[mlir][sparse] introduce sparse_tensor.coiterate operation.

Author: Peiming Liu

mlir/include/mlir/Dialect/SparseTensor/IR/SparseTensor.h

@@ -61,37 +61,62 @@ struct COOSegment {
 /// A simple wrapper to encode a bitset of (at most 64) levels, currently used
 /// by `sparse_tensor.iterate` operation for the set of levels on which the
 /// coordinates should be loaded.
-class LevelSet {
-  uint64_t bits = 0;
+class I64BitSet {
+  uint64_t storage = 0;
 
 public:
-  LevelSet() = default;
-  explicit LevelSet(uint64_t bits) : bits(bits) {}
-  operator uint64_t() const { return bits; }
+  using const_set_bits_iterator = llvm::const_set_bits_iterator_impl<I64BitSet>;
+  const_set_bits_iterator begin() const {
+    return const_set_bits_iterator(*this);
+  }
+  const_set_bits_iterator end() const {
+    return const_set_bits_iterator(*this, -1);
+  }
+  iterator_range<const_set_bits_iterator> bits() const {
+    return make_range(begin(), end());
+  }
+
+  I64BitSet() = default;
+  explicit I64BitSet(uint64_t bits) : storage(bits) {}
+  operator uint64_t() const { return storage; }
 
-  LevelSet &set(unsigned i) {
+  I64BitSet &set(unsigned i) {
     assert(i < 64);
-    bits |= static_cast<uint64_t>(0x01u) << i;
+    storage |= static_cast<uint64_t>(0x01u) << i;
     return *this;
   }
 
-  LevelSet &operator|=(LevelSet lhs) {
-    bits |= static_cast<uint64_t>(lhs);
+  I64BitSet &operator|=(I64BitSet lhs) {
+    storage |= static_cast<uint64_t>(lhs);
     return *this;
   }
 
-  LevelSet &lshift(unsigned offset) {
-    bits = bits << offset;
+  I64BitSet &lshift(unsigned offset) {
+    storage = storage << offset;
     return *this;
   }
 
+  // Needed by `llvm::const_set_bits_iterator_impl`.
+  int find_first() const { return min(); }
+  int find_next(unsigned prev) const {
+    if (prev >= max())
+      return -1;
+
+    uint64_t b = storage >> (prev + 1);
+    if (b == 0)
+      return -1;
+
+    return llvm::countr_zero(b) + prev + 1;
+  }
+
   bool operator[](unsigned i) const {
     assert(i < 64);
-    return (bits & (1 << i)) != 0;
+    return (storage & (1 << i)) != 0;
   }
-  unsigned max() const { return 64 - llvm::countl_zero(bits); }
-  unsigned count() const { return llvm::popcount(bits); }
-  bool empty() const { return bits == 0; }
+  unsigned min() const { return llvm::countr_zero(storage); }
+  unsigned max() const { return 64 - llvm::countl_zero(storage); }
+  unsigned count() const { return llvm::popcount(storage); }
+  bool empty() const { return storage == 0; }
 };
 
 } // namespace sparse_tensor

mlir/include/mlir/Dialect/SparseTensor/IR/SparseTensorAttrDefs.td

@@ -24,16 +24,17 @@ class SparseTensor_Attr<string name,
 // sparse tensor levels.
 //===----------------------------------------------------------------------===//
 
-def LevelSetAttr :
-    TypedAttrBase<
-      I64, "IntegerAttr",
+def I64BitSetAttr : TypedAttrBase<I64, "IntegerAttr",
       And<[CPred<"::llvm::isa<::mlir::IntegerAttr>($_self)">,
            CPred<"::llvm::cast<::mlir::IntegerAttr>($_self).getType().isInteger(64)">]>,
       "LevelSet attribute"> {
-  let returnType = [{::mlir::sparse_tensor::LevelSet}];
-  let convertFromStorage = [{::mlir::sparse_tensor::LevelSet($_self.getValue().getZExtValue())}];
+  let returnType = [{::mlir::sparse_tensor::I64BitSet}];
+  let convertFromStorage = [{::mlir::sparse_tensor::I64BitSet($_self.getValue().getZExtValue())}];
 }
 
+def I64BitSetArrayAttr :
+    TypedArrayAttrBase<I64BitSetAttr, "I64BitSet array attribute">;
+
 //===----------------------------------------------------------------------===//
 // These attributes are just like `IndexAttr` except that they clarify whether
 // the index refers to a dimension (an axis of the semantic tensor) or a level

mlir/include/mlir/Dialect/SparseTensor/IR/SparseTensorOps.td

@@ -1306,7 +1306,7 @@ def SparseTensor_SelectOp : SparseTensor_Op<"select", [Pure, SameOperandsAndResu
 
 def SparseTensor_YieldOp : SparseTensor_Op<"yield", [Pure, Terminator,
     ParentOneOf<["BinaryOp", "UnaryOp", "ReduceOp", "SelectOp",
-                 "ForeachOp", "IterateOp"]>]> {
+                 "ForeachOp", "IterateOp", "CoIterateOp"]>]> {
   let summary = "Yield from sparse_tensor set-like operations";
   let description = [{
       Yields a value from within a `binary`, `unary`, `reduce`,
@@ -1604,14 +1604,14 @@ def IterateOp : SparseTensor_Op<"iterate",
 
   let arguments = (ins AnySparseIterSpace:$iterSpace,
                        Variadic<AnyType>:$initArgs,
-                       LevelSetAttr:$crdUsedLvls);
+                       I64BitSetAttr:$crdUsedLvls);
   let results = (outs Variadic<AnyType>:$results);
   let regions = (region SizedRegion<1>:$region);
 
   let skipDefaultBuilders = 1;
   let builders = [
     OpBuilder<(ins "Value":$iterSpace, "ValueRange":$initArgs)>,
-    OpBuilder<(ins "Value":$iterSpace, "ValueRange":$initArgs, "LevelSet" :$crdUsedLvls)>
+    OpBuilder<(ins "Value":$iterSpace, "ValueRange":$initArgs, "I64BitSet" :$crdUsedLvls)>
   ];
 
   let extraClassDeclaration = [{
@@ -1644,6 +1644,123 @@ def IterateOp : SparseTensor_Op<"iterate",
   let hasCustomAssemblyFormat = 1;
 }
 
+def SparseTensor_CoIterateOp : SparseTensor_Op<"coiterate",
+    [AttrSizedOperandSegments,
+     SingleBlockImplicitTerminator<"sparse_tensor::YieldOp">,
+     RecursiveMemoryEffects]> {
+  let summary = "CoIterates over a set of sparse iteration spaces";
+  let description = [{
+      The `sparse_tensor.coiterate` operation represents a loop (nest) over
+      the a set of iteration spaces.
+      The operation can have multiple regions, with each of them defining a
+      case to compute a result at the current iterations. The case condition
+      is defined solely based on the pattern of specified iterators.
+      For example:
+      ```mlir
+      %ret = sparse_tensor.coiterate (%sp1, %sp2) at(%coord) iter_args(%arg = %init)
+           : (!sparse_tensor.iter_space<#CSR, lvls = 0>,
+              !sparse_tensor.iter_space<#COO, lvls = 0>)
+           -> index
+      case %it1, _ {
+        // %coord is specifed in space %sp1 but *NOT* specified in space %sp2.
+      }
+      case %it1, %it2 {
+        // %coord is specifed in *BOTH* spaces %sp1 and %sp2.
+      }
+      ```
+
+      `sparse_tensor.coiterate` can also operate on loop-carried variables.
+      It returns the final values after loop termination.
+      The initial values of the variables are passed as additional SSA operands
+      to the iterator SSA value and used coordinate SSA values.
+      Each operation region has variadic arguments for specified (used), one argument
+      for each loop-carried variable, representing the value of the variable
+      at the current iteration, followed by a list of arguments for iterators.
+      The body region must contain exactly one block that terminates with
+      `sparse_tensor.yield`.
+
+      The results of an `sparse_tensor.coiterate` hold the final values after
+      the last iteration. If the `sparse_tensor.coiterate` defines any values,
+      a yield must be explicitly present in every region defined in the operation.
+      The number and types of the `sparse_tensor.coiterate` results must match
+      the initial values in the iter_args binding and the yield operands.
+
+
+      A `sparse_tensor.coiterate` example that does elementwise addition between two
+      sparse vectors.
+
+
+      ```mlir
+      %ret = sparse_tensor.coiterate (%sp1, %sp2) at(%coord) iter_args(%arg = %init)
+           : (!sparse_tensor.iter_space<#CSR, lvls = 0>,
+              !sparse_tensor.iter_space<#CSR, lvls = 0>)
+           -> tensor<?xindex, #CSR>
+      case %it1, _ {
+         // v = v1 + 0 = v1
+         %v1 = sparse_tensor.extract_value %t1 at %it1 : index
+         %yield = sparse_tensor.insert %v1 into %arg[%coord]
+         sparse_tensor.yield %yield
+      }
+      case _, %it2 {
+         // v = v2 + 0 = v2
+         %v2 = sparse_tensor.extract_value %t2 at %it2 : index
+         %yield = sparse_tensor.insert %v1 into %arg[%coord]
+         sparse_tensor.yield %yield
+      }
+      case %it1, %it2 {
+         // v = v1 + v2
+         %v1 = sparse_tensor.extract_value %t1 at %it1 : index
+         %v2 = sparse_tensor.extract_value %t2 at %it2 : index
+         %v = arith.addi %v1, %v2 : index
+         %yield = sparse_tensor.insert %v into %arg[%coord]
+         sparse_tensor.yield %yield
+      }
+      ```
+  }];
+
+  let arguments = (ins Variadic<AnySparseIterSpace>:$iterSpaces,
+                       Variadic<AnyType>:$initArgs,
+                       I64BitSetAttr:$crdUsedLvls,
+                       I64BitSetArrayAttr:$cases);
+  let results = (outs Variadic<AnyType>:$results);
+  let regions = (region VariadicRegion<SizedRegion<1>>:$caseRegions);
+
+  let extraClassDeclaration = [{
+    unsigned getSpaceDim() {
+      return llvm::cast<::mlir::sparse_tensor::IterSpaceType>(
+                 getIterSpaces().front().getType())
+          .getSpaceDim();
+    }
+    I64BitSet getRegionDefinedSpace(unsigned regionIdx) {
+      return I64BitSet(llvm::cast<IntegerAttr>(getCases()[regionIdx])
+                           .getValue().getZExtValue());
+    }
+    // The block arguments starts with referenced coordinates, follows by
+    // user-provided iteration arguments and ends with iterators.
+    Block::BlockArgListType getCrds(unsigned regionIdx) {
+      return getRegion(regionIdx).getArguments()
+          .take_front(getCrdUsedLvls().count());
+    }
+    unsigned getNumRegionIterArgs(unsigned regionIdx) {
+      return getInitArgs().size();
+    }
+    Block::BlockArgListType getRegionIterArgs(unsigned regionIdx) {
+      return getRegion(regionIdx).getArguments()
+          .slice(getCrdUsedLvls().count(), getNumRegionIterArgs(regionIdx));
+    }
+    Block::BlockArgListType getRegionIterators(unsigned regionIdx) {
+      return getRegion(regionIdx).getArguments()
+          .take_back(getRegionDefinedSpace(regionIdx).count());
+    }
+  }];
+
+  // TODO:
+  // let hasVerifier = 1;
+  // let hasRegionVerifier = 1;
+  // let hasCanonicalizer = 1;
+  let hasCustomAssemblyFormat = 1;
+}
+
 //===----------------------------------------------------------------------===//
 // Sparse Tensor Debugging and Test-Only Operations.
 //===----------------------------------------------------------------------===//