
+        out = relay.op._make.concatenate(cb_3, 1)
+        ce_4 = relay.annotation.compiler_end(out, target)
+        f = relay.Function([x, y], ce_4)
+        mod = tvm.IRModule.from_expr(f)
+        return mod
+
+    result = transform.AnnotateTarget(target)(before())
+    expected = transform.InferType()(after())
+    assert tvm.ir.structural_equal(expected, result)
+
+
 def test_composite_function():
     def before():
         a = relay.var('a', shape=(10, 10))
@@ -265,8 +337,37 @@ def test_composite_function():
     assert tvm.ir.structural_equal(expected, result)
 
 
+def test_multiple_runs():
+    @reg.register("nn.relu", "target.A")
+    def relu(attrs, args):  # pylint: disable=unused-variable
+        return True
+
+    @reg.register("add", "target.B")
+    def add(attrs, args):  # pylint: disable=unused-variable
+        return True
+
+    def before():
+        x = relay.var("x", shape=(10, 5))
+        a_1 = relay.nn.relu(x)
+        a_2 = relay.abs(a_1)
+        a_3 = relay.nn.relu(a_1)
+        out = relay.add(a_2, a_3)
+
+        f = relay.Function([x], out)
+        mod = tvm.IRModule.from_expr(f)
+        return mod
+
+    mod = transform.AnnotateTarget("A")(before())
+    mod = transform.AnnotateTarget("B")(mod)
+    expected = transform.AnnotateTarget(["A", "B"])(before())
+    assert tvm.ir.structural_equal(expected, mod)
+
+
 if __name__ == "__main__":
-    test_multiple_ends()
     test_extern_dnnl()
-    #test_extern_dnnl_mobilenet()
     test_composite_function()
+    #test_extern_dnnl_mobilenet()
+    test_multiple_ends()
+    test_type_propagation()
+    test_tuple()
+    test_multiple_runs()

tests/python/relay/test_pass_merge_compiler_regions.py

@@ -30,9 +30,9 @@ def test_diamond_graph_fanouts():
     X = not supported by target
 
        O         O
-      / \       /               \
+      / \\      /               \\
      O   X --> O    +       +    X
-      \ /              \ /
+     \\ /             \\ /
        O                O
 
     Note that we can't just merge the three supported operators together,
@@ -45,17 +45,20 @@ def test_diamond_graph_fanouts():
         ce_1 = compiler_end(O_1, "test")
         ce_2 = compiler_end(O_1, "test")
         cb_2 = compiler_begin(ce_1, "test")
+        cb_3 = compiler_begin(ce_2, "default")
         O_2 = relay.nn.relu(cb_2)
         ce_3 = compiler_end(O_2, "test")
 
-        X = relay.tanh(ce_2)
 
-        cb_3 = compiler_begin(ce_3, "test")
-        cb_4 = compiler_begin(X, "test")
-        O_3 = relay.add(cb_3, cb_4)
-        ce_4 = compiler_end(O_3, "test")
+        X = relay.tanh(cb_3)
+        ce_4 = compiler_end(X, "default")
 
-        diamond = relay.Function([data], ce_4)
+        cb_4 = compiler_begin(ce_3, "test")
+        cb_5 = compiler_begin(ce_4, "test")
+        O_3 = relay.add(cb_4, cb_5)
+        ce_5 = compiler_end(O_3, "test")
+
+        diamond = relay.Function([data], ce_5)
         return diamond
 
     def expected():
@@ -66,14 +69,16 @@ def test_diamond_graph_fanouts():
         O_2 = relay.nn.relu(O_1)
         ce_3 = compiler_end(O_2, "test")
 
-        X = relay.tanh(ce_2)
+        cb_3 = compiler_begin(ce_2, "default")
+        X = relay.tanh(cb_3)
+        ce_4 = compiler_end(X, "default")
 
-        cb_3 = compiler_begin(ce_3, "test")
-        cb_4 = compiler_begin(X, "test")
-        O_3 = relay.add(cb_3, cb_4)
-        ce_4 = compiler_end(O_3, "test")
+        cb_4 = compiler_begin(ce_3, "test")
+        cb_5 = compiler_begin(ce_4, "test")
+        O_3 = relay.add(cb_4, cb_5)
+        ce_5 = compiler_end(O_3, "test")
 
-        func = relay.Function([data], ce_4)
+        func = relay.Function([data], ce_5)
         return func
 
     result = run_opt_pass(diamond_graph_fanouts(), relay.transform.MergeCompilerRegions())
@@ -85,7 +90,7 @@ def test_example_graph():
     """This tests the merging algorithm on the example used in the RFC.
 
     See the RFC here: https://discuss.tvm.ai/t/relay-improved-graph-partitioning-algorithm/5830
-    Blue nodes are adds, red nodes are subtracts.
+    Blue nodes are adds (target: test), red nodes are subtracts (target: default).
     """
     def annotated():
         in_1 = relay.var('in_1', shape=(10, 10), dtype='float32')
@@ -112,21 +117,30 @@ def test_example_graph():
         node2 = relay.add(begin4, begin5)
         end2 = compiler_end(node2, "test")
 
-        node3 = relay.subtract(in_5, in_6)
-        node4 = relay.subtract(in_7, node3)
+        dbegin0 = compiler_begin(in_5, "default")
+        dbegin1 = compiler_begin(in_6, "default")
+        node3 = relay.subtract(dbegin0, dbegin1)
+        dbegin2 = compiler_begin(in_7, "default")
+        dend1 = compiler_end(node3, "default")
+        dbegin3 = compiler_begin(dend1, "default")
+        node4 = relay.subtract(dbegin2, dbegin3)
+        dend2 = compiler_end(node4, "default")
 
         begin6 = compiler_begin(end2, "test")
-        begin7 = compiler_begin(node4, "test")
+        begin7 = compiler_begin(dend2, "test")
         node5 = relay.add(begin6, begin7)
         end3 = compiler_end(node5, "test")
         end4 = compiler_end(node5, "test")
-        node6 = relay.subtract(in_8, end3)
+        dbegin4 = compiler_begin(in_8, "default")
+        dbegin5 = compiler_begin(end3, "default")
+        node6 = relay.subtract(dbegin4, dbegin5)
         begin8 = compiler_begin(in_9, "test")
         begin9 = compiler_begin(end4, "test")
         node7 = relay.add(begin8, begin9)
         end5 = compiler_end(node7, "test")
 
-        begin10 = compiler_begin(node6, "test")
+        dend3 = compiler_end(node6, "default")
+        begin10 = compiler_begin(dend3, "test")
         begin11 = compiler_begin(end5, "test")
         node8 = relay.add(begin10, begin11)
         end6 = compiler_end(node8, "test")
@@ -159,20 +173,27 @@ def test_example_graph():
         node1 = relay.add(begin2, begin3)
         node2 = relay.add(node0, node1)
 
-        node3 = relay.subtract(in_5, in_6)
-        node4 = relay.subtract(in_7, node3)
+        dbegin0 = compiler_begin(in_5, "default")
+        dbegin1 = compiler_begin(in_6, "default")
+        dbegin2 = compiler_begin(in_7, "default")
+        node3 = relay.subtract(dbegin0, dbegin1)
+        node4 = relay.subtract(dbegin2, node3)
+        dend0 = compiler_end(node4, "default")
 
-        begin4 = compiler_begin(node4, "test")
+        begin4 = compiler_begin(dend0, "test")
         begin5 = compiler_begin(in_9, "test")
         node5 = relay.add(node2, begin4)
         end1 = compiler_end(node5, "test")
 
-        node6 = relay.subtract(in_8, end1)
+        dbegin4 = compiler_begin(end1, "default")
+        dbegin5 = compiler_begin(in_8, "default")
+        node6 = relay.subtract(dbegin5, dbegin4)
+        dend1 = compiler_end(node6, "default")
 
         node7 = relay.add(begin5, node5)
         end2 = compiler_end(node7, "test")
         begin6 = compiler_begin(end2, "test")
-        begin7 = compiler_begin(node6, "test")
+        begin7 = compiler_begin(dend1, "test")
 
         node8 = relay.add(begin7, begin6)
 

tests/python/relay/test_pass_partition_graph.py

@@ -17,6 +17,7 @@
 """Unit tests for graph partitioning."""
 import os
 import sys
+
 import numpy as np
 import pytest
 
@@ -26,8 +27,12 @@ from tvm import relay
 from tvm import runtime
 from tvm.relay import transform
 from tvm.contrib import util
-from tvm.relay.op.annotation import compiler_begin, compiler_end
+from tvm.relay import transform
+from tvm.relay.backend import compile_engine
 from tvm.relay.expr_functor import ExprMutator
+from tvm.relay.op.annotation import compiler_begin, compiler_end
+from tvm.runtime import container
+
 
 # Leverage the pass manager to write a simple white list based annotator
 @transform.function_pass(opt_level=0)
@@ -188,6 +193,7 @@ def check_result(mod, map_inputs, out_shape, result, tol=1e-5, target="llvm",
         return lib
 
     def check_vm_result():
+        compile_engine.get().clear()
         with relay.build_config(opt_level=3):
             exe = relay.vm.compile(mod, target=target, params=params)
         code, lib = exe.save()
@@ -199,6 +205,7 @@ def check_result(mod, map_inputs, out_shape, result, tol=1e-5, target="llvm",
         tvm.testing.assert_allclose(out.asnumpy(), result, rtol=tol, atol=tol)
 
     def check_graph_runtime_result():
+        compile_engine.get().clear()
         with relay.build_config(opt_level=3):
             json, lib, param = relay.build(mod, target=target, params=params)
         lib = update_lib(lib)
@@ -449,9 +456,9 @@ def test_extern_dnnl_mobilenet():
     mod, params = relay.testing.mobilenet.get_workload(
         batch_size=1, dtype='float32')
 
-    op_list = ["nn.conv2d", "nn.dense", "nn.relu", "add"]
     mod["main"] = relay.build_module.bind_params_by_name(mod["main"], params)
-    mod = WhiteListAnnotator(op_list, "dnnl")(mod)
+    mod = transform.AnnotateTarget(["dnnl"])(mod)
+    mod = transform.MergeCompilerRegions()(mod)
     mod = transform.PartitionGraph()(mod)
     i_data = np.random.uniform(0, 1, ishape).astype(dtype)
 
@@ -851,6 +858,7 @@ if __name__ == "__main__":
     test_extern_ccompiler_default_ops()
     test_extern_ccompiler()
     test_extern_dnnl()
+    # TODO(@comaniac, @zhiics): Fix constant node and re-open this case.
     #test_extern_dnnl_mobilenet()
     test_function_lifting()
     test_function_lifting_inline()