maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/mlir/lib/Transforms/MaterializeVectors.cpp

779 lines
33 KiB
C++
Raw Blame History

//===- MaterializeVectors.cpp - MaterializeVectors Pass Impl ----*- C++ -*-===//
//
// Copyright 2019 The MLIR Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// =============================================================================
//
// This file implements target-dependent materialization of super-vectors to
// vectors of the proper size for the hardware.
//
//===----------------------------------------------------------------------===//
#include "mlir/AffineOps/AffineOps.h"
#include "mlir/Analysis/AffineAnalysis.h"
#include "mlir/Analysis/Dominance.h"
#include "mlir/Analysis/LoopAnalysis.h"
#include "mlir/Analysis/NestedMatcher.h"
#include "mlir/Analysis/SliceAnalysis.h"
#include "mlir/Analysis/Utils.h"
#include "mlir/Analysis/VectorAnalysis.h"
#include "mlir/IR/AffineExpr.h"
#include "mlir/IR/AffineMap.h"
#include "mlir/IR/Attributes.h"
#include "mlir/IR/Builders.h"
#include "mlir/IR/Location.h"
#include "mlir/IR/OperationSupport.h"
#include "mlir/IR/Types.h"
#include "mlir/Pass/Pass.h"
#include "mlir/StandardOps/Ops.h"
#include "mlir/SuperVectorOps/SuperVectorOps.h"
#include "mlir/Support/Functional.h"
#include "mlir/Support/LLVM.h"
#include "mlir/Transforms/Passes.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
///
/// Implements target-dependent materialization of virtual super-vectors to
/// vectors of the proper size for the hardware.
///
/// While the physical vector size is target-dependent, the pass is written in
/// a target-independent way: the target vector size is specified as a parameter
/// to the pass. This pass is thus a partial lowering that opens the "greybox"
/// that is the super-vector abstraction. In particular, this pass can turn the
/// vector_transfer_read and vector_transfer_write ops in either:
/// 1. a loop nest with either scalar and vector load/store instructions; or
/// 2. a loop-nest with DmaStartOp / DmaWaitOp; or
/// 3. a pre-existing blackbox library call that can be written manually or
/// synthesized using search and superoptimization.
/// An important feature that either of these 3 target lowering abstractions
/// must handle is the handling of "non-effecting" padding with the proper
/// neutral element in order to guarantee that all "partial tiles" are actually
/// "full tiles" in practice.
///
/// In particular this pass is a MLIR-MLIR rewriting and does not concern itself
/// with target-specific instruction-selection and register allocation. These
/// will happen downstream in LLVM.
///
/// In this sense, despite performing lowering to a target-dependent size, this
/// pass is still target-agnostic.
///
/// Implementation details
/// ======================
/// The current decisions made by the super-vectorization pass guarantee that
/// use-def chains do not escape an enclosing vectorized AffineForOp. In other
/// words, this pass operates on a scoped program slice. Furthermore, since we
/// do not vectorize in the presence of conditionals for now, sliced chains are
/// guaranteed not to escape the innermost scope, which has to be either the top
/// Function scope or the innermost loop scope, by construction. As a
/// consequence, the implementation just starts from vector_transfer_write
/// operations and builds the slice scoped the innermost loop enclosing the
/// current vector_transfer_write. These assumptions and the implementation
/// details are subject to revision in the future.
///
/// Example
/// ========
/// In the following, the single vector_transfer_write op operates on a
/// vector<4x4x4xf32>. Let's assume the HW supports vector<4x4xf32>.
/// Materialization is achieved by instantiating each occurrence of the leading
/// dimension of vector<4x4x4xf32> into a vector<4x4xf32>.
/// The program transformation that implements this instantiation is a
/// multi-loop unroll-and-jam (it can be partial or full depending on the ratio
/// of super-vector shape to HW-vector shape).
///
/// As a simple case, the following:
///
/// ```mlir
/// mlfunc @materialize(%M : index, %N : index, %O : index, %P : index) {
/// %A = alloc (%M, %N, %O, %P) : memref<?x?x?x?xf32, 0>
/// %f1 = constant splat<vector<4x4x4xf32>, 1.000000e+00> :
/// vector<4x4x4xf32> affine.for %i0 = 0 to %M step 4 {
/// affine.for %i1 = 0 to %N step 4 {
/// affine.for %i2 = 0 to %O {
/// affine.for %i3 = 0 to %P step 4 {
/// vector_transfer_write %f1, %A, %i0, %i1, %i2, %i3
/// {permutation_map: (d0, d1, d2, d3) -> (d3, d1, d0)} :
/// vector<4x4x4xf32>, memref<?x?x?x?xf32, 0>,
/// index, index, index, index
/// }}}}
/// return
/// }
/// ```
///
/// is instantiated by unroll-and-jam (just unroll in this case) into:
///
/// ```mlir
/// mlfunc @materialize(%M : index, %N : index, %O : index, %P : index) {
/// %A = alloc (%M, %N, %O, %P) : memref<?x?x?x?xf32, 0>
/// %f1 = constant splat<vector<4x4xf32>, 1.000000e+00> : vector<4x4x4xf32>
/// affine.for %i0 = 0 to %arg0 step 4 {
/// affine.for %i1 = 0 to %arg1 step 4 {
/// affine.for %i2 = 0 to %arg2 {
/// affine.for %i3 = 0 to %arg3 step 4 {
/// %1 = affine.apply (d0, d1, d2, d3) -> (d0, d1, d2, d3)
/// (%i0, %i1, %i2, %i3)
/// vector_transfer_write f1, %0, %1#0, %1#1, %1#2, %1#3
/// {permutation_map: (d0, d1, d2, d3) -> (d1, d0)} :
/// vector<4x4xf32>, memref<?x?x?x?xf32>,
/// index, index, index, index
/// %2 = affine.apply (d0, d1, d2, d3) -> (d0, d1, d2, d3 + 1)
/// (%i0, %i1, %i2, %i3)
/// vector_transfer_write {{.*}}, %0, %2#0, %2#1, %2#2, %2#3
/// {permutation_map: (d0, d1, d2, d3) -> (d1, d0)} :
/// vector<4x4xf32>, memref<?x?x?x?xf32>,
/// index, index, index, index
/// %3 = affine.apply (d0, d1, d2, d3) -> (d0, d1, d2, d3 + 2)
/// (%i0, %i1, %i2, %i3)
/// vector_transfer_write {{.*}}, %0, %3#0, %3#1, %3#2, %3#3
/// {permutation_map: (d0, d1, d2, d3) -> (d1, d0)} :
/// vector<4x4xf32>, memref<?x?x?x?xf32>,
/// index, index, index, index
/// %4 = affine.apply (d0, d1, d2, d3) -> (d0, d1, d2, d3 + 3)
/// (%i0, %i1, %i2, %i3)
/// vector_transfer_write {{.*}}, %0, %4#0, %4#1, %4#2, %4#3
/// {permutation_map: (d0, d1, d2, d3) -> (d1, d0)} :
/// vector<4x4xf32>, memref<?x?x?x?xf32>,
/// index, index, index, index
/// }}}}
/// return
/// }
/// ```
using llvm::dbgs;
using llvm::DenseSet;
using llvm::SetVector;
using namespace mlir;
using functional::makePtrDynCaster;
using functional::map;
static llvm::cl::list<int>
clVectorSize("vector-size",
llvm::cl::desc("Specify the HW vector size for vectorization"),
llvm::cl::ZeroOrMore);
#define DEBUG_TYPE "materialize-vect"
namespace {
struct MaterializationState {
/// In practice, the determination of the HW-specific vector type to use when
/// lowering a super-vector type must be based on the elemental type. The
/// elemental type must be retrieved from the super-vector type. In the future
/// information about hardware vector type for a particular elemental type
/// will be part of the contract between MLIR and the backend.
///
/// For example, 8xf32 has the same size as 16xf16 but the targeted HW itself
/// may exhibit the following property:
/// 1. have a special unit for a 128xf16 datapath;
/// 2. no F16 FPU support on the regular 8xf32/16xf16 vector datapath.
///
/// For now, we just assume hwVectorSize has the proper information regardless
/// of the type and we assert everything is f32.
/// TODO(ntv): relax the assumptions on admissible element type once a
/// contract exists.
MaterializationState() : hwVectorSize(clVectorSize.size(), 0) {
std::copy(clVectorSize.begin(), clVectorSize.end(), hwVectorSize.begin());
}
SmallVector<int64_t, 8> hwVectorSize;
VectorType superVectorType;
VectorType hwVectorType;
SmallVector<unsigned, 8> hwVectorInstance;
DenseMap<Value *, Value *> *substitutionsMap;
};
struct MaterializeVectorsPass : public FunctionPass<MaterializeVectorsPass> {
void runOnFunction() override;
};
} // end anonymous namespace
/// Given a shape with sizes greater than 0 along all dimensions,
/// returns the distance, in number of elements, between a slice in a dimension
/// and the next slice in the same dimension.
/// e.g. shape[3, 4, 5] -> strides[20, 5, 1]
static SmallVector<unsigned, 8> makeStrides(ArrayRef<unsigned> shape) {
SmallVector<unsigned, 8> tmp;
tmp.reserve(shape.size());
unsigned running = 1;
for (auto rit = shape.rbegin(), reit = shape.rend(); rit != reit; ++rit) {
assert(*rit > 0 && "size must be greater than 0 along all dimensions of "
"shape");
tmp.push_back(running);
running *= *rit;
}
return SmallVector<unsigned, 8>(tmp.rbegin(), tmp.rend());
}
/// Given a shape with sizes greater than 0 along all dimensions, returns the
/// delinearized components of linearIndex along shape.
static SmallVector<unsigned, 8> delinearize(unsigned linearIndex,
ArrayRef<unsigned> shape) {
SmallVector<unsigned, 8> res;
res.reserve(shape.size());
auto strides = makeStrides(shape);
for (unsigned idx = 0; idx < strides.size(); ++idx) {
assert(strides[idx] > 0);
auto val = linearIndex / strides[idx];
res.push_back(val);
assert((val >= 0 && val < shape[idx]) &&
"delinearization is out of bounds");
linearIndex %= strides[idx];
}
// Sanity check.
assert(linearIndex == 0 && "linear index constructed from shape must "
"have 0 remainder after delinearization");
return res;
}
static Instruction *instantiate(FuncBuilder *b, Instruction *opInst,
VectorType hwVectorType,
DenseMap<Value *, Value *> *substitutionsMap);
/// Not all Values belong to a program slice scoped within the immediately
/// enclosing loop.
/// One simple example is constants defined outside the innermost loop scope.
/// For such cases the substitutionsMap has no entry and we allow an additional
/// insertion.
/// For now, this is limited to ConstantOp because we do not vectorize loop
/// indices and will need to be extended in the future.
///
/// If substitution fails, returns nullptr.
static Value *substitute(Value *v, VectorType hwVectorType,
DenseMap<Value *, Value *> *substitutionsMap) {
auto it = substitutionsMap->find(v);
if (it == substitutionsMap->end()) {
auto *opInst = v->getDefiningOp();
if (opInst->isa<ConstantOp>()) {
FuncBuilder b(opInst);
auto *inst = instantiate(&b, opInst, hwVectorType, substitutionsMap);
auto res =
substitutionsMap->insert(std::make_pair(v, inst->getResult(0)));
assert(res.second && "Insertion failed");
return res.first->second;
}
v->getDefiningOp()->emitError("Missing substitution");
return nullptr;
}
return it->second;
}
/// Returns a list of single result AffineApplyOps that reindex the
/// `memRefIndices` by the multi-dimensional `hwVectorInstance`. This is used by
/// the function that materializes a vector_transfer operation to use hardware
/// vector types instead of super-vector types.
///
/// The general problem this function solves is as follows:
/// Assume a vector_transfer operation at the super-vector granularity that has
/// `l` enclosing loops (AffineForOp). Assume the vector transfer operation
/// operates on a MemRef of rank `r`, a super-vector of rank `s` and a hardware
/// vector of rank `h`. For the purpose of illustration assume l==4, r==3, s==2,
/// h==1 and that the super-vector is vector<3x32xf32> and the hardware vector
/// is vector<8xf32>. Assume the following MLIR snippet after
/// super-vectorization has been applied:
///
/// ```mlir
/// affine.for %i0 = 0 to %M {
/// affine.for %i1 = 0 to %N step 3 {
/// affine.for %i2 = 0 to %O {
/// affine.for %i3 = 0 to %P step 32 {
/// %r = vector_transfer_read(%A, map(%i..)#0, map(%i..)#1, map(%i..)#2)
/// -> vector<3x32xf32>
/// ...
/// }}}}
/// ```
///
/// where map denotes an AffineMap operating on enclosing loops with properties
/// compatible for vectorization (i.e. some contiguity left unspecified here).
/// Note that the vectorized loops are %i1 and %i3.
/// This function translates the vector_transfer_read operation to multiple
/// instances of vector_transfer_read that operate on vector<8x32>.
///
/// Without loss of generality, we assume hwVectorInstance is: {2, 1}.
/// The only constraints on hwVectorInstance is they belong to:
/// [0, 2] x [0, 3], which is the span of ratio of super-vector shape to
/// hardware vector shape in our example.
///
/// This function instantiates the iteration <2, 1> of vector_transfer_read
/// into the set of operations in pseudo-MLIR:
///
/// ```mlir
/// map2 = (d0, d1, d2, d3) -> (d0, d1 + 2, d2, d3 + 1 * 8)
/// map3 = map o map2 // where o denotes composition
/// %r = vector_transfer_read(%A, map3(%i..)#0, map3(%i..)#1, map3(%i..)#2)
/// -> vector<3x32xf32>
/// ```
///
/// Practical considerations
/// ========================
/// For now, `map` is assumed to be the identity map and the indices are
/// specified just as vector_transfer_read(%A, %i0, %i1, %i2, %i3). This will be
/// extended in the future once we have a proper Op for vector transfers.
/// Additionally, the example above is specified in pseudo-MLIR form; once we
/// have proper support for generic maps we can generate the code and show
/// actual MLIR.
///
/// TODO(ntv): support a concrete AffineMap and compose with it.
/// TODO(ntv): these implementation details should be captured in a
/// vectorization trait at the op level directly.
static SmallVector<mlir::Value *, 8>
reindexAffineIndices(FuncBuilder *b, VectorType hwVectorType,
ArrayRef<unsigned> hwVectorInstance,
ArrayRef<Value *> memrefIndices) {
auto vectorShape = hwVectorType.getShape();
assert(hwVectorInstance.size() >= vectorShape.size());
unsigned numIndices = memrefIndices.size();
auto numMemRefIndices = numIndices - hwVectorInstance.size();
auto numSuperVectorIndices = hwVectorInstance.size() - vectorShape.size();
SmallVector<AffineExpr, 8> affineExprs;
// TODO(ntv): support a concrete map and composition.
unsigned i = 0;
// The first numMemRefIndices correspond to AffineForOp that have not been
// vectorized, the transformation is the identity on those.
for (i = 0; i < numMemRefIndices; ++i) {
auto d_i = b->getAffineDimExpr(i);
affineExprs.push_back(d_i);
}
// The next numSuperVectorIndices correspond to super-vector dimensions that
// do not have a hardware vector dimension counterpart. For those we only
// need to increment the index by the corresponding hwVectorInstance.
for (i = numMemRefIndices; i < numMemRefIndices + numSuperVectorIndices;
++i) {
auto d_i = b->getAffineDimExpr(i);
auto offset = hwVectorInstance[i - numMemRefIndices];
affineExprs.push_back(d_i + offset);
}
// The remaining indices correspond to super-vector dimensions that
// have a hardware vector dimension counterpart. For those we to increment the
// index by "hwVectorInstance" multiples of the corresponding hardware
// vector size.
for (; i < numIndices; ++i) {
auto d_i = b->getAffineDimExpr(i);
auto offset = hwVectorInstance[i - numMemRefIndices];
auto stride = vectorShape[i - numMemRefIndices - numSuperVectorIndices];
affineExprs.push_back(d_i + offset * stride);
}
// Create a bunch of single result AffineApplyOp.
SmallVector<mlir::Value *, 8> res;
res.reserve(affineExprs.size());
for (auto expr : affineExprs) {
auto map = AffineMap::get(numIndices, 0, expr, {});
res.push_back(makeComposedAffineApply(b, b->getInsertionPoint()->getLoc(),
map, memrefIndices));
}
return res;
}
/// Returns attributes with the following substitutions applied:
/// - constant splat is replaced by constant splat of `hwVectorType`.
/// TODO(ntv): add more substitutions on a per-need basis.
static SmallVector<NamedAttribute, 1>
materializeAttributes(Instruction *opInst, VectorType hwVectorType) {
SmallVector<NamedAttribute, 1> res;
for (auto a : opInst->getAttrs()) {
if (auto splat = a.second.dyn_cast<SplatElementsAttr>()) {
auto attr = SplatElementsAttr::get(hwVectorType, splat.getValue());
res.push_back(NamedAttribute(a.first, attr));
} else {
res.push_back(a);
}
}
return res;
}
/// Creates an instantiated version of `opInst`.
/// Ops other than VectorTransferReadOp/VectorTransferWriteOp require no
/// affine reindexing. Just substitute their Value operands and be done. For
/// this case the actual instance is irrelevant. Just use the values in
/// substitutionsMap.
///
/// If the underlying substitution fails, this fails too and returns nullptr.
static Instruction *instantiate(FuncBuilder *b, Instruction *opInst,
VectorType hwVectorType,
DenseMap<Value *, Value *> *substitutionsMap) {
assert(!opInst->isa<VectorTransferReadOp>() &&
"Should call the function specialized for VectorTransferReadOp");
assert(!opInst->isa<VectorTransferWriteOp>() &&
"Should call the function specialized for VectorTransferWriteOp");
if (opInst->getNumRegions() != 0)
return nullptr;
bool fail = false;
auto operands = map(
[hwVectorType, substitutionsMap, &fail](Value *v) -> Value * {
auto *res =
fail ? nullptr : substitute(v, hwVectorType, substitutionsMap);
fail |= !res;
return res;
},
opInst->getOperands());
if (fail)
return nullptr;
auto attrs = materializeAttributes(opInst, hwVectorType);
OperationState state(b->getContext(), opInst->getLoc(),
opInst->getName().getStringRef(), operands,
{hwVectorType}, attrs);
return b->createOperation(state);
}
/// Computes the permutationMap required for a VectorTransferOp from the memref
/// to the `hwVectorType`.
/// This is achieved by returning the projection of the permutationMap along the
/// dimensions of the super-vector type that remain in the hwVectorType.
/// In particular, if a dimension is fully instantiated (i.e. unrolled) then it
/// is projected out in the final result.
template <typename VectorTransferOpTy>
static AffineMap projectedPermutationMap(VectorTransferOpTy transfer,
VectorType hwVectorType) {
static_assert(
std::is_same<VectorTransferOpTy, VectorTransferReadOp>::value ||
std::is_same<VectorTransferOpTy, VectorTransferWriteOp>::value,
"Must be called on a VectorTransferOp");
auto superVectorType = transfer.getVectorType();
auto optionalRatio = shapeRatio(superVectorType, hwVectorType);
assert(optionalRatio &&
(optionalRatio->size() == superVectorType.getShape().size()) &&
"Shape and ratio not of the same size");
unsigned dim = 0;
SmallVector<AffineExpr, 4> keep;
MLIRContext *context = transfer.getContext();
functional::zipApply(
[&dim, &keep, context](int64_t shape, int64_t ratio) {
assert(shape >= ratio && "shape dim must be greater than ratio dim");
if (shape != ratio) {
// HW vector is not full instantiated along this dim, keep it.
keep.push_back(getAffineDimExpr(dim, context));
}
++dim;
},
superVectorType.getShape(), *optionalRatio);
auto permutationMap = transfer.getPermutationMap();
LLVM_DEBUG(permutationMap.print(dbgs() << "\npermutationMap: "));
if (keep.empty()) {
return permutationMap;
}
auto projectionMap = AffineMap::get(optionalRatio->size(), 0, keep, {});
LLVM_DEBUG(projectionMap.print(dbgs() << "\nprojectionMap: "));
return simplifyAffineMap(projectionMap.compose(permutationMap));
}
/// Creates an instantiated version of `read` for the instance of
/// `hwVectorInstance` when lowering from a super-vector type to
/// `hwVectorType`. `hwVectorInstance` represents one particular instance of
/// `hwVectorType` int the covering of the super-vector type. For a more
/// detailed description of the problem, see the description of
/// reindexAffineIndices.
static Instruction *instantiate(FuncBuilder *b, VectorTransferReadOp read,
VectorType hwVectorType,
ArrayRef<unsigned> hwVectorInstance,
DenseMap<Value *, Value *> *substitutionsMap) {
SmallVector<Value *, 8> indices =
map(makePtrDynCaster<Value>(), read.getIndices());
auto affineIndices =
reindexAffineIndices(b, hwVectorType, hwVectorInstance, indices);
auto map = projectedPermutationMap(read, hwVectorType);
if (!map) {
return nullptr;
}
auto cloned = b->create<VectorTransferReadOp>(read.getLoc(), hwVectorType,
read.getMemRef(), affineIndices,
map, read.getPaddingValue());
return cloned.getOperation();
}
/// Creates an instantiated version of `write` for the instance of
/// `hwVectorInstance` when lowering from a super-vector type to
/// `hwVectorType`. `hwVectorInstance` represents one particular instance of
/// `hwVectorType` int the covering of th3e super-vector type. For a more
/// detailed description of the problem, see the description of
/// reindexAffineIndices.
static Instruction *instantiate(FuncBuilder *b, VectorTransferWriteOp write,
VectorType hwVectorType,
ArrayRef<unsigned> hwVectorInstance,
DenseMap<Value *, Value *> *substitutionsMap) {
SmallVector<Value *, 8> indices =
map(makePtrDynCaster<Value>(), write.getIndices());
auto affineIndices =
reindexAffineIndices(b, hwVectorType, hwVectorInstance, indices);
auto cloned = b->create<VectorTransferWriteOp>(
write.getLoc(),
substitute(write.getVector(), hwVectorType, substitutionsMap),
write.getMemRef(), affineIndices,
projectedPermutationMap(write, hwVectorType));
return cloned.getOperation();
}
/// Returns `true` if inst instance is properly cloned and inserted, false
/// otherwise.
/// The multi-dimensional `hwVectorInstance` belongs to the shapeRatio of
/// super-vector type to hw vector type.
/// A cloned instance of `inst` is formed as follows:
/// 1. vector_transfer_read: the return `superVectorType` is replaced by
/// `hwVectorType`. Additionally, affine indices are reindexed with
/// `reindexAffineIndices` using `hwVectorInstance` and vector type
/// information;
/// 2. vector_transfer_write: the `valueToStore` type is simply substituted.
/// Since we operate on a topologically sorted slice, a substitution must
/// have been registered for non-constant ops. Additionally, affine indices
/// are reindexed in the same way as for vector_transfer_read;
/// 3. constant ops are splats of the super-vector type by construction.
/// They are cloned to a splat on the hw vector type with the same value;
/// 4. remaining ops are cloned to version of the op that returns a hw vector
/// type, all operands are substituted according to `substitutions`. Thanks
/// to the topological order of a slice, the substitution is always
/// possible.
///
/// Returns true on failure.
static bool instantiateMaterialization(Instruction *inst,
MaterializationState *state) {
LLVM_DEBUG(dbgs() << "\ninstantiate: " << *inst);
// Create a builder here for unroll-and-jam effects.
FuncBuilder b(inst);
// AffineApplyOp are ignored: instantiating the proper vector op will take
// care of AffineApplyOps by composing them properly.
if (inst->isa<AffineApplyOp>()) {
return false;
}
if (inst->getNumRegions() != 0)
return inst->emitError("NYI path Op with region");
if (auto write = inst->dyn_cast<VectorTransferWriteOp>()) {
auto *clone = instantiate(&b, write, state->hwVectorType,
state->hwVectorInstance, state->substitutionsMap);
return clone == nullptr;
}
if (auto read = inst->dyn_cast<VectorTransferReadOp>()) {
auto *clone = instantiate(&b, read, state->hwVectorType,
state->hwVectorInstance, state->substitutionsMap);
if (!clone) {
return true;
}
state->substitutionsMap->insert(
std::make_pair(read.getResult(), clone->getResult(0)));
return false;
}
// The only op with 0 results reaching this point must, by construction, be
// VectorTransferWriteOps and have been caught above. Ops with >= 2 results
// are not yet supported. So just support 1 result.
if (inst->getNumResults() != 1) {
return inst->emitError("NYI: ops with != 1 results");
}
if (inst->getResult(0)->getType() != state->superVectorType) {
return inst->emitError("Op does not return a supervector.");
}
auto *clone =
instantiate(&b, inst, state->hwVectorType, state->substitutionsMap);
if (!clone) {
return true;
}
state->substitutionsMap->insert(
std::make_pair(inst->getResult(0), clone->getResult(0)));
return false;
}
/// Takes a slice and rewrites the operations in it so that occurrences
/// of `superVectorType` are replaced by `hwVectorType`.
///
/// Implementation
/// ==============
/// 1. computes the shape ratio of super-vector to HW vector shapes. This
/// gives for each op in the slice, how many instantiations are required
/// in each dimension;
/// 2. performs the concrete materialization. Note that in a first
/// implementation we use full unrolling because it pragmatically removes
/// the need to explicitly materialize an AllocOp. Thanks to the properties
/// of super-vectors, this unrolling is always possible and simple:
/// vectorizing to a super-vector abstraction already achieved the
/// equivalent of loop strip-mining + loop sinking and encoded this in the
/// vector type.
///
/// Returns true on failure.
///
/// TODO(ntv): materialized allocs.
/// TODO(ntv): full loops + materialized allocs.
/// TODO(ntv): partial unrolling + materialized allocs.
static bool emitSlice(MaterializationState *state,
SetVector<Instruction *> *slice) {
auto ratio = shapeRatio(state->superVectorType, state->hwVectorType);
assert(ratio.hasValue() &&
"ratio of super-vector to HW-vector shape is not integral");
// The number of integer points in a hyperrectangular region is:
// shape[0] * strides[0].
auto numValueToUnroll = (*ratio)[0] * makeStrides(*ratio)[0];
// Full unrolling to hardware vectors in a first approximation.
for (unsigned idx = 0; idx < numValueToUnroll; ++idx) {
// Fresh RAII instanceIndices and substitutionsMap.
MaterializationState scopedState = *state;
scopedState.hwVectorInstance = delinearize(idx, *ratio);
DenseMap<Value *, Value *> substitutionMap;
scopedState.substitutionsMap = &substitutionMap;
// slice are topologically sorted, we can just clone them in order.
for (auto *inst : *slice) {
auto fail = instantiateMaterialization(inst, &scopedState);
if (fail) {
inst->emitError("Unhandled super-vector materialization failure");
return true;
}
}
}
LLVM_DEBUG(dbgs() << "\nMLFunction is now\n");
LLVM_DEBUG((*slice)[0]->getFunction()->print(dbgs()));
// slice are topologically sorted, we can just erase them in reverse
// order. Reverse iterator does not just work simply with an operator*
// dereference.
for (int idx = slice->size() - 1; idx >= 0; --idx) {
LLVM_DEBUG(dbgs() << "\nErase: ");
LLVM_DEBUG((*slice)[idx]->print(dbgs()));
(*slice)[idx]->erase();
}
return false;
}
/// Materializes super-vector types into concrete hw vector types as follows:
/// 1. start from super-vector terminators (current vector_transfer_write
/// ops);
/// 2. collect all the instructions that can be reached by transitive use-defs
/// chains;
/// 3. get the superVectorType for this particular terminator and the
/// corresponding hardware vector type (for now limited to F32)
/// TODO(ntv): be more general than F32.
/// 4. emit the transitive useDef set to operate on the finer-grain vector
/// types.
///
/// Notes
/// =====
/// The `slice` is sorted in topological order by construction.
/// Additionally, this set is limited to instructions in the same lexical scope
/// because we currently disallow vectorization of defs that come from another
/// scope.
/// TODO(ntv): please document return value.
static bool materialize(Function *f,
const SetVector<Instruction *> &terminators,
MaterializationState *state) {
DenseSet<Instruction *> seen;
DominanceInfo domInfo(f);
for (auto *term : terminators) {
// Short-circuit test, a given terminator may have been reached by some
// other previous transitive use-def chains.
if (seen.count(term) > 0) {
continue;
}
auto terminator = term->cast<VectorTransferWriteOp>();
LLVM_DEBUG(dbgs() << "\nFrom terminator:" << *term);
// Get the transitive use-defs starting from terminator, limited to the
// current enclosing scope of the terminator. See the top of the function
// Note for the justification of this restriction.
// TODO(ntv): relax scoping constraints.
auto *enclosingScope = term->getParentInst();
auto keepIfInSameScope = [enclosingScope, &domInfo](Instruction *inst) {
assert(inst && "NULL inst");
if (!enclosingScope) {
// by construction, everyone is always under the top scope (null scope).
return true;
}
return domInfo.properlyDominates(enclosingScope, inst);
};
SetVector<Instruction *> slice =
getSlice(term, keepIfInSameScope, keepIfInSameScope);
assert(!slice.empty());
// Sanity checks: transitive slice must be completely disjoint from
// what we have seen so far.
LLVM_DEBUG(dbgs() << "\nTransitive use-defs:");
for (auto *ud : slice) {
LLVM_DEBUG(dbgs() << "\nud:" << *ud);
assert(seen.count(ud) == 0 &&
"Transitive use-defs not disjoint from already seen");
seen.insert(ud);
}
// Emit the current slice.
// Set scoped super-vector and corresponding hw vector types.
state->superVectorType = terminator.getVectorType();
assert((state->superVectorType.getElementType() ==
FloatType::getF32(term->getContext())) &&
"Only f32 supported for now");
state->hwVectorType = VectorType::get(
state->hwVectorSize, state->superVectorType.getElementType());
auto fail = emitSlice(state, &slice);
if (fail) {
return true;
}
LLVM_DEBUG(dbgs() << "\nMLFunction is now\n");
LLVM_DEBUG(f->print(dbgs()));
}
return false;
}
void MaterializeVectorsPass::runOnFunction() {
// Thread-safe RAII local context, BumpPtrAllocator freed on exit.
NestedPatternContext mlContext;
// TODO(ntv): Check to see if this supports arbitrary top-level code.
Function *f = &getFunction();
if (f->getBlocks().size() != 1)
return;
using matcher::Op;
LLVM_DEBUG(dbgs() << "\nMaterializeVectors on Function\n");
LLVM_DEBUG(f->print(dbgs()));
MaterializationState state;
// Get the hardware vector type.
// TODO(ntv): get elemental type from super-vector type rather than force f32.
auto subVectorType =
VectorType::get(state.hwVectorSize, FloatType::getF32(f->getContext()));
// Capture terminators; i.e. vector_transfer_write ops involving a strict
// super-vector of subVectorType.
auto filter = [subVectorType](Instruction &inst) {
if (!inst.isa<VectorTransferWriteOp>()) {
return false;
}
return matcher::operatesOnSuperVectors(inst, subVectorType);
};
auto pat = Op(filter);
SmallVector<NestedMatch, 8> matches;
pat.match(f, &matches);
SetVector<Instruction *> terminators;
for (auto m : matches) {
terminators.insert(m.getMatchedInstruction());
}
if (materialize(f, terminators, &state))
signalPassFailure();
}
FunctionPassBase *mlir::createMaterializeVectorsPass() {
return new MaterializeVectorsPass();
}
static PassRegistration<MaterializeVectorsPass>
pass("materialize-vectors", "Materializes super-vectors to vectors of the "
"proper size for the hardware");
#undef DEBUG_TYPE
Reference in New Issue View Git Blame Copy Permalink