Google Git
Sign in
llvm / llvm-project / 9e3e1aad3161f4ce5301c3a59c7313ad83240a6d / . / mlir / lib / Transforms / Utils / LoopUtils.cpp
blob: c43ee3f70ea6395d345f0babbe9431b2c99e639f [file] [log] [blame]
//===- LoopUtils.cpp ---- Misc utilities for loop transformation ----------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file implements miscellaneous loop transformation routines.
//
//===----------------------------------------------------------------------===//
#include "mlir/Transforms/LoopUtils.h"
#include "mlir/Analysis/AffineAnalysis.h"
#include "mlir/Analysis/LoopAnalysis.h"
#include "mlir/Analysis/SliceAnalysis.h"
#include "mlir/Analysis/Utils.h"
#include "mlir/Dialect/Affine/IR/AffineOps.h"
#include "mlir/Dialect/Affine/IR/AffineValueMap.h"
#include "mlir/Dialect/Arithmetic/IR/Arithmetic.h"
#include "mlir/Dialect/MemRef/IR/MemRef.h"
#include "mlir/Dialect/SCF/SCF.h"
#include "mlir/IR/AffineMap.h"
#include "mlir/IR/BlockAndValueMapping.h"
#include "mlir/IR/BuiltinOps.h"
#include "mlir/IR/IntegerSet.h"
#include "mlir/Support/MathExtras.h"
#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
#include "mlir/Transforms/RegionUtils.h"
#include "mlir/Transforms/Utils.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/MapVector.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
#define DEBUG_TYPE "LoopUtils"
using namespace mlir;
using llvm::SmallMapVector;
namespace {
// This structure is to pass and return sets of loop parameters without
// confusing the order.
struct LoopParams {
Value lowerBound;
Value upperBound;
Value step;
};
} // namespace
/// Computes the cleanup loop lower bound of the loop being unrolled with
/// the specified unroll factor; this bound will also be upper bound of the main
/// part of the unrolled loop. Computes the bound as an AffineMap with its
/// operands or a null map when the trip count can't be expressed as an affine
/// expression.
static void getCleanupLoopLowerBound(AffineForOp forOp, unsigned unrollFactor,
AffineMap &map,
SmallVectorImpl<Value> &operands) {
auto lbMap = forOp.getLowerBoundMap();
// Single result lower bound map only.
if (lbMap.getNumResults() != 1) {
map = AffineMap();
return;
}
AffineMap tripCountMap;
SmallVector<Value, 4> tripCountOperands;
getTripCountMapAndOperands(forOp, &tripCountMap, &tripCountOperands);
// Sometimes the trip count cannot be expressed as an affine expression.
if (!tripCountMap) {
map = AffineMap();
return;
}
OpBuilder b(forOp);
auto lb = b.create<AffineApplyOp>(forOp.getLoc(), lbMap,
forOp.getLowerBoundOperands());
// For each upper bound expr, get the range.
// Eg: affine.for %i = lb to min (ub1, ub2),
// where tripCountExprs yield (tr1, tr2), we create affine.apply's:
// lb + tr1 - tr1 % ufactor, lb + tr2 - tr2 % ufactor; the results of all
// these affine.apply's make up the cleanup loop lower bound.
SmallVector<AffineExpr, 4> bumpExprs(tripCountMap.getNumResults());
SmallVector<Value, 4> bumpValues(tripCountMap.getNumResults());
int64_t step = forOp.getStep();
for (unsigned i = 0, e = tripCountMap.getNumResults(); i < e; i++) {
auto tripCountExpr = tripCountMap.getResult(i);
bumpExprs[i] = (tripCountExpr - tripCountExpr % unrollFactor) * step;
auto bumpMap = AffineMap::get(tripCountMap.getNumDims(),
tripCountMap.getNumSymbols(), bumpExprs[i]);
bumpValues[i] =
b.create<AffineApplyOp>(forOp.getLoc(), bumpMap, tripCountOperands);
}
SmallVector<AffineExpr, 4> newUbExprs(tripCountMap.getNumResults());
for (unsigned i = 0, e = bumpExprs.size(); i < e; i++)
newUbExprs[i] = b.getAffineDimExpr(0) + b.getAffineDimExpr(i + 1);
operands.clear();
operands.push_back(lb);
operands.append(bumpValues.begin(), bumpValues.end());
map = AffineMap::get(1 + tripCountMap.getNumResults(), 0, newUbExprs,
b.getContext());
// Simplify the map + operands.
fullyComposeAffineMapAndOperands(&map, &operands);
map = simplifyAffineMap(map);
canonicalizeMapAndOperands(&map, &operands);
// Remove any affine.apply's that became dead from the simplification above.
for (auto v : bumpValues)
if (v.use_empty())
v.getDefiningOp()->erase();
if (lb.use_empty())
lb.erase();
}
// Build the IR that performs ceil division of a positive value by a constant:
// ceildiv(a, B) = divis(a + (B-1), B)
// where divis is rounding-to-zero division.
static Value ceilDivPositive(OpBuilder &builder, Location loc, Value dividend,
int64_t divisor) {
assert(divisor > 0 && "expected positive divisor");
assert(dividend.getType().isIndex() && "expected index-typed value");
Value divisorMinusOneCst =
builder.create<arith::ConstantIndexOp>(loc, divisor - 1);
Value divisorCst = builder.create<arith::ConstantIndexOp>(loc, divisor);
Value sum = builder.create<arith::AddIOp>(loc, dividend, divisorMinusOneCst);
return builder.create<arith::DivSIOp>(loc, sum, divisorCst);
}
// Build the IR that performs ceil division of a positive value by another
// positive value:
// ceildiv(a, b) = divis(a + (b - 1), b)
// where divis is rounding-to-zero division.
static Value ceilDivPositive(OpBuilder &builder, Location loc, Value dividend,
Value divisor) {
assert(dividend.getType().isIndex() && "expected index-typed value");
Value cstOne = builder.create<arith::ConstantIndexOp>(loc, 1);
Value divisorMinusOne = builder.create<arith::SubIOp>(loc, divisor, cstOne);
Value sum = builder.create<arith::AddIOp>(loc, dividend, divisorMinusOne);
return builder.create<arith::DivSIOp>(loc, sum, divisor);
}
/// Helper to replace uses of loop carried values (iter_args) and loop
/// yield values while promoting single iteration affine.for and scf.for ops.
template <typename AffineOrSCFForOp>
static void replaceIterArgsAndYieldResults(AffineOrSCFForOp forOp) {
static_assert(
llvm::is_one_of<AffineOrSCFForOp, AffineForOp, scf::ForOp>::value,
"only for affine.for and scf.for ops");
// Replace uses of iter arguments with iter operands (initial values).
auto iterOperands = forOp.getIterOperands();
auto iterArgs = forOp.getRegionIterArgs();
for (auto e : llvm::zip(iterOperands, iterArgs))
std::get<1>(e).replaceAllUsesWith(std::get<0>(e));
// Replace uses of loop results with the values yielded by the loop.
auto outerResults = forOp.getResults();
auto innerResults = forOp.getBody()->getTerminator()->getOperands();
for (auto e : llvm::zip(outerResults, innerResults))
std::get<0>(e).replaceAllUsesWith(std::get<1>(e));
}
/// Promotes the loop body of a forOp to its containing block if the forOp
/// was known to have a single iteration.
// TODO: extend this for arbitrary affine bounds.
LogicalResult mlir::promoteIfSingleIteration(AffineForOp forOp) {
Optional<uint64_t> tripCount = getConstantTripCount(forOp);
if (!tripCount || tripCount.getValue() != 1)
return failure();
if (forOp.getLowerBoundMap().getNumResults() != 1)
return failure();
// Replaces all IV uses to its single iteration value.
auto iv = forOp.getInductionVar();
auto *parentBlock = forOp->getBlock();
if (!iv.use_empty()) {
if (forOp.hasConstantLowerBound()) {
OpBuilder topBuilder(forOp->getParentOfType<FuncOp>().getBody());
auto constOp = topBuilder.create<arith::ConstantIndexOp>(
forOp.getLoc(), forOp.getConstantLowerBound());
iv.replaceAllUsesWith(constOp);
} else {
auto lbOperands = forOp.getLowerBoundOperands();
auto lbMap = forOp.getLowerBoundMap();
OpBuilder builder(parentBlock, Block::iterator(forOp));
if (lbMap == builder.getDimIdentityMap()) {
// No need of generating an affine.apply.
iv.replaceAllUsesWith(lbOperands[0]);
} else {
auto affineApplyOp =
builder.create<AffineApplyOp>(forOp.getLoc(), lbMap, lbOperands);
iv.replaceAllUsesWith(affineApplyOp);
}
}
}
replaceIterArgsAndYieldResults(forOp);
// Move the loop body operations, except for its terminator, to the loop's
// containing block.
forOp.getBody()->back().erase();
parentBlock->getOperations().splice(Block::iterator(forOp),
forOp.getBody()->getOperations());
forOp.erase();
return success();
}
/// Promotes the loop body of a forOp to its containing block if the forOp
/// it can be determined that the loop has a single iteration.
LogicalResult mlir::promoteIfSingleIteration(scf::ForOp forOp) {
auto lbCstOp = forOp.lowerBound().getDefiningOp<arith::ConstantIndexOp>();
auto ubCstOp = forOp.upperBound().getDefiningOp<arith::ConstantIndexOp>();
auto stepCstOp = forOp.step().getDefiningOp<arith::ConstantIndexOp>();
if (!lbCstOp || !ubCstOp || !stepCstOp || lbCstOp.value() < 0 ||
ubCstOp.value() < 0 || stepCstOp.value() < 0)
return failure();
int64_t tripCount =
mlir::ceilDiv(ubCstOp.value() - lbCstOp.value(), stepCstOp.value());
if (tripCount != 1)
return failure();
auto iv = forOp.getInductionVar();
iv.replaceAllUsesWith(lbCstOp);
replaceIterArgsAndYieldResults(forOp);
// Move the loop body operations, except for its terminator, to the loop's
// containing block.
auto *parentBlock = forOp->getBlock();
forOp.getBody()->getTerminator()->erase();
parentBlock->getOperations().splice(Block::iterator(forOp),
forOp.getBody()->getOperations());
forOp.erase();
return success();
}
/// Promotes all single iteration 'for' ops in `f`, i.e., moves
/// their body into the containing Block.
void mlir::promoteSingleIterationLoops(FuncOp f) {
// Gathers all innermost loops through a post order pruned walk.
f.walk([](Operation *op) {
if (auto forOp = dyn_cast<AffineForOp>(op))
(void)promoteIfSingleIteration(forOp);
else if (auto forOp = dyn_cast<scf::ForOp>(op))
(void)promoteIfSingleIteration(forOp);
});
}
/// Generates an affine.for op with the specified lower and upper bounds
/// while generating the right IV remappings to realize shifts for operations in
/// its body. The operations that go into the loop body are specified in
/// opGroupQueue starting from the specified offset, and in that order. The
/// first element of the pair specifies the shift applied to that group of
/// operations; the shift is multiplied by the loop step before being applied.
/// Returns nullptr if the generated loop simplifies to a single iteration one.
static AffineForOp generateShiftedLoop(
AffineMap lbMap, AffineMap ubMap,
const std::vector<std::pair<uint64_t, ArrayRef<Operation *>>> &opGroupQueue,
unsigned offset, AffineForOp srcForOp, OpBuilder b) {
auto lbOperands = srcForOp.getLowerBoundOperands();
auto ubOperands = srcForOp.getUpperBoundOperands();
assert(lbMap.getNumInputs() == lbOperands.size());
assert(ubMap.getNumInputs() == ubOperands.size());
auto loopChunk = b.create<AffineForOp>(srcForOp.getLoc(), lbOperands, lbMap,
ubOperands, ubMap, srcForOp.getStep());
auto loopChunkIV = loopChunk.getInductionVar();
auto srcIV = srcForOp.getInductionVar();
BlockAndValueMapping operandMap;
auto bodyBuilder = OpBuilder::atBlockTerminator(loopChunk.getBody());
for (auto it = opGroupQueue.begin() + offset, e = opGroupQueue.end(); it != e;
++it) {
uint64_t shift = it->first;
auto ops = it->second;
// All 'same shift' operations get added with their operands being
// remapped to results of cloned operations, and their IV used remapped.
// Generate the remapping if the shift is not zero: remappedIV = newIV -
// shift.
if (!srcIV.use_empty() && shift != 0) {
auto ivRemap = bodyBuilder.create<AffineApplyOp>(
srcForOp.getLoc(),
bodyBuilder.getSingleDimShiftAffineMap(
-static_cast<int64_t>(srcForOp.getStep() * shift)),
loopChunkIV);
operandMap.map(srcIV, ivRemap);
} else {
operandMap.map(srcIV, loopChunkIV);
}
for (auto *op : ops)
bodyBuilder.clone(*op, operandMap);
};
if (succeeded(promoteIfSingleIteration(loopChunk)))
return AffineForOp();
return loopChunk;
}
// The skewing of operations with respect to one another can be used for
// example to allow overlap of asynchronous operations (such as DMA
// communication) with computation, or just relative shifting of operations
// for better register reuse, locality or parallelism. As such, the shifts are
// typically expected to be at most of the order of the number of operations.
// This method should not be used as a substitute for loop distribution/fission.
// This method uses an algorithm// in time linear in the number of operations
// in the body of the for loop - (using the 'sweep line' paradigm). This method
// asserts preservation of SSA dominance. A check for that as well as that for
// memory-based dependence preservation check rests with the users of this
// method.
LogicalResult mlir::affineForOpBodySkew(AffineForOp forOp,
ArrayRef<uint64_t> shifts,
bool unrollPrologueEpilogue) {
assert(forOp.getBody()->getOperations().size() == shifts.size() &&
"too few/many shifts");
if (forOp.getBody()->begin() == std::prev(forOp.getBody()->end()))
return success();
// If the trip counts aren't constant, we would need versioning and
// conditional guards (or context information to prevent such versioning). The
// better way to pipeline for such loops is to first tile them and extract
// constant trip count "full tiles" before applying this.
auto mayBeConstTripCount = getConstantTripCount(forOp);
if (!mayBeConstTripCount.hasValue()) {
LLVM_DEBUG(forOp.emitRemark("non-constant trip count loop not handled"));
return success();
}
uint64_t tripCount = mayBeConstTripCount.getValue();
assert(isOpwiseShiftValid(forOp, shifts) &&
"shifts will lead to an invalid transformation\n");
int64_t step = forOp.getStep();
unsigned numChildOps = shifts.size();
// Do a linear time (counting) sort for the shifts.
uint64_t maxShift = *std::max_element(shifts.begin(), shifts.end());
if (maxShift >= numChildOps) {
// Large shifts are not the typical use case.
forOp.emitWarning("not shifting because shifts are unrealistically large");
return success();
}
// An array of operation groups sorted by shift amount; each group has all
// operations with the same shift in the order in which they appear in the
// body of the 'affine.for' op.
std::vector<std::vector<Operation *>> sortedOpGroups(maxShift + 1);
unsigned pos = 0;
for (auto &op : forOp.getBody()->without_terminator()) {
auto shift = shifts[pos++];
sortedOpGroups[shift].push_back(&op);
}
// Unless the shifts have a specific pattern (which actually would be the
// common use case), prologue and epilogue are not meaningfully defined.
// Nevertheless, if 'unrollPrologueEpilogue' is set, we will treat the first
// loop generated as the prologue and the last as epilogue and unroll these
// fully.
AffineForOp prologue, epilogue;
// Do a sweep over the sorted shifts while storing open groups in a
// vector, and generating loop portions as necessary during the sweep. A block
// of operations is paired with its shift.
std::vector<std::pair<uint64_t, ArrayRef<Operation *>>> opGroupQueue;
auto origLbMap = forOp.getLowerBoundMap();
uint64_t lbShift = 0;
OpBuilder b(forOp);
for (uint64_t d = 0, e = sortedOpGroups.size(); d < e; ++d) {
// If nothing is shifted by d, continue.
if (sortedOpGroups[d].empty())
continue;
if (!opGroupQueue.empty()) {
assert(d > 0 &&
"Queue expected to be empty when the first block is found");
// The interval for which the loop needs to be generated here is:
// [lbShift, min(lbShift + tripCount, d)) and the body of the
// loop needs to have all operations in opQueue in that order.
AffineForOp res;
if (lbShift + tripCount * step < d * step) {
res = generateShiftedLoop(
b.getShiftedAffineMap(origLbMap, lbShift),
b.getShiftedAffineMap(origLbMap, lbShift + tripCount * step),
opGroupQueue, /*offset=*/0, forOp, b);
// Entire loop for the queued op groups generated, empty it.
opGroupQueue.clear();
lbShift += tripCount * step;
} else {
res = generateShiftedLoop(b.getShiftedAffineMap(origLbMap, lbShift),
b.getShiftedAffineMap(origLbMap, d),
opGroupQueue, /*offset=*/0, forOp, b);
lbShift = d * step;
}
if (res) {
// Simplify/canonicalize the affine.for.
RewritePatternSet patterns(res.getContext());
AffineForOp::getCanonicalizationPatterns(patterns, res.getContext());
bool erased;
(void)applyOpPatternsAndFold(res, std::move(patterns), &erased);
if (!erased && !prologue)
prologue = res;
if (!erased)
epilogue = res;
}
} else {
// Start of first interval.
lbShift = d * step;
}
// Augment the list of operations that get into the current open interval.
opGroupQueue.push_back({d, sortedOpGroups[d]});
}
// Those operations groups left in the queue now need to be processed (FIFO)
// and their loops completed.
for (unsigned i = 0, e = opGroupQueue.size(); i < e; ++i) {
uint64_t ubShift = (opGroupQueue[i].first + tripCount) * step;
epilogue = generateShiftedLoop(b.getShiftedAffineMap(origLbMap, lbShift),
b.getShiftedAffineMap(origLbMap, ubShift),
opGroupQueue, /*offset=*/i, forOp, b);
lbShift = ubShift;
if (!prologue)
prologue = epilogue;
}
// Erase the original for op.
forOp.erase();
if (unrollPrologueEpilogue && prologue)
(void)loopUnrollFull(prologue);
if (unrollPrologueEpilogue && !epilogue && epilogue != prologue)
(void)loopUnrollFull(epilogue);
return success();
}
/// Checks the legality of tiling of a hyper-rectangular loop nest by simply
/// checking if there is a 'negative' dependence in the memrefs present in
/// the loop nest. If yes then tiling is invalid.
static bool
checkTilingLegalityImpl(MutableArrayRef<mlir::AffineForOp> origLoops) {
assert(!origLoops.empty() && "no original loops provided");
// We first find out all dependences we intend to check.
SmallVector<Operation *, 8> loadAndStoreOps;
origLoops[0]->walk([&](Operation *op) {
if (isa<AffineReadOpInterface, AffineWriteOpInterface>(op))
loadAndStoreOps.push_back(op);
});
unsigned numOps = loadAndStoreOps.size();
unsigned numLoops = origLoops.size();
FlatAffineValueConstraints dependenceConstraints;
for (unsigned d = 1; d <= numLoops + 1; ++d) {
for (unsigned i = 0; i < numOps; ++i) {
Operation *srcOp = loadAndStoreOps[i];
MemRefAccess srcAccess(srcOp);
for (unsigned j = 0; j < numOps; ++j) {
Operation *dstOp = loadAndStoreOps[j];
MemRefAccess dstAccess(dstOp);
SmallVector<DependenceComponent, 2> depComps;
dependenceConstraints.reset();
DependenceResult result = checkMemrefAccessDependence(
srcAccess, dstAccess, d, &dependenceConstraints, &depComps);
// Skip if there is no dependence in this case.
if (!hasDependence(result))
continue;
// Check whether there is any negative direction vector in the
// dependence components found above, which means that dependence is
// violated by the default hyper-rect tiling method.
LLVM_DEBUG(llvm::dbgs() << "Checking whether tiling legality violated "
"for dependence at depth: "
<< Twine(d) << " between:\n";);
LLVM_DEBUG(srcAccess.opInst->dump(););
LLVM_DEBUG(dstAccess.opInst->dump(););
for (unsigned k = 0, e = depComps.size(); k < e; k++) {
DependenceComponent depComp = depComps[k];
if (depComp.lb.hasValue() && depComp.ub.hasValue() &&
depComp.lb.getValue() < depComp.ub.getValue() &&
depComp.ub.getValue() < 0) {
LLVM_DEBUG(llvm::dbgs()
<< "Dependence component lb = "
<< Twine(depComp.lb.getValue())
<< " ub = " << Twine(depComp.ub.getValue())
<< " is negative at depth: " << Twine(d)
<< " and thus violates the legality rule.\n");
return false;
}
}
}
}
}
return true;
}
/// Checks whether hyper-rectangular loop tiling of the nest
/// represented by `origLoops` is valid. The validity condition is from Irigoin
/// and Triolet, which states that two tiles cannot depend on each other. We
/// simplify such condition to just checking whether there is any negative
/// dependence direction, since we have the prior knowledge that the tiling
/// results will be hyper-rectangles, which are scheduled in the
/// lexicographically increasing order on the vector of loop indices. This
/// function will return failure when any dependence component is negative along
/// any of `origLoops`.
LogicalResult
checkTilingLegality(MutableArrayRef<mlir::AffineForOp> origLoops) {
return success(checkTilingLegalityImpl(origLoops));
}
/// Check if the input data is valid and whether tiled code will be legal or
/// not.
template <typename t>
void performPreTilingChecks(MutableArrayRef<AffineForOp> input,
ArrayRef<t> tileSizes) {
// Check if the supplied for op's are all successively nested.
assert(!input.empty() && "no loops in input band");
assert(input.size() == tileSizes.size() && "Too few/many tile sizes");
assert(isPerfectlyNested(input) && "input loops not perfectly nested");
// Perform tiling legality test.
if (failed(checkTilingLegality(input)))
input[0].emitRemark("tiled code is illegal due to dependences");
}
/// Move the loop body of AffineForOp 'src' from 'src' into the specified
/// location in destination's body, ignoring the terminator.
static void moveLoopBodyImpl(AffineForOp src, AffineForOp dest,
Block::iterator loc) {
auto &ops = src.getBody()->getOperations();
dest.getBody()->getOperations().splice(loc, ops, ops.begin(),
std::prev(ops.end()));
}
/// Move the loop body of AffineForOp 'src' from 'src' to the start of dest
/// body.
void moveLoopBody(AffineForOp src, AffineForOp dest) {
moveLoopBodyImpl(src, dest, dest.getBody()->begin());
}
/// Constructs tiled loop nest, without setting the loop bounds and move the
/// body of the original loop nest to the tiled loop nest.
void constructTiledLoopNest(MutableArrayRef<AffineForOp> origLoops,
AffineForOp rootAffineForOp, unsigned width,
MutableArrayRef<AffineForOp> tiledLoops) {
Location loc = rootAffineForOp.getLoc();
// The outermost among the loops as we add more..
Operation *topLoop = rootAffineForOp.getOperation();
AffineForOp innermostPointLoop;
// Add intra-tile (or point) loops.
for (unsigned i = 0; i < width; i++) {
OpBuilder b(topLoop);
// Loop bounds will be set later.
AffineForOp pointLoop = b.create<AffineForOp>(loc, 0, 0);
pointLoop.getBody()->getOperations().splice(
pointLoop.getBody()->begin(), topLoop->getBlock()->getOperations(),
topLoop);
tiledLoops[2 * width - 1 - i] = pointLoop;
topLoop = pointLoop.getOperation();
if (i == 0)
innermostPointLoop = pointLoop;
}
// Add tile space loops;
for (unsigned i = width; i < 2 * width; i++) {
OpBuilder b(topLoop);
// Loop bounds will be set later.
AffineForOp tileSpaceLoop = b.create<AffineForOp>(loc, 0, 0);
tileSpaceLoop.getBody()->getOperations().splice(
tileSpaceLoop.getBody()->begin(), topLoop->getBlock()->getOperations(),
topLoop);
tiledLoops[2 * width - i - 1] = tileSpaceLoop;
topLoop = tileSpaceLoop.getOperation();
}
// Move the loop body of the original nest to the new one.
moveLoopBody(origLoops.back(), innermostPointLoop);
}
/// Checks whether a loop nest is hyper-rectangular or not.
LogicalResult checkIfHyperRectangular(MutableArrayRef<AffineForOp> input,
AffineForOp rootAffineForOp,
unsigned width) {
FlatAffineValueConstraints cst;
SmallVector<Operation *, 8> ops(input.begin(), input.end());
(void)getIndexSet(ops, &cst);
if (!cst.isHyperRectangular(0, width)) {
rootAffineForOp.emitError("tiled code generation unimplemented for the "
"non-hyperrectangular case");
return failure();
}
return success();
}
/// Set lower and upper bounds of intra-tile loops for parametric tiling.
// TODO: Handle non-constant lower bounds.
static void setIntraTileBoundsParametric(OpBuilder &b, AffineForOp origLoop,
AffineForOp newInterTileLoop,
AffineForOp newIntraTileLoop,
Value tileSize) {
// The lower bound for the intra-tile loop is represented by an affine map
// as (%i, %t0)->((%i - %origlb) * %t0 + %origlb). Similarly, the upper bound
// for the intra-tile loop is represented by an affine map as (%i, %t0)->((%i
// - %origlb) * %t0) + (%t0 * %origLoopStep) + %origlb), where %i is loop IV
// of the corresponding inter-tile loop, %t0 is the corresponding tiling
// parameter, %origlb is lower bound and %origLoopStep is the loop step of the
// corresponding inter-tile loop.
assert(origLoop.hasConstantLowerBound() &&
"expected input loops to have constant lower bound.");
// Get lower bound of original loop as an affine expression.
AffineExpr origLowerBoundExpr;
origLowerBoundExpr =
b.getAffineConstantExpr(origLoop.getConstantLowerBound());
// Add dim operands from original lower/upper bound.
SmallVector<Value, 4> lbOperands, ubOperands;
AffineBound lb = origLoop.getLowerBound();
AffineBound ub = origLoop.getUpperBound();
lbOperands.reserve(lb.getNumOperands() + 2);
ubOperands.reserve(ub.getNumOperands() + 2);
AffineMap origLbMap = lb.getMap();
AffineMap origUbMap = ub.getMap();
for (unsigned j = 0, e = origLbMap.getNumDims(); j < e; ++j)
lbOperands.push_back(lb.getOperand(j));
for (unsigned j = 0, e = origUbMap.getNumDims(); j < e; ++j)
ubOperands.push_back(ub.getOperand(j));
// Add a new dim operand in lb/ubOperands corresponding to the origLoop
// IV.
lbOperands.push_back(newInterTileLoop.getInductionVar());
ubOperands.push_back(newInterTileLoop.getInductionVar());
// Get loop IV as an affine expression for lower/upper bound. Size of
// lb/ubOperands is guaranteed to be atleast one.
AffineExpr lbLoopIvExpr = b.getAffineDimExpr(lbOperands.size() - 1);
AffineExpr ubLoopIvExpr = b.getAffineDimExpr(ubOperands.size() - 1);
// Add symbol operands from original lower/upper bound.
for (unsigned j = 0, e = origLbMap.getNumSymbols(); j < e; ++j)
lbOperands.push_back(lb.getOperand(origLbMap.getNumDims() + j));
for (unsigned j = 0, e = origUbMap.getNumSymbols(); j < e; ++j)
ubOperands.push_back(ub.getOperand(origUbMap.getNumDims() + j));
// Add a new symbol operand which is the tile size for this loop.
lbOperands.push_back(tileSize);
ubOperands.push_back(tileSize);
SmallVector<AffineExpr, 4> lbBoundExprs;
SmallVector<AffineExpr, 4> ubBoundExprs;
lbBoundExprs.reserve(origLbMap.getNumResults());
ubBoundExprs.reserve(origUbMap.getNumResults());
// Get tiling parameter as an affine expression for lb/ub.
AffineExpr lbTileParameter = b.getAffineSymbolExpr(origLbMap.getNumSymbols());
AffineExpr ubTileParameter = b.getAffineSymbolExpr(origUbMap.getNumSymbols());
// Insert lb as inter-tile ((loop IV - origlb) * tilingParameter) + origlb.
lbBoundExprs.push_back(
((lbLoopIvExpr - origLowerBoundExpr) * lbTileParameter) +
origLowerBoundExpr);
// Get the origLoopStep as an affine expression.
AffineExpr origLoopStep = b.getAffineConstantExpr(origLoop.getStep());
// Insert ub as inter-tile ((loop IV - origlb) * tilingParameter) +
// (tilingParameter * origLoopStep) + origlb.
ubBoundExprs.push_back(
((ubLoopIvExpr - origLowerBoundExpr) * ubTileParameter) +
(ubTileParameter * origLoopStep) + origLowerBoundExpr);
ubBoundExprs.append(origUbMap.getResults().begin(),
origUbMap.getResults().end());
AffineMap lbMap =
AffineMap::get(origLbMap.getNumDims() + 1, origLbMap.getNumSymbols() + 1,
lbBoundExprs, b.getContext());
newIntraTileLoop.setLowerBound(lbOperands, lbMap);
AffineMap ubMap =
AffineMap::get(origUbMap.getNumDims() + 1, origUbMap.getNumSymbols() + 1,
ubBoundExprs, b.getContext());
newIntraTileLoop.setUpperBound(ubOperands, ubMap);
// Original loop step must be preserved.
newIntraTileLoop.setStep(origLoop.getStep());
}
/// Set lower and upper bounds of inter-tile loops for parametric tiling.
// TODO: Handle non-constant lower bounds.
static void setInterTileBoundsParametric(OpBuilder &b, AffineForOp origLoop,
AffineForOp newLoop, Value tileSize) {
OperandRange newLbOperands = origLoop.getLowerBoundOperands();
// The lower bounds for inter-tile loops are same as the corresponding lower
// bounds of original loops.
newLoop.setLowerBound(newLbOperands, origLoop.getLowerBoundMap());
// The new upper bound map for inter-tile loops, assuming constant lower
// bounds, are now originalLowerBound + ceildiv((originalUpperBound -
// originalLowerBound), tiling parameter); where tiling parameter is the
// respective tile size for that loop. For e.g. if the original ubmap was
// ()->(1024), the new map will be
// ()[s0]->(ceildiv((1024 -lb) % s0)), where s0 is the tiling parameter.
// Therefore a new symbol operand is inserted in the map and the result
// expression is overwritten.
assert(origLoop.hasConstantLowerBound() &&
"expected input loops to have constant lower bound.");
// Get lower bound of original loop as an affine expression.
AffineExpr origLowerBoundExpr;
origLowerBoundExpr =
b.getAffineConstantExpr(origLoop.getConstantLowerBound());
// Add dim operands from original upper bound.
SmallVector<Value, 4> ubOperands;
AffineBound ub = origLoop.getUpperBound();
ubOperands.reserve(ub.getNumOperands() + 1);
AffineMap origUbMap = ub.getMap();
for (unsigned j = 0, e = origUbMap.getNumDims(); j < e; ++j)
ubOperands.push_back(ub.getOperand(j));
// Add symbol operands from original upper bound.
for (unsigned j = 0, e = origUbMap.getNumSymbols(); j < e; ++j)
ubOperands.push_back(ub.getOperand(origUbMap.getNumDims() + j));
// Add a new symbol operand which is the tile size for this loop.
ubOperands.push_back(tileSize);
// Get tiling parameter as an affine expression.
AffineExpr tileParameter = b.getAffineSymbolExpr(origUbMap.getNumSymbols());
SmallVector<AffineExpr, 4> boundExprs;
boundExprs.reserve(origUbMap.getNumResults());
int64_t origUpperBound;
AffineExpr origUpperBoundExpr;
// If upper bound for the original loop is constant, then the constant can
// be obtained as an affine expression straight away.
if (origLoop.hasConstantUpperBound()) {
origUpperBound = origLoop.getConstantUpperBound();
// Get original constant upper bound as an affine expression.
origUpperBoundExpr = b.getAffineConstantExpr(origUpperBound);
// Insert the bound as originalLowerBoundceildiv((originalUpperBound -
// originalLowerBound), tilingParameter).
boundExprs.push_back(
origLowerBoundExpr +
(origUpperBoundExpr - origLowerBoundExpr).ceilDiv(tileParameter));
} else {
// If upper bound for the original loop is not constant then two cases
// are possible, although there handeling is the same, 1.) The result of
// ubmap has only one result expression. For e.g.
// affine.for %i = 5 to %ub
//
// A symbol operand is added which represents the tiling parameter. The
// new loop bounds here will be like ()[s0, s1] -> ((s0 - 5) ceildiv s1 + 5)
// where 's0' is the original upper bound and 's1' is the tiling
// parameter. 2.) When ubMap has more than one result expression. For e.g.
// #map0 = affine_map<()[s0, s1] -> (s0, s1)
// affine.for %i = 5 to min #map0()[%s0, %s1]
//
// A symbol operand is added which represents the tiling parameter. The
// new loop bounds will be like ()[s0, s1, s2] -> ((s0 - 5) ceildiv s2 + 5,
// (s1 -5) ceildiv s2 + 5), where s2 is the tiling parameter.
// Insert the bounds as originalLowerBound + ceildiv((originalUpperBound -
// originalLowerBound), tilingParameter).
for (AffineExpr origUpperBoundExpr : origUbMap.getResults())
boundExprs.push_back(
origLowerBoundExpr +
(origUpperBoundExpr - origLowerBoundExpr).ceilDiv(tileParameter));
}
AffineMap ubMap =
AffineMap::get(origUbMap.getNumDims(), origUbMap.getNumSymbols() + 1,
boundExprs, b.getContext());
newLoop.setUpperBound(ubOperands, ubMap);
// Original loop step must be preserved.
newLoop.setStep(origLoop.getStep());
}
/// Constructs and sets new loop bounds after tiling for the case of
/// hyper-rectangular index sets, where the bounds of one dimension do not
/// depend on other dimensions and tiling parameters are captured from SSA
/// values. Bounds of each dimension can thus be treated independently,
/// and deriving the new bounds is much simpler and faster than for the case of
/// tiling arbitrary polyhedral shapes.
static void constructParametricallyTiledIndexSetHyperRect(
MutableArrayRef<AffineForOp> origLoops,
MutableArrayRef<AffineForOp> newLoops, ArrayRef<Value> tileSizes) {
assert(!origLoops.empty() && "expected atleast one loop in band");
assert(origLoops.size() == tileSizes.size() &&
"expected tiling parameter for each loop in band.");
OpBuilder b(origLoops[0].getOperation());
unsigned width = origLoops.size();
// Set bounds for tile space loops.
for (unsigned i = 0; i < width; ++i) {
setInterTileBoundsParametric(b, origLoops[i], newLoops[i], tileSizes[i]);
}
// Set bounds for intra-tile loops.
for (unsigned i = 0; i < width; ++i) {
setIntraTileBoundsParametric(b, origLoops[i], newLoops[i],
newLoops[i + width], tileSizes[i]);
}
}
/// Constructs and sets new loop bounds after tiling for the case of
/// hyper-rectangular index sets, where the bounds of one dimension do not
/// depend on other dimensions. Bounds of each dimension can thus be treated
/// independently, and deriving the new bounds is much simpler and faster
/// than for the case of tiling arbitrary polyhedral shapes.
static void
constructTiledIndexSetHyperRect(MutableArrayRef<AffineForOp> origLoops,
MutableArrayRef<AffineForOp> newLoops,
ArrayRef<unsigned> tileSizes) {
assert(!origLoops.empty());
assert(origLoops.size() == tileSizes.size());
OpBuilder b(origLoops[0].getOperation());
unsigned width = origLoops.size();
// Bounds for tile space loops.
for (unsigned i = 0; i < width; i++) {
OperandRange newLbOperands = origLoops[i].getLowerBoundOperands();
OperandRange newUbOperands = origLoops[i].getUpperBoundOperands();
newLoops[i].setLowerBound(newLbOperands, origLoops[i].getLowerBoundMap());
newLoops[i].setUpperBound(newUbOperands, origLoops[i].getUpperBoundMap());
// If the step size of original loop is x and tileSize is y then after
// tiling the tile space loops' step size becomes x*y.
newLoops[i].setStep(tileSizes[i] * origLoops[i].getStep());
}
// Bounds for intra-tile loops.
for (unsigned i = 0; i < width; i++) {
int64_t largestDiv = getLargestDivisorOfTripCount(origLoops[i]);
Optional<uint64_t> mayBeConstantCount = getConstantTripCount(origLoops[i]);
// The lower bound is just the tile-space loop.
AffineMap lbMap = b.getDimIdentityMap();
newLoops[width + i].setLowerBound(
/*operands=*/newLoops[i].getInductionVar(), lbMap);
// The step sizes of intra-tile loops is just the original loops' step size.
newLoops[width + i].setStep(origLoops[i].getStep());
// Set the upper bound.
if (mayBeConstantCount && mayBeConstantCount.getValue() < tileSizes[i]) {
// Trip count is less than the tile size: upper bound is lower bound +
// trip count * stepSize.
AffineMap ubMap = b.getSingleDimShiftAffineMap(
mayBeConstantCount.getValue() * origLoops[i].getStep());
newLoops[width + i].setUpperBound(
/*operands=*/newLoops[i].getInductionVar(), ubMap);
} else if (largestDiv % tileSizes[i] != 0) {
// Intra-tile loop ii goes from i to min(i + tileSize * stepSize, ub_i).
// Construct the upper bound map; the operands are the original operands
// with 'i' (tile-space loop) appended to it. The new upper bound map is
// the original one with an additional expression i + tileSize * stepSize
// appended.
// Add dim operands from original upper bound.
SmallVector<Value, 4> ubOperands;
AffineBound ub = origLoops[i].getUpperBound();
ubOperands.reserve(ub.getNumOperands() + 1);
AffineMap origUbMap = ub.getMap();
for (unsigned j = 0, e = origUbMap.getNumDims(); j < e; ++j)
ubOperands.push_back(ub.getOperand(j));
// Add dim operand for new loop upper bound.
ubOperands.push_back(newLoops[i].getInductionVar());
// Add symbol operands from original upper bound.
for (unsigned j = 0, e = origUbMap.getNumSymbols(); j < e; ++j)
ubOperands.push_back(ub.getOperand(origUbMap.getNumDims() + j));
SmallVector<AffineExpr, 4> boundExprs;
boundExprs.reserve(1 + origUbMap.getNumResults());
AffineExpr dim = b.getAffineDimExpr(origUbMap.getNumDims());
// The new upper bound map is the original one with an additional
// expression i + tileSize * stepSize (of original loop) appended.
boundExprs.push_back(dim + tileSizes[i] * origLoops[i].getStep());
boundExprs.append(origUbMap.getResults().begin(),
origUbMap.getResults().end());
AffineMap ubMap =
AffineMap::get(origUbMap.getNumDims() + 1, origUbMap.getNumSymbols(),
boundExprs, b.getContext());
newLoops[width + i].setUpperBound(/*operands=*/ubOperands, ubMap);
} else {
// No need of the min expression.
AffineExpr dim = b.getAffineDimExpr(0);
AffineMap ubMap =
AffineMap::get(1, 0, dim + tileSizes[i] * origLoops[i].getStep());
newLoops[width + i].setUpperBound(newLoops[i].getInductionVar(), ubMap);
}
}
}
/// Tiles the specified band of perfectly nested loops creating tile-space loops
/// and intra-tile loops. A band is a contiguous set of loops.
// TODO: handle non hyper-rectangular spaces.
LogicalResult
mlir::tilePerfectlyNested(MutableArrayRef<AffineForOp> input,
ArrayRef<unsigned> tileSizes,
SmallVectorImpl<AffineForOp> *tiledNest) {
performPreTilingChecks(input, tileSizes);
MutableArrayRef<AffineForOp> origLoops = input;
AffineForOp rootAffineForOp = origLoops[0];
// Note that width is at least one since band isn't empty.
unsigned width = input.size();
SmallVector<AffineForOp, 6> tiledLoops(2 * width);
// Construct a tiled loop nest without setting their bounds. Bounds are
// set later.
constructTiledLoopNest(origLoops, rootAffineForOp, width, tiledLoops);
SmallVector<Value, 8> origLoopIVs;
extractForInductionVars(input, &origLoopIVs);
if (failed(checkIfHyperRectangular(input, rootAffineForOp, width)))
return failure();
// Set loop bounds for the tiled loop nest.
constructTiledIndexSetHyperRect(origLoops, tiledLoops, tileSizes);
// Replace original IVs with intra-tile loop IVs.
for (unsigned i = 0; i < width; i++)
origLoopIVs[i].replaceAllUsesWith(tiledLoops[i + width].getInductionVar());
// Erase the old loop nest.
rootAffineForOp.erase();
if (tiledNest)
*tiledNest = std::move(tiledLoops);
return success();
}
/// Tiles the specified band of perfectly nested loops creating tile-space
/// loops and intra-tile loops, using SSA values as tiling parameters. A band
/// is a contiguous set of loops.
// TODO: handle non hyper-rectangular spaces.
LogicalResult
mlir::tilePerfectlyNestedParametric(MutableArrayRef<AffineForOp> input,
ArrayRef<Value> tileSizes,
SmallVectorImpl<AffineForOp> *tiledNest) {
performPreTilingChecks(input, tileSizes);
MutableArrayRef<AffineForOp> origLoops = input;
AffineForOp rootAffineForOp = origLoops[0];
// Note that width is at least one since band isn't empty.
unsigned width = input.size();
SmallVector<AffineForOp, 6> tiledLoops(2 * width);
// Construct a tiled loop nest without setting their bounds. Bounds are
// set later.
constructTiledLoopNest(origLoops, rootAffineForOp, width, tiledLoops);
SmallVector<Value, 8> origLoopIVs;
extractForInductionVars(input, &origLoopIVs);
if (failed(checkIfHyperRectangular(input, rootAffineForOp, width)))
return failure();
// Set loop bounds for the tiled loop nest.
constructParametricallyTiledIndexSetHyperRect(origLoops, tiledLoops,
tileSizes);
// Replace original IVs with intra-tile loop IVs.
for (unsigned i = 0; i < width; i++)
origLoopIVs[i].replaceAllUsesWith(tiledLoops[i + width].getInductionVar());
// Erase the old loop nest.
rootAffineForOp.erase();
if (tiledNest)
*tiledNest = std::move(tiledLoops);
return success();
}
/// Collect perfectly nested loops starting from `rootForOps`. Loops are
/// perfectly nested if each loop is the first and only non-terminator operation
/// in the parent loop. Collect at most `maxLoops` loops and append them to
/// `forOps`.
template <typename T>
static void getPerfectlyNestedLoopsImpl(
SmallVectorImpl<T> &forOps, T rootForOp,
unsigned maxLoops = std::numeric_limits<unsigned>::max()) {
for (unsigned i = 0; i < maxLoops; ++i) {
forOps.push_back(rootForOp);
Block &body = rootForOp.region().front();
if (body.begin() != std::prev(body.end(), 2))
return;
rootForOp = dyn_cast<T>(&body.front());
if (!rootForOp)
return;
}
}
/// Get perfectly nested sequence of loops starting at root of loop nest
/// (the first op being another AffineFor, and the second op - a terminator).
/// A loop is perfectly nested iff: the first op in the loop's body is another
/// AffineForOp, and the second op is a terminator).
void mlir::getPerfectlyNestedLoops(SmallVectorImpl<AffineForOp> &nestedLoops,
AffineForOp root) {
getPerfectlyNestedLoopsImpl(nestedLoops, root);
}
void mlir::getPerfectlyNestedLoops(SmallVectorImpl<scf::ForOp> &nestedLoops,
scf::ForOp root) {
getPerfectlyNestedLoopsImpl(nestedLoops, root);
}
/// Identify valid and profitable bands of loops to tile. This is currently just
/// a temporary placeholder to test the mechanics of tiled code generation.
/// Returns all maximal outermost perfect loop nests to tile.
void mlir::getTileableBands(FuncOp f,
std::vector<SmallVector<AffineForOp, 6>> *bands) {
// Get maximal perfect nest of 'affine.for' insts starting from root
// (inclusive).
for (AffineForOp forOp : f.getOps<AffineForOp>()) {
SmallVector<AffineForOp, 6> band;
getPerfectlyNestedLoops(band, forOp);
bands->push_back(band);
}
}
/// Unrolls this loop completely.
LogicalResult mlir::loopUnrollFull(AffineForOp forOp) {
Optional<uint64_t> mayBeConstantTripCount = getConstantTripCount(forOp);
if (mayBeConstantTripCount.hasValue()) {
uint64_t tripCount = mayBeConstantTripCount.getValue();
if (tripCount == 0)
return success();
if (tripCount == 1)
return promoteIfSingleIteration(forOp);
return loopUnrollByFactor(forOp, tripCount);
}
return failure();
}
/// Unrolls this loop by the specified factor or by the trip count (if constant)
/// whichever is lower.
LogicalResult mlir::loopUnrollUpToFactor(AffineForOp forOp,
uint64_t unrollFactor) {
Optional<uint64_t> mayBeConstantTripCount = getConstantTripCount(forOp);
if (mayBeConstantTripCount.hasValue() &&
mayBeConstantTripCount.getValue() < unrollFactor)
return loopUnrollByFactor(forOp, mayBeConstantTripCount.getValue());
return loopUnrollByFactor(forOp, unrollFactor);
}
/// Generates unrolled copies of AffineForOp or scf::ForOp 'loopBodyBlock', with
/// associated 'forOpIV' by 'unrollFactor', calling 'ivRemapFn' to remap
/// 'forOpIV' for each unrolled body. If specified, annotates the Ops in each
/// unrolled iteration using annotateFn.
static void generateUnrolledLoop(
Block *loopBodyBlock, Value forOpIV, uint64_t unrollFactor,
function_ref<Value(unsigned, Value, OpBuilder)> ivRemapFn,
function_ref<void(unsigned, Operation *, OpBuilder)> annotateFn,
ValueRange iterArgs, ValueRange yieldedValues) {
// Builder to insert unrolled bodies just before the terminator of the body of
// 'forOp'.
auto builder = OpBuilder::atBlockTerminator(loopBodyBlock);
if (!annotateFn)
annotateFn = [](unsigned, Operation *, OpBuilder) {};
// Keep a pointer to the last non-terminator operation in the original block
// so that we know what to clone (since we are doing this in-place).
Block::iterator srcBlockEnd = std::prev(loopBodyBlock->end(), 2);
// Unroll the contents of 'forOp' (append unrollFactor - 1 additional copies).
SmallVector<Value, 4> lastYielded(yieldedValues);
for (unsigned i = 1; i < unrollFactor; i++) {
BlockAndValueMapping operandMap;
// Prepare operand map.
operandMap.map(iterArgs, lastYielded);
// If the induction variable is used, create a remapping to the value for
// this unrolled instance.
if (!forOpIV.use_empty()) {
Value ivUnroll = ivRemapFn(i, forOpIV, builder);
operandMap.map(forOpIV, ivUnroll);
}
// Clone the original body of 'forOp'.
for (auto it = loopBodyBlock->begin(); it != std::next(srcBlockEnd); it++) {
Operation *clonedOp = builder.clone(*it, operandMap);
annotateFn(i, clonedOp, builder);
}
// Update yielded values.
for (unsigned i = 0, e = lastYielded.size(); i < e; i++)
lastYielded[i] = operandMap.lookup(yieldedValues[i]);
}
// Make sure we annotate the Ops in the original body. We do this last so that
// any annotations are not copied into the cloned Ops above.
for (auto it = loopBodyBlock->begin(); it != std::next(srcBlockEnd); it++)
annotateFn(0, &*it, builder);
// Update operands of the yield statement.
loopBodyBlock->getTerminator()->setOperands(lastYielded);
}
/// Helper to generate cleanup loop for unroll or unroll-and-jam when the trip
/// count is not a multiple of `unrollFactor`.
static void generateCleanupLoopForUnroll(AffineForOp forOp,
uint64_t unrollFactor) {
// Insert the cleanup loop right after 'forOp'.
OpBuilder builder(forOp->getBlock(), std::next(Block::iterator(forOp)));
auto cleanupForOp = cast<AffineForOp>(builder.clone(*forOp));
// Update uses of `forOp` results. `cleanupForOp` should use `forOp` result
// and produce results for the original users of `forOp` results.
auto results = forOp.getResults();
auto cleanupResults = cleanupForOp.getResults();
auto cleanupIterOperands = cleanupForOp.getIterOperands();
for (auto e : llvm::zip(results, cleanupResults, cleanupIterOperands)) {
std::get<0>(e).replaceAllUsesWith(std::get<1>(e));
cleanupForOp->replaceUsesOfWith(std::get<2>(e), std::get<0>(e));
}
AffineMap cleanupMap;
SmallVector<Value, 4> cleanupOperands;
getCleanupLoopLowerBound(forOp, unrollFactor, cleanupMap, cleanupOperands);
assert(cleanupMap &&
"cleanup loop lower bound map for single result lower bound maps "
"can always be determined");
cleanupForOp.setLowerBound(cleanupOperands, cleanupMap);
// Promote the loop body up if this has turned into a single iteration loop.
(void)promoteIfSingleIteration(cleanupForOp);
// Adjust upper bound of the original loop; this is the same as the lower
// bound of the cleanup loop.
forOp.setUpperBound(cleanupOperands, cleanupMap);
}
/// Unrolls this loop by the specified factor. Returns success if the loop
/// is successfully unrolled.
LogicalResult mlir::loopUnrollByFactor(
AffineForOp forOp, uint64_t unrollFactor,
function_ref<void(unsigned, Operation *, OpBuilder)> annotateFn) {
assert(unrollFactor > 0 && "unroll factor should be positive");
if (unrollFactor == 1)
return promoteIfSingleIteration(forOp);
// Nothing in the loop body other than the terminator.
if (llvm::hasSingleElement(forOp.getBody()->getOperations()))
return success();
// Loops where the lower bound is a max expression isn't supported for
// unrolling since the trip count can be expressed as an affine function when
// both the lower bound and the upper bound are multi-result maps. However,
// one meaningful way to do such unrolling would be to specialize the loop for
// the 'hotspot' case and unroll that hotspot.
if (forOp.getLowerBoundMap().getNumResults() != 1)
return failure();
// If the trip count is lower than the unroll factor, no unrolled body.
// TODO: option to specify cleanup loop unrolling.
Optional<uint64_t> mayBeConstantTripCount = getConstantTripCount(forOp);
if (mayBeConstantTripCount.hasValue() &&
mayBeConstantTripCount.getValue() < unrollFactor)
return failure();
// Generate the cleanup loop if trip count isn't a multiple of unrollFactor.
if (getLargestDivisorOfTripCount(forOp) % unrollFactor != 0)
generateCleanupLoopForUnroll(forOp, unrollFactor);
ValueRange iterArgs(forOp.getRegionIterArgs());
auto yieldedValues = forOp.getBody()->getTerminator()->getOperands();
// Scale the step of loop being unrolled by unroll factor.
int64_t step = forOp.getStep();
forOp.setStep(step * unrollFactor);
generateUnrolledLoop(
forOp.getBody(), forOp.getInductionVar(), unrollFactor,
[&](unsigned i, Value iv, OpBuilder b) {
// iv' = iv + i * step
auto d0 = b.getAffineDimExpr(0);
auto bumpMap = AffineMap::get(1, 0, d0 + i * step);
return b.create<AffineApplyOp>(forOp.getLoc(), bumpMap, iv);
},
/*annotateFn=*/annotateFn,
/*iterArgs=*/iterArgs, /*yieldedValues=*/yieldedValues);
// Promote the loop body up if this has turned into a single iteration loop.
(void)promoteIfSingleIteration(forOp);
return success();
}
/// Unrolls 'forOp' by 'unrollFactor', returns success if the loop is unrolled.
LogicalResult mlir::loopUnrollByFactor(
scf::ForOp forOp, uint64_t unrollFactor,
function_ref<void(unsigned, Operation *, OpBuilder)> annotateFn) {
assert(unrollFactor > 0 && "expected positive unroll factor");
if (unrollFactor == 1)
return promoteIfSingleIteration(forOp);
// Return if the loop body is empty.
if (llvm::hasSingleElement(forOp.getBody()->getOperations()))
return success();
// Compute tripCount = ceilDiv((upperBound - lowerBound), step) and populate
// 'upperBoundUnrolled' and 'stepUnrolled' for static and dynamic cases.
OpBuilder boundsBuilder(forOp);
auto loc = forOp.getLoc();
auto step = forOp.step();
Value upperBoundUnrolled;
Value stepUnrolled;
bool generateEpilogueLoop = true;
auto lbCstOp = forOp.lowerBound().getDefiningOp<arith::ConstantIndexOp>();
auto ubCstOp = forOp.upperBound().getDefiningOp<arith::ConstantIndexOp>();
auto stepCstOp = forOp.step().getDefiningOp<arith::ConstantIndexOp>();
if (lbCstOp && ubCstOp && stepCstOp) {
// Constant loop bounds computation.
int64_t lbCst = lbCstOp.value();
int64_t ubCst = ubCstOp.value();
int64_t stepCst = stepCstOp.value();
assert(lbCst >= 0 && ubCst >= 0 && stepCst >= 0 &&
"expected positive loop bounds and step");
int64_t tripCount = mlir::ceilDiv(ubCst - lbCst, stepCst);
int64_t tripCountEvenMultiple = tripCount - (tripCount % unrollFactor);
int64_t upperBoundUnrolledCst = lbCst + tripCountEvenMultiple * stepCst;
assert(upperBoundUnrolledCst <= ubCst);
int64_t stepUnrolledCst = stepCst * unrollFactor;
// Create constant for 'upperBoundUnrolled' and set epilogue loop flag.
generateEpilogueLoop = upperBoundUnrolledCst < ubCst;
if (generateEpilogueLoop)
upperBoundUnrolled = boundsBuilder.create<arith::ConstantIndexOp>(
loc, upperBoundUnrolledCst);
else
upperBoundUnrolled = ubCstOp;
// Create constant for 'stepUnrolled'.
stepUnrolled = stepCst == stepUnrolledCst
? step
: boundsBuilder.create<arith::ConstantIndexOp>(
loc, stepUnrolledCst);
} else {
// Dynamic loop bounds computation.
// TODO: Add dynamic asserts for negative lb/ub/step, or
// consider using ceilDiv from AffineApplyExpander.
auto lowerBound = forOp.lowerBound();
auto upperBound = forOp.upperBound();
Value diff =
boundsBuilder.create<arith::SubIOp>(loc, upperBound, lowerBound);
Value tripCount = ceilDivPositive(boundsBuilder, loc, diff, step);
Value unrollFactorCst =
boundsBuilder.create<arith::ConstantIndexOp>(loc, unrollFactor);
Value tripCountRem =
boundsBuilder.create<arith::RemSIOp>(loc, tripCount, unrollFactorCst);
// Compute tripCountEvenMultiple = tripCount - (tripCount % unrollFactor)
Value tripCountEvenMultiple =
boundsBuilder.create<arith::SubIOp>(loc, tripCount, tripCountRem);
// Compute upperBoundUnrolled = lowerBound + tripCountEvenMultiple * step
upperBoundUnrolled = boundsBuilder.create<arith::AddIOp>(
loc, lowerBound,
boundsBuilder.create<arith::MulIOp>(loc, tripCountEvenMultiple, step));
// Scale 'step' by 'unrollFactor'.
stepUnrolled =
boundsBuilder.create<arith::MulIOp>(loc, step, unrollFactorCst);
}
// Create epilogue clean up loop starting at 'upperBoundUnrolled'.
if (generateEpilogueLoop) {
OpBuilder epilogueBuilder(forOp->getContext());
epilogueBuilder.setInsertionPoint(forOp->getBlock(),
std::next(Block::iterator(forOp)));
auto epilogueForOp = cast<scf::ForOp>(epilogueBuilder.clone(*forOp));
epilogueForOp.setLowerBound(upperBoundUnrolled);
// Update uses of loop results.
auto results = forOp.getResults();
auto epilogueResults = epilogueForOp.getResults();
auto epilogueIterOperands = epilogueForOp.getIterOperands();
for (auto e : llvm::zip(results, epilogueResults, epilogueIterOperands)) {
std::get<0>(e).replaceAllUsesWith(std::get<1>(e));
epilogueForOp->replaceUsesOfWith(std::get<2>(e), std::get<0>(e));
}
(void)promoteIfSingleIteration(epilogueForOp);
}
// Create unrolled loop.
forOp.setUpperBound(upperBoundUnrolled);
forOp.setStep(stepUnrolled);
auto iterArgs = ValueRange(forOp.getRegionIterArgs());
auto yieldedValues = forOp.getBody()->getTerminator()->getOperands();
generateUnrolledLoop(
forOp.getBody(), forOp.getInductionVar(), unrollFactor,
[&](unsigned i, Value iv, OpBuilder b) {
// iv' = iv + step * i;
auto stride = b.create<arith::MulIOp>(
loc, step, b.create<arith::ConstantIndexOp>(loc, i));
return b.create<arith::AddIOp>(loc, iv, stride);
},
annotateFn, iterArgs, yieldedValues);
// Promote the loop body up if this has turned into a single iteration loop.
(void)promoteIfSingleIteration(forOp);
return success();
}
LogicalResult mlir::loopUnrollJamUpToFactor(AffineForOp forOp,
uint64_t unrollJamFactor) {
Optional<uint64_t> mayBeConstantTripCount = getConstantTripCount(forOp);
if (mayBeConstantTripCount.hasValue() &&
mayBeConstantTripCount.getValue() < unrollJamFactor)
return loopUnrollJamByFactor(forOp, mayBeConstantTripCount.getValue());
return loopUnrollJamByFactor(forOp, unrollJamFactor);
}
/// Check if all control operands of all loops are defined outside of `forOp`
/// and return false if not.
static bool areInnerBoundsInvariant(AffineForOp forOp) {
auto walkResult = forOp.walk([&](AffineForOp aForOp) {
for (auto controlOperand : aForOp.getControlOperands()) {
if (!forOp.isDefinedOutsideOfLoop(controlOperand))
return WalkResult::interrupt();
}
return WalkResult::advance();
});
if (walkResult.wasInterrupted())
return false;
return true;
}
// Gathers all maximal sub-blocks of operations that do not themselves
// include a for op (a operation could have a descendant for op though
// in its tree). Ignore the block terminators.
struct JamBlockGatherer {
// Store iterators to the first and last op of each sub-block found.
std::vector<std::pair<Block::iterator, Block::iterator>> subBlocks;
// This is a linear time walk.
void walk(Operation *op) {
for (auto &region : op->getRegions())
for (auto &block : region)
walk(block);
}
void walk(Block &block) {
for (auto it = block.begin(), e = std::prev(block.end()); it != e;) {
auto subBlockStart = it;
while (it != e && !isa<AffineForOp>(&*it))
++it;
if (it != subBlockStart)
subBlocks.push_back({subBlockStart, std::prev(it)});
// Process all for ops that appear next.
while (it != e && isa<AffineForOp>(&*it))
walk(&*it++);
}
}
};
/// Unrolls and jams this loop by the specified factor.
LogicalResult mlir::loopUnrollJamByFactor(AffineForOp forOp,
uint64_t unrollJamFactor) {
assert(unrollJamFactor > 0 && "unroll jam factor should be positive");
if (unrollJamFactor == 1)
return promoteIfSingleIteration(forOp);
// Nothing in the loop body other than the terminator.
if (llvm::hasSingleElement(forOp.getBody()->getOperations()))
return success();
// Loops where both lower and upper bounds are multi-result maps won't be
// unrolled (since the trip can't be expressed as an affine function in
// general).
// TODO: this may not be common, but we could support the case
// where the lower bound is a multi-result map and the ub is a single result
// one.
if (forOp.getLowerBoundMap().getNumResults() != 1)
return failure();
Optional<uint64_t> mayBeConstantTripCount = getConstantTripCount(forOp);
// If the trip count is lower than the unroll jam factor, no unroll jam.
if (mayBeConstantTripCount.hasValue() &&
mayBeConstantTripCount.getValue() < unrollJamFactor) {
LLVM_DEBUG(llvm::dbgs() << "[failed] trip count < unroll-jam factor\n");
return failure();
}
// If any control operand of any inner loop of `forOp` is defined within
// `forOp`, no unroll jam.
if (!areInnerBoundsInvariant(forOp))
return failure();
// Gather all sub-blocks to jam upon the loop being unrolled.
JamBlockGatherer jbg;
jbg.walk(forOp);
auto &subBlocks = jbg.subBlocks;
// Collect loops with iter_args.
SmallVector<AffineForOp, 4> loopsWithIterArgs;
forOp.walk([&](AffineForOp aForOp) {
if (aForOp.getNumIterOperands() > 0)
loopsWithIterArgs.push_back(aForOp);
});
// Get supported reductions to be used for creating reduction ops at the end.
SmallVector<LoopReduction> reductions;
if (forOp.getNumIterOperands() > 0)
getSupportedReductions(forOp, reductions);
// Generate the cleanup loop if trip count isn't a multiple of
// unrollJamFactor.
if (getLargestDivisorOfTripCount(forOp) % unrollJamFactor != 0)
generateCleanupLoopForUnroll(forOp, unrollJamFactor);
// `operandMaps[i - 1]` carries old->new operand mapping for the ith unrolled
// iteration. There are (`unrollJamFactor` - 1) iterations.
SmallVector<BlockAndValueMapping, 4> operandMaps(unrollJamFactor - 1);
// For any loop with iter_args, replace it with a new loop that has
// `unrollJamFactor` copies of its iterOperands, iter_args and yield
// operands.
SmallVector<AffineForOp, 4> newLoopsWithIterArgs;
OpBuilder builder(forOp.getContext());
for (AffineForOp oldForOp : loopsWithIterArgs) {
SmallVector<Value, 4> dupIterOperands, dupIterArgs, dupYieldOperands;
ValueRange oldIterOperands = oldForOp.getIterOperands();
ValueRange oldIterArgs = oldForOp.getRegionIterArgs();
ValueRange oldYieldOperands =
cast<AffineYieldOp>(oldForOp.getBody()->getTerminator()).getOperands();
// Get additional iterOperands, iterArgs, and yield operands. We will
// fix iterOperands and yield operands after cloning of sub-blocks.
for (unsigned i = unrollJamFactor - 1; i >= 1; --i) {
dupIterOperands.append(oldIterOperands.begin(), oldIterOperands.end());
dupIterArgs.append(oldIterArgs.begin(), oldIterArgs.end());
dupYieldOperands.append(oldYieldOperands.begin(), oldYieldOperands.end());
}
// Create a new loop with additional iterOperands, iter_args and yield
// operands. This new loop will take the loop body of the original loop.
AffineForOp newForOp = mlir::replaceForOpWithNewYields(
builder, oldForOp, dupIterOperands, dupYieldOperands, dupIterArgs);
newLoopsWithIterArgs.push_back(newForOp);
// `forOp` has been replaced with a new loop.
if (oldForOp == forOp)
forOp = newForOp;
assert(oldForOp.use_empty() && "old for op should not have any user");
oldForOp.erase();
// Update `operandMaps` for `newForOp` iterArgs and results.
ValueRange newIterArgs = newForOp.getRegionIterArgs();
unsigned oldNumIterArgs = oldIterArgs.size();
ValueRange newResults = newForOp.getResults();
unsigned oldNumResults = newResults.size() / unrollJamFactor;
assert(oldNumIterArgs == oldNumResults &&
"oldNumIterArgs must be the same as oldNumResults");
for (unsigned i = unrollJamFactor - 1; i >= 1; --i) {
for (unsigned j = 0; j < oldNumIterArgs; ++j) {
// `newForOp` has `unrollJamFactor` - 1 new sets of iterArgs and
// results. Update `operandMaps[i - 1]` to map old iterArgs and results
// to those in the `i`th new set.
operandMaps[i - 1].map(newIterArgs[j],
newIterArgs[i * oldNumIterArgs + j]);
operandMaps[i - 1].map(newResults[j],
newResults[i * oldNumResults + j]);
}
}
}
// Scale the step of loop being unroll-jammed by the unroll-jam factor.
int64_t step = forOp.getStep();
forOp.setStep(step * unrollJamFactor);
auto forOpIV = forOp.getInductionVar();
// Unroll and jam (appends unrollJamFactor - 1 additional copies).
for (unsigned i = unrollJamFactor - 1; i >= 1; --i) {
for (auto &subBlock : subBlocks) {
// Builder to insert unroll-jammed bodies. Insert right at the end of
// sub-block.
OpBuilder builder(subBlock.first->getBlock(), std::next(subBlock.second));
// If the induction variable is used, create a remapping to the value for
// this unrolled instance.
if (!forOpIV.use_empty()) {
// iv' = iv + i * step, i = 1 to unrollJamFactor-1.
auto d0 = builder.getAffineDimExpr(0);
auto bumpMap = AffineMap::get(1, 0, d0 + i * step);
auto ivUnroll =
builder.create<AffineApplyOp>(forOp.getLoc(), bumpMap, forOpIV);
operandMaps[i - 1].map(forOpIV, ivUnroll);
}
// Clone the sub-block being unroll-jammed.
for (auto it = subBlock.first; it != std::next(subBlock.second); ++it)
builder.clone(*it, operandMaps[i - 1]);
}
// Fix iterOperands and yield op operands of newly created loops.
for (auto newForOp : newLoopsWithIterArgs) {
unsigned oldNumIterOperands =
newForOp.getNumIterOperands() / unrollJamFactor;
unsigned numControlOperands = newForOp.getNumControlOperands();
auto yieldOp = cast<AffineYieldOp>(newForOp.getBody()->getTerminator());
unsigned oldNumYieldOperands = yieldOp.getNumOperands() / unrollJamFactor;
assert(oldNumIterOperands == oldNumYieldOperands &&
"oldNumIterOperands must be the same as oldNumYieldOperands");
for (unsigned j = 0; j < oldNumIterOperands; ++j) {
// The `i`th duplication of an old iterOperand or yield op operand
// needs to be replaced with a mapped value from `operandMaps[i - 1]`
// if such mapped value exists.
newForOp.setOperand(numControlOperands + i * oldNumIterOperands + j,
operandMaps[i - 1].lookupOrDefault(
newForOp.getOperand(numControlOperands + j)));
yieldOp.setOperand(
i * oldNumYieldOperands + j,
operandMaps[i - 1].lookupOrDefault(yieldOp.getOperand(j)));
}
}
}
if (forOp.getNumResults() > 0) {
// Create reduction ops to combine every `unrollJamFactor` related results
// into one value. For example, for %0:2 = affine.for ... and addf, we add
// %1 = arith.addf %0#0, %0#1, and replace the following uses of %0#0 with
// %1.
builder.setInsertionPointAfter(forOp);
auto loc = forOp.getLoc();
unsigned oldNumResults = forOp.getNumResults() / unrollJamFactor;
for (LoopReduction &reduction : reductions) {
unsigned pos = reduction.iterArgPosition;
Value lhs = forOp.getResult(pos);
Value rhs;
SmallPtrSet<Operation *, 4> newOps;
for (unsigned i = unrollJamFactor - 1; i >= 1; --i) {
rhs = forOp.getResult(i * oldNumResults + pos);
// Create ops based on reduction type.
lhs = getReductionOp(reduction.kind, builder, loc, lhs, rhs);
if (!lhs)
return failure();
Operation *op = lhs.getDefiningOp();
assert(op && "Reduction op should have been created");
newOps.insert(op);
}
// Replace all uses except those in newly created reduction ops.
forOp.getResult(pos).replaceAllUsesExcept(lhs, newOps);
}
}
// Promote the loop body up if this has turned into a single iteration loop.
(void)promoteIfSingleIteration(forOp);
return success();
}
/// Performs loop interchange on 'forOpA' and 'forOpB', where 'forOpB' is
/// nested within 'forOpA' as the only non-terminator operation in its block.
void mlir::interchangeLoops(AffineForOp forOpA, AffineForOp forOpB) {
assert(&*forOpA.getBody()->begin() == forOpB.getOperation());
auto &forOpABody = forOpA.getBody()->getOperations();
auto &forOpBBody = forOpB.getBody()->getOperations();
// 1) Splice forOpA's non-terminator operations (which is just forOpB) just
// before forOpA (in ForOpA's parent's block) this should leave 'forOpA's
// body containing only the terminator.
forOpA->getBlock()->getOperations().splice(Block::iterator(forOpA),
forOpABody, forOpABody.begin(),
std::prev(forOpABody.end()));
// 2) Splice forOpB's non-terminator operations into the beginning of forOpA's
// body (this leaves forOpB's body containing only the terminator).
forOpABody.splice(forOpABody.begin(), forOpBBody, forOpBBody.begin(),
std::prev(forOpBBody.end()));
// 3) Splice forOpA into the beginning of forOpB's body.
forOpBBody.splice(forOpBBody.begin(), forOpA->getBlock()->getOperations(),
Block::iterator(forOpA));
}
// Checks each dependence component against the permutation to see if the
// desired loop interchange would violate dependences by making the
// dependence component lexicographically negative.
static bool checkLoopInterchangeDependences(
const std::vector<SmallVector<DependenceComponent, 2>> &depCompsVec,
ArrayRef<AffineForOp> loops, ArrayRef<unsigned> loopPermMap) {
// Invert permutation map.
unsigned maxLoopDepth = loops.size();
SmallVector<unsigned, 4> loopPermMapInv;
loopPermMapInv.resize(maxLoopDepth);
for (unsigned i = 0; i < maxLoopDepth; ++i)
loopPermMapInv[loopPermMap[i]] = i;
// Check each dependence component against the permutation to see if the
// desired loop interchange permutation would make the dependence vectors
// lexicographically negative.
// Example 1: [-1, 1][0, 0]
// Example 2: [0, 0][-1, 1]
for (unsigned i = 0, e = depCompsVec.size(); i < e; ++i) {
const SmallVector<DependenceComponent, 2> &depComps = depCompsVec[i];
assert(depComps.size() >= maxLoopDepth);
// Check if the first non-zero dependence component is positive.
// This iterates through loops in the desired order.
for (unsigned j = 0; j < maxLoopDepth; ++j) {
unsigned permIndex = loopPermMapInv[j];
assert(depComps[permIndex].lb.hasValue());
int64_t depCompLb = depComps[permIndex].lb.getValue();
if (depCompLb > 0)
break;
if (depCompLb < 0)
return false;
}
}
return true;
}
/// Checks if the loop interchange permutation 'loopPermMap' of the perfectly
/// nested sequence of loops in 'loops' would violate dependences.
bool mlir::isValidLoopInterchangePermutation(ArrayRef<AffineForOp> loops,
ArrayRef<unsigned> loopPermMap) {
// Gather dependence components for dependences between all ops in loop nest
// rooted at 'loops[0]', at loop depths in range [1, maxLoopDepth].
assert(loopPermMap.size() == loops.size());
unsigned maxLoopDepth = loops.size();
std::vector<SmallVector<DependenceComponent, 2>> depCompsVec;
getDependenceComponents(loops[0], maxLoopDepth, &depCompsVec);
return checkLoopInterchangeDependences(depCompsVec, loops, loopPermMap);
}
/// Returns true if `loops` is a perfectly nested loop nest, where loops appear
/// in it from outermost to innermost.
bool LLVM_ATTRIBUTE_UNUSED
mlir::isPerfectlyNested(ArrayRef<AffineForOp> loops) {
assert(!loops.empty() && "no loops provided");
// We already know that the block can't be empty.
auto hasTwoElements = [](Block *block) {
auto secondOpIt = std::next(block->begin());
return secondOpIt != block->end() && &*secondOpIt == &block->back();
};
auto enclosingLoop = loops.front();
for (auto loop : loops.drop_front()) {
auto parentForOp = dyn_cast<AffineForOp>(loop->getParentOp());
// parentForOp's body should be just this loop and the terminator.
if (parentForOp != enclosingLoop || !hasTwoElements(parentForOp.getBody()))
return false;
enclosingLoop = loop;
}
return true;
}
// input[i] should move from position i -> permMap[i]. Returns the position in
// `input` that becomes the new outermost loop.
unsigned mlir::permuteLoops(MutableArrayRef<AffineForOp> input,
ArrayRef<unsigned> permMap) {
assert(input.size() == permMap.size() && "invalid permutation map size");
// Check whether the permutation spec is valid. This is a small vector - we'll
// just sort and check if it's iota.
SmallVector<unsigned, 4> checkPermMap(permMap.begin(), permMap.end());
llvm::sort(checkPermMap);
if (llvm::any_of(llvm::enumerate(checkPermMap),
[](const auto &en) { return en.value() != en.index(); }))
assert(false && "invalid permutation map");
// Nothing to do.
if (input.size() < 2)
return 0;
assert(isPerfectlyNested(input) && "input not perfectly nested");
// Compute the inverse mapping, invPermMap: since input[i] goes to position
// permMap[i], position i of the permuted nest is at input[invPermMap[i]].
SmallVector<std::pair<unsigned, unsigned>, 4> invPermMap;
for (unsigned i = 0, e = input.size(); i < e; ++i)
invPermMap.push_back({permMap[i], i});
llvm::sort(invPermMap);
// Move the innermost loop body to the loop that would be the innermost in the
// permuted nest (only if the innermost loop is going to change).
if (permMap.back() != input.size() - 1) {
auto *destBody = input[invPermMap.back().second].getBody();
auto *srcBody = input.back().getBody();
destBody->getOperations().splice(destBody->begin(),
srcBody->getOperations(), srcBody->begin(),
std::prev(srcBody->end()));
}
// We'll move each loop in `input` in the reverse order so that its body is
// empty when we are moving it; this incurs zero copies and no erasing.
for (int i = input.size() - 1; i >= 0; --i) {
// If this has to become the outermost loop after permutation, add it to the
// parent block of the original root.
if (permMap[i] == 0) {
// If the root remains the same, nothing to do.
if (i == 0)
continue;
// Make input[i] the new outermost loop moving it into parentBlock.
auto *parentBlock = input[0]->getBlock();
parentBlock->getOperations().splice(Block::iterator(input[0]),
input[i]->getBlock()->getOperations(),
Block::iterator(input[i]));
continue;
}
// If the parent in the permuted order is the same as in the original,
// nothing to do.
unsigned parentPosInInput = invPermMap[permMap[i] - 1].second;
if (i > 0 && static_cast<unsigned>(i - 1) == parentPosInInput)
continue;
// Move input[i] to its surrounding loop in the transformed nest.
auto *destBody = input[parentPosInInput].getBody();
destBody->getOperations().splice(destBody->begin(),
input[i]->getBlock()->getOperations(),
Block::iterator(input[i]));
}
return invPermMap[0].second;
}
// Sinks all sequential loops to the innermost levels (while preserving
// relative order among them) and moves all parallel loops to the
// outermost (while again preserving relative order among them).
AffineForOp mlir::sinkSequentialLoops(AffineForOp forOp) {
SmallVector<AffineForOp, 4> loops;
getPerfectlyNestedLoops(loops, forOp);
if (loops.size() < 2)
return forOp;
// Gather dependence components for dependences between all ops in loop nest
// rooted at 'loops[0]', at loop depths in range [1, maxLoopDepth].
unsigned maxLoopDepth = loops.size();
std::vector<SmallVector<DependenceComponent, 2>> depCompsVec;
getDependenceComponents(loops[0], maxLoopDepth, &depCompsVec);
// Mark loops as either parallel or sequential.
SmallVector<bool, 8> isParallelLoop(maxLoopDepth, true);
for (unsigned i = 0, e = depCompsVec.size(); i < e; ++i) {
SmallVector<DependenceComponent, 2> &depComps = depCompsVec[i];
assert(depComps.size() >= maxLoopDepth);
for (unsigned j = 0; j < maxLoopDepth; ++j) {
DependenceComponent &depComp = depComps[j];
assert(depComp.lb.hasValue() && depComp.ub.hasValue());
if (depComp.lb.getValue() != 0 || depComp.ub.getValue() != 0)
isParallelLoop[j] = false;
}
}
// Count the number of parallel loops.
unsigned numParallelLoops = 0;
for (unsigned i = 0, e = isParallelLoop.size(); i < e; ++i)
if (isParallelLoop[i])
++numParallelLoops;
// Compute permutation of loops that sinks sequential loops (and thus raises
// parallel loops) while preserving relative order.
SmallVector<unsigned, 4> loopPermMap(maxLoopDepth);
unsigned nextSequentialLoop = numParallelLoops;
unsigned nextParallelLoop = 0;
for (unsigned i = 0; i < maxLoopDepth; ++i) {
if (isParallelLoop[i]) {
loopPermMap[i] = nextParallelLoop++;
} else {
loopPermMap[i] = nextSequentialLoop++;
}
}
// Check if permutation 'loopPermMap' would violate dependences.
if (!checkLoopInterchangeDependences(depCompsVec, loops, loopPermMap))
return forOp;
// Perform loop interchange according to permutation 'loopPermMap'.
unsigned loopNestRootIndex = permuteLoops(loops, loopPermMap);
return loops[loopNestRootIndex];
}
// Factors out common behavior to add a new `iv` (resp. `iv` + `offset`) to the
// lower (resp. upper) loop bound. When called for both the lower and upper
// bounds, the resulting IR resembles:
//
// ```mlir
// affine.for %i = max (`iv, ...) to min (`iv` + `offset`) {
// ...
// }
// ```
static void augmentMapAndBounds(OpBuilder &b, Value iv, AffineMap *map,
SmallVector<Value, 4> *operands,
int64_t offset = 0) {
auto bounds = llvm::to_vector<4>(map->getResults());
bounds.push_back(b.getAffineDimExpr(map->getNumDims()) + offset);
operands->insert(operands->begin() + map->getNumDims(), iv);
*map = AffineMap::get(map->getNumDims() + 1, map->getNumSymbols(), bounds,
b.getContext());
canonicalizeMapAndOperands(map, operands);
}
// Stripmines `forOp` by `factor` and sinks it under each of the `targets`.
// Stripmine-sink is a primitive building block for generalized tiling of
// imperfectly nested loops.
// This transformation is purely mechanical and does not check legality,
// profitability or even structural correctness. It is the user's
// responsibility to specify `targets` that are dominated by `forOp`.
// Returns the new AffineForOps, one per `targets`, nested immediately under
// each of the `targets`.
static SmallVector<AffineForOp, 8>
stripmineSink(AffineForOp forOp, uint64_t factor,
ArrayRef<AffineForOp> targets) {
auto originalStep = forOp.getStep();
auto scaledStep = originalStep * factor;
forOp.setStep(scaledStep);
OpBuilder b(forOp->getBlock(), std::next(Block::iterator(forOp)));
// Lower-bound map creation.
auto lbMap = forOp.getLowerBoundMap();
SmallVector<Value, 4> lbOperands(forOp.getLowerBoundOperands());
augmentMapAndBounds(b, forOp.getInductionVar(), &lbMap, &lbOperands);
// Upper-bound map creation.
auto ubMap = forOp.getUpperBoundMap();
SmallVector<Value, 4> ubOperands(forOp.getUpperBoundOperands());
augmentMapAndBounds(b, forOp.getInductionVar(), &ubMap, &ubOperands,
/*offset=*/scaledStep);
auto iv = forOp.getInductionVar();
SmallVector<AffineForOp, 8> innerLoops;
for (auto t : targets) {
// Insert newForOp before the terminator of `t`.
auto b = OpBuilder::atBlockTerminator(t.getBody());
auto newForOp = b.create<AffineForOp>(t.getLoc(), lbOperands, lbMap,
ubOperands, ubMap, originalStep);
auto begin = t.getBody()->begin();
// Skip terminator and `newForOp` which is just before the terminator.
auto nOps = t.getBody()->getOperations().size() - 2;
newForOp.getBody()->getOperations().splice(
newForOp.getBody()->getOperations().begin(),
t.getBody()->getOperations(), begin, std::next(begin, nOps));
replaceAllUsesInRegionWith(iv, newForOp.getInductionVar(),
newForOp.region());
innerLoops.push_back(newForOp);
}
return innerLoops;
}
static Loops stripmineSink(scf::ForOp forOp, Value factor,
ArrayRef<scf::ForOp> targets) {
auto originalStep = forOp.step();
auto iv = forOp.getInductionVar();
OpBuilder b(forOp);
forOp.setStep(b.create<arith::MulIOp>(forOp.getLoc(), originalStep, factor));
Loops innerLoops;
for (auto t : targets) {
// Save information for splicing ops out of t when done
auto begin = t.getBody()->begin();
auto nOps = t.getBody()->getOperations().size();
// Insert newForOp before the terminator of `t`.
auto b = OpBuilder::atBlockTerminator((t.getBody()));
Value stepped = b.create<arith::AddIOp>(t.getLoc(), iv, forOp.step());
Value less = b.create<arith::CmpIOp>(t.getLoc(), arith::CmpIPredicate::slt,
forOp.upperBound(), stepped);
Value ub =
b.create<SelectOp>(t.getLoc(), less, forOp.upperBound(), stepped);
// Splice [begin, begin + nOps - 1) into `newForOp` and replace uses.
auto newForOp = b.create<scf::ForOp>(t.getLoc(), iv, ub, originalStep);
newForOp.getBody()->getOperations().splice(
newForOp.getBody()->getOperations().begin(),
t.getBody()->getOperations(), begin, std::next(begin, nOps - 1));
replaceAllUsesInRegionWith(iv, newForOp.getInductionVar(),
newForOp.region());
innerLoops.push_back(newForOp);
}
return innerLoops;
}
// Stripmines a `forOp` by `factor` and sinks it under a single `target`.
// Returns the new AffineForOps, nested immediately under `target`.
template <typename ForType, typename SizeType>
static ForType stripmineSink(ForType forOp, SizeType factor, ForType target) {
// TODO: Use cheap structural assertions that targets are nested under
// forOp and that targets are not nested under each other when DominanceInfo
// exposes the capability. It seems overkill to construct a whole function
// dominance tree at this point.
auto res = stripmineSink(forOp, factor, ArrayRef<ForType>{target});
assert(res.size() == 1 && "Expected 1 inner forOp");
return res[0];
}
template <typename ForType, typename SizeType>
static SmallVector<SmallVector<ForType, 8>, 8>
tileImpl(ArrayRef<ForType> forOps, ArrayRef<SizeType> sizes,
ArrayRef<ForType> targets) {
SmallVector<SmallVector<ForType, 8>, 8> res;
SmallVector<ForType, 8> currentTargets(targets.begin(), targets.end());
for (auto it : llvm::zip(forOps, sizes)) {
auto step = stripmineSink(std::get<0>(it), std::get<1>(it), currentTargets);
res.push_back(step);
currentTargets = step;
}
return res;
}
SmallVector<SmallVector<AffineForOp, 8>, 8>
mlir::tile(ArrayRef<AffineForOp> forOps, ArrayRef<uint64_t> sizes,
ArrayRef<AffineForOp> targets) {
return tileImpl(forOps, sizes, targets);
}
SmallVector<Loops, 8> mlir::tile(ArrayRef<scf::ForOp> forOps,
ArrayRef<Value> sizes,
ArrayRef<scf::ForOp> targets) {
return tileImpl(forOps, sizes, targets);
}
template <typename ForType, typename SizeType>
static SmallVector<ForType, 8>
tileImpl(ArrayRef<ForType> forOps, ArrayRef<SizeType> sizes, ForType target) {
SmallVector<ForType, 8> res;
for (auto loops : tile(forOps, sizes, ArrayRef<ForType>{target})) {
assert(loops.size() == 1);
res.push_back(loops[0]);
}
return res;
}
SmallVector<AffineForOp, 8> mlir::tile(ArrayRef<AffineForOp> forOps,
ArrayRef<uint64_t> sizes,
AffineForOp target) {
return tileImpl(forOps, sizes, target);
}
Loops mlir::tile(ArrayRef<scf::ForOp> forOps, ArrayRef<Value> sizes,
scf::ForOp target) {
return tileImpl(forOps, sizes, target);
}
Loops mlir::tilePerfectlyNested(scf::ForOp rootForOp, ArrayRef<Value> sizes) {
// Collect perfectly nested loops. If more size values provided than nested
// loops available, truncate `sizes`.
SmallVector<scf::ForOp, 4> forOps;
forOps.reserve(sizes.size());
getPerfectlyNestedLoopsImpl(forOps, rootForOp, sizes.size());
if (forOps.size() < sizes.size())
sizes = sizes.take_front(forOps.size());
return ::tile(forOps, sizes, forOps.back());
}
// Hoist the ops within `outer` that appear before `inner`.
// Such ops include the ops that have been introduced by parametric tiling.
// Ops that come from triangular loops (i.e. that belong to the program slice
// rooted at `outer`) and ops that have side effects cannot be hoisted.
// Return failure when any op fails to hoist.
static LogicalResult hoistOpsBetween(scf::ForOp outer, scf::ForOp inner) {
SetVector<Operation *> forwardSlice;
getForwardSlice(
outer.getInductionVar(), &forwardSlice,
[&inner](Operation *op) { return op != inner.getOperation(); });
LogicalResult status = success();
SmallVector<Operation *, 8> toHoist;
for (auto &op : outer.getBody()->without_terminator()) {
// Stop when encountering the inner loop.
if (&op == inner.getOperation())
break;
// Skip over non-hoistable ops.
if (forwardSlice.count(&op) > 0) {
status = failure();
continue;
}
// Skip intermediate scf::ForOp, these are not considered a failure.
if (isa<scf::ForOp>(op))
continue;
// Skip other ops with regions.
if (op.getNumRegions() > 0) {
status = failure();
continue;
}
// Skip if op has side effects.
// TODO: loads to immutable memory regions are ok.
if (!MemoryEffectOpInterface::hasNoEffect(&op)) {
status = failure();
continue;
}
toHoist.push_back(&op);
}
auto *outerForOp = outer.getOperation();
for (auto *op : toHoist)
op->moveBefore(outerForOp);
return status;
}
// Traverse the interTile and intraTile loops and try to hoist ops such that
// bands of perfectly nested loops are isolated.
// Return failure if either perfect interTile or perfect intraTile bands cannot
// be formed.
static LogicalResult tryIsolateBands(const TileLoops &tileLoops) {
LogicalResult status = success();
const Loops &interTile = tileLoops.first;
const Loops &intraTile = tileLoops.second;
auto size = interTile.size();
assert(size == intraTile.size());
if (size <= 1)
return success();
for (unsigned s = 1; s < size; ++s)
status = succeeded(status) ? hoistOpsBetween(intraTile[0], intraTile[s])
: failure();
for (unsigned s = 1; s < size; ++s)
status = succeeded(status) ? hoistOpsBetween(interTile[0], interTile[s])
: failure();
return status;
}
TileLoops mlir::extractFixedOuterLoops(scf::ForOp rootForOp,
ArrayRef<int64_t> sizes) {
// Collect perfectly nested loops. If more size values provided than nested
// loops available, truncate `sizes`.
SmallVector<scf::ForOp, 4> forOps;
forOps.reserve(sizes.size());
getPerfectlyNestedLoopsImpl(forOps, rootForOp, sizes.size());
if (forOps.size() < sizes.size())
sizes = sizes.take_front(forOps.size());
// Compute the tile sizes such that i-th outer loop executes size[i]
// iterations. Given that the loop current executes
// numIterations = ceildiv((upperBound - lowerBound), step)
// iterations, we need to tile with size ceildiv(numIterations, size[i]).
SmallVector<Value, 4> tileSizes;
tileSizes.reserve(sizes.size());
for (unsigned i = 0, e = sizes.size(); i < e; ++i) {
assert(sizes[i] > 0 && "expected strictly positive size for strip-mining");
auto forOp = forOps[i];
OpBuilder builder(forOp);
auto loc = forOp.getLoc();
Value diff = builder.create<arith::SubIOp>(loc, forOp.upperBound(),
forOp.lowerBound());
Value numIterations = ceilDivPositive(builder, loc, diff, forOp.step());
Value iterationsPerBlock =
ceilDivPositive(builder, loc, numIterations, sizes[i]);
tileSizes.push_back(iterationsPerBlock);
}
// Call parametric tiling with the given sizes.
auto intraTile = tile(forOps, tileSizes, forOps.back());
TileLoops tileLoops = std::make_pair(forOps, intraTile);
// TODO: for now we just ignore the result of band isolation.
// In the future, mapping decisions may be impacted by the ability to
// isolate perfectly nested bands.
(void)tryIsolateBands(tileLoops);
return tileLoops;
}
/// Return the new lower bound, upper bound, and step in that order. Insert any
/// additional bounds calculations before the given builder and any additional
/// conversion back to the original loop induction value inside the given Block.
static LoopParams normalizeLoop(OpBuilder &boundsBuilder,
OpBuilder &insideLoopBuilder, Location loc,
Value lowerBound, Value upperBound, Value step,
Value inductionVar) {
// Check if the loop is already known to have a constant zero lower bound or
// a constant one step.
bool isZeroBased = false;
if (auto ubCst = lowerBound.getDefiningOp<arith::ConstantIndexOp>())
isZeroBased = ubCst.value() == 0;
bool isStepOne = false;
if (auto stepCst = step.getDefiningOp<arith::ConstantIndexOp>())
isStepOne = stepCst.value() == 1;
// Compute the number of iterations the loop executes: ceildiv(ub - lb, step)
// assuming the step is strictly positive. Update the bounds and the step
// of the loop to go from 0 to the number of iterations, if necessary.
// TODO: introduce support for negative steps or emit dynamic asserts
// on step positivity, whatever gets implemented first.
if (isZeroBased && isStepOne)
return {/*lowerBound=*/lowerBound, /*upperBound=*/upperBound,
/*step=*/step};
Value diff = boundsBuilder.create<arith::SubIOp>(loc, upperBound, lowerBound);
Value newUpperBound = ceilDivPositive(boundsBuilder, loc, diff, step);
Value newLowerBound =
isZeroBased ? lowerBound
: boundsBuilder.create<arith::ConstantIndexOp>(loc, 0);
Value newStep =
isStepOne ? step : boundsBuilder.create<arith::ConstantIndexOp>(loc, 1);
// Insert code computing the value of the original loop induction variable
// from the "normalized" one.
Value scaled =
isStepOne
? inductionVar
: insideLoopBuilder.create<arith::MulIOp>(loc, inductionVar, step);
Value shifted =
isZeroBased
? scaled
: insideLoopBuilder.create<arith::AddIOp>(loc, scaled, lowerBound);
SmallPtrSet<Operation *, 2> preserve{scaled.getDefiningOp(),
shifted.getDefiningOp()};
inductionVar.replaceAllUsesExcept(shifted, preserve);
return {/*lowerBound=*/newLowerBound, /*upperBound=*/newUpperBound,
/*step=*/newStep};
}
/// Transform a loop with a strictly positive step
/// for %i = %lb to %ub step %s
/// into a 0-based loop with step 1
/// for %ii = 0 to ceildiv(%ub - %lb, %s) step 1 {
/// %i = %ii * %s + %lb
/// Insert the induction variable remapping in the body of `inner`, which is
/// expected to be either `loop` or another loop perfectly nested under `loop`.
/// Insert the definition of new bounds immediate before `outer`, which is
/// expected to be either `loop` or its parent in the loop nest.
static void normalizeLoop(scf::ForOp loop, scf::ForOp outer, scf::ForOp inner) {
OpBuilder builder(outer);
OpBuilder innerBuilder = OpBuilder::atBlockBegin(inner.getBody());
auto loopPieces =
normalizeLoop(builder, innerBuilder, loop.getLoc(), loop.lowerBound(),
loop.upperBound(), loop.step(), loop.getInductionVar());
loop.setLowerBound(loopPieces.lowerBound);
loop.setUpperBound(loopPieces.upperBound);
loop.setStep(loopPieces.step);
}
void mlir::coalesceLoops(MutableArrayRef<scf::ForOp> loops) {
if (loops.size() < 2)
return;
scf::ForOp innermost = loops.back();
scf::ForOp outermost = loops.front();
// 1. Make sure all loops iterate from 0 to upperBound with step 1. This
// allows the following code to assume upperBound is the number of iterations.
for (auto loop : loops)
normalizeLoop(loop, outermost, innermost);
// 2. Emit code computing the upper bound of the coalesced loop as product
// of the number of iterations of all loops.
OpBuilder builder(outermost);
Location loc = outermost.getLoc();
Value upperBound = outermost.upperBound();
for (auto loop : loops.drop_front())
upperBound =
builder.create<arith::MulIOp>(loc, upperBound, loop.upperBound());
outermost.setUpperBound(upperBound);
builder.setInsertionPointToStart(outermost.getBody());
// 3. Remap induction variables. For each original loop, the value of the
// induction variable can be obtained by dividing the induction variable of
// the linearized loop by the total number of iterations of the loops nested
// in it modulo the number of iterations in this loop (remove the values
// related to the outer loops):
// iv_i = floordiv(iv_linear, product-of-loop-ranges-until-i) mod range_i.
// Compute these iteratively from the innermost loop by creating a "running
// quotient" of division by the range.
Value previous = outermost.getInductionVar();
for (unsigned i = 0, e = loops.size(); i < e; ++i) {
unsigned idx = loops.size() - i - 1;
if (i != 0)
previous = builder.create<arith::DivSIOp>(loc, previous,
loops[idx + 1].upperBound());
Value iv = (i == e - 1) ? previous
: builder.create<arith::RemSIOp>(
loc, previous, loops[idx].upperBound());
replaceAllUsesInRegionWith(loops[idx].getInductionVar(), iv,
loops.back().region());
}
// 4. Move the operations from the innermost just above the second-outermost
// loop, delete the extra terminator and the second-outermost loop.
scf::ForOp second = loops[1];
innermost.getBody()->back().erase();
outermost.getBody()->getOperations().splice(
Block::iterator(second.getOperation()),
innermost.getBody()->getOperations());
second.erase();
}
LogicalResult mlir::coalesceLoops(MutableArrayRef<AffineForOp> loops) {
if (loops.size() < 2)
return success();
AffineForOp innermost = loops.back();
AffineForOp outermost = loops.front();
AffineBound ub = outermost.getUpperBound();
AffineMap origUbMap = ub.getMap();
Location loc = outermost.getLoc();
OpBuilder builder(outermost);
for (AffineForOp loop : loops) {
// We only work on normalized loops.
if (loop.getStep() != 1 || !loop.hasConstantLowerBound() ||
loop.getConstantLowerBound() != 0)
return failure();
}
SmallVector<Value, 4> upperBoundSymbols;
SmallVector<Value, 4> ubOperands(ub.getOperands().begin(),
ub.getOperands().end());
// 1. Store the upper bound of the outermost loop in a variable.
Value prev;
if (!llvm::hasSingleElement(origUbMap.getResults()))
prev = builder.create<AffineMinOp>(loc, origUbMap, ubOperands);
else
prev = builder.create<AffineApplyOp>(loc, origUbMap, ubOperands);
upperBoundSymbols.push_back(prev);
// 2. Emit code computing the upper bound of the coalesced loop as product of
// the number of iterations of all loops.
for (AffineForOp loop : loops.drop_front()) {
ub = loop.getUpperBound();
origUbMap = ub.getMap();
ubOperands = ub.getOperands();
Value upperBound;
// If upper bound map has more than one result, take their minimum.
if (!llvm::hasSingleElement(origUbMap.getResults()))
upperBound = builder.create<AffineMinOp>(loc, origUbMap, ubOperands);
else
upperBound = builder.create<AffineApplyOp>(loc, origUbMap, ubOperands);
upperBoundSymbols.push_back(upperBound);
SmallVector<Value, 4> operands;
operands.push_back(prev);
operands.push_back(upperBound);
// Maintain running product of loop upper bounds.
prev = builder.create<AffineApplyOp>(
loc,
AffineMap::get(/*numDims=*/1,
/*numSymbols=*/1,
builder.getAffineDimExpr(0) *
builder.getAffineSymbolExpr(0)),
operands);
}
// Set upper bound of the coalesced loop.
AffineMap newUbMap = AffineMap::get(
/*numDims=*/0,
/*numSymbols=*/1, builder.getAffineSymbolExpr(0), builder.getContext());
outermost.setUpperBound(prev, newUbMap);
builder.setInsertionPointToStart(outermost.getBody());
// 3. Remap induction variables. For each original loop, the value of the
// induction variable can be obtained by dividing the induction variable of
// the linearized loop by the total number of iterations of the loops nested
// in it modulo the number of iterations in this loop (remove the values
// related to the outer loops):
// iv_i = floordiv(iv_linear, product-of-loop-ranges-until-i) mod range_i.
// Compute these iteratively from the innermost loop by creating a "running
// quotient" of division by the range.
Value previous = outermost.getInductionVar();
for (unsigned idx = loops.size(); idx > 0; --idx) {
if (idx != loops.size()) {
SmallVector<Value, 4> operands;
operands.push_back(previous);
operands.push_back(upperBoundSymbols[idx]);
previous = builder.create<AffineApplyOp>(
loc,
AffineMap::get(
/*numDims=*/1, /*numSymbols=*/1,
builder.getAffineDimExpr(0).floorDiv(
builder.getAffineSymbolExpr(0))),
operands);
}
// Modified value of the induction variables of the nested loops after
// coalescing.
Value inductionVariable;
if (idx == 1) {
inductionVariable = previous;
} else {
SmallVector<Value, 4> applyOperands;
applyOperands.push_back(previous);
applyOperands.push_back(upperBoundSymbols[idx - 1]);
inductionVariable = builder.create<AffineApplyOp>(
loc,
AffineMap::get(
/*numDims=*/1, /*numSymbols=*/1,
builder.getAffineDimExpr(0) % builder.getAffineSymbolExpr(0)),
applyOperands);
}
replaceAllUsesInRegionWith(loops[idx - 1].getInductionVar(),
inductionVariable, loops.back().region());
}
// 4. Move the operations from the innermost just above the second-outermost
// loop, delete the extra terminator and the second-outermost loop.
AffineForOp secondOutermostLoop = loops[1];
innermost.getBody()->back().erase();
outermost.getBody()->getOperations().splice(
Block::iterator(secondOutermostLoop.getOperation()),
innermost.getBody()->getOperations());
secondOutermostLoop.erase();
return success();
}
void mlir::collapseParallelLoops(
scf::ParallelOp loops, ArrayRef<std::vector<unsigned>> combinedDimensions) {
OpBuilder outsideBuilder(loops);
Location loc = loops.getLoc();
// Presort combined dimensions.
auto sortedDimensions = llvm::to_vector<3>(combinedDimensions);
for (auto &dims : sortedDimensions)
std::sort(dims.begin(), dims.end());
// Normalize ParallelOp's iteration pattern.
SmallVector<Value, 3> normalizedLowerBounds, normalizedSteps,
normalizedUpperBounds;
for (unsigned i = 0, e = loops.getNumLoops(); i < e; ++i) {
OpBuilder insideLoopBuilder = OpBuilder::atBlockBegin(loops.getBody());
auto resultBounds =
normalizeLoop(outsideBuilder, insideLoopBuilder, loc,
loops.lowerBound()[i], loops.upperBound()[i],
loops.step()[i], loops.getBody()->getArgument(i));
normalizedLowerBounds.push_back(resultBounds.lowerBound);
normalizedUpperBounds.push_back(resultBounds.upperBound);
normalizedSteps.push_back(resultBounds.step);
}
// Combine iteration spaces.
SmallVector<Value, 3> lowerBounds, upperBounds, steps;
auto cst0 = outsideBuilder.create<arith::ConstantIndexOp>(loc, 0);
auto cst1 = outsideBuilder.create<arith::ConstantIndexOp>(loc, 1);
for (unsigned i = 0, e = sortedDimensions.size(); i < e; ++i) {
Value newUpperBound = outsideBuilder.create<arith::ConstantIndexOp>(loc, 1);
for (auto idx : sortedDimensions[i]) {
newUpperBound = outsideBuilder.create<arith::MulIOp>(
loc, newUpperBound, normalizedUpperBounds[idx]);
}
lowerBounds.push_back(cst0);
steps.push_back(cst1);
upperBounds.push_back(newUpperBound);
}
// Create new ParallelLoop with conversions to the original induction values.
// The loop below uses divisions to get the relevant range of values in the
// new induction value that represent each range of the original induction
// value. The remainders then determine based on that range, which iteration
// of the original induction value this represents. This is a normalized value
// that is un-normalized already by the previous logic.
auto newPloop = outsideBuilder.create<scf::ParallelOp>(
loc, lowerBounds, upperBounds, steps,
[&](OpBuilder &insideBuilder, Location, ValueRange ploopIVs) {
for (unsigned i = 0, e = combinedDimensions.size(); i < e; ++i) {
Value previous = ploopIVs[i];
unsigned numberCombinedDimensions = combinedDimensions[i].size();
// Iterate over all except the last induction value.
for (unsigned j = numberCombinedDimensions - 1; j > 0; --j) {
unsigned idx = combinedDimensions[i][j];
// Determine the current induction value's current loop iteration
Value iv = insideBuilder.create<arith::RemSIOp>(
loc, previous, normalizedUpperBounds[idx]);
replaceAllUsesInRegionWith(loops.getBody()->getArgument(idx), iv,
loops.region());
// Remove the effect of the current induction value to prepare for
// the next value.
previous = insideBuilder.create<arith::DivSIOp>(
loc, previous, normalizedUpperBounds[idx]);
}
// The final induction value is just the remaining value.
unsigned idx = combinedDimensions[i][0];
replaceAllUsesInRegionWith(loops.getBody()->getArgument(idx),
previous, loops.region());
}
});
// Replace the old loop with the new loop.
loops.getBody()->back().erase();
newPloop.getBody()->getOperations().splice(
Block::iterator(newPloop.getBody()->back()),
loops.getBody()->getOperations());
loops.erase();
}
void mlir::mapLoopToProcessorIds(scf::ForOp forOp, ArrayRef<Value> processorId,
ArrayRef<Value> numProcessors) {
assert(processorId.size() == numProcessors.size());
if (processorId.empty())
return;
OpBuilder b(forOp);
Location loc(forOp.getLoc());
AffineExpr lhs, rhs;
bindSymbols(forOp.getContext(), lhs, rhs);
auto mulMap = AffineMap::get(0, 2, lhs * rhs);
auto addMap = AffineMap::get(0, 2, lhs + rhs);
Value linearIndex = processorId.front();
for (unsigned i = 1, e = processorId.size(); i < e; ++i) {
auto mulApplyOp = b.create<AffineApplyOp>(
loc, mulMap, ValueRange{linearIndex, numProcessors[i]});
linearIndex = b.create<AffineApplyOp>(
loc, addMap, ValueRange{mulApplyOp, processorId[i]});
}
auto mulApplyOp = b.create<AffineApplyOp>(
loc, mulMap, ValueRange{linearIndex, forOp.step()});
Value lb = b.create<AffineApplyOp>(
loc, addMap, ValueRange{mulApplyOp, forOp.lowerBound()});
forOp.setLowerBound(lb);
Value step = forOp.step();
for (auto numProcs : numProcessors)
step = b.create<AffineApplyOp>(loc, mulMap, ValueRange{numProcs, step});
forOp.setStep(step);
}
/// Given a memref region, determine the lowest depth at which transfers can be
/// placed for it, and return the corresponding block, start and end positions
/// in the block for placing incoming (read) and outgoing (write) copies
/// respectively. The lowest depth depends on whether the region being accessed
/// is hoistable with respect to one or more immediately surrounding loops.
static void
findHighestBlockForPlacement(const MemRefRegion &region, Block &block,
Block::iterator &begin, Block::iterator &end,
Block **copyPlacementBlock,
Block::iterator *copyInPlacementStart,
Block::iterator *copyOutPlacementStart) {
const auto *cst = region.getConstraints();
SmallVector<Value, 4> symbols;
cst->getValues(cst->getNumDimIds(), cst->getNumDimAndSymbolIds(), &symbols);
SmallVector<AffineForOp, 4> enclosingFors;
getLoopIVs(*block.begin(), &enclosingFors);
// Walk up loop parents till we find an IV on which this region is
// symbolic/variant.
auto it = enclosingFors.rbegin();
for (auto e = enclosingFors.rend(); it != e; ++it) {
// TODO: also need to be checking this for regions symbols that
// aren't loop IVs, whether we are within their resp. defs' dominance scope.
if (llvm::is_contained(symbols, it->getInductionVar()))
break;
}
if (it != enclosingFors.rbegin()) {
auto lastInvariantIV = *std::prev(it);
*copyInPlacementStart = Block::iterator(lastInvariantIV.getOperation());
*copyOutPlacementStart = std::next(*copyInPlacementStart);
*copyPlacementBlock = lastInvariantIV->getBlock();
} else {
*copyInPlacementStart = begin;
*copyOutPlacementStart = end;
*copyPlacementBlock = &block;
}
}
// Info comprising stride and number of elements transferred every stride.
struct StrideInfo {
int64_t stride;
int64_t numEltPerStride;
};
/// Returns striding information for a copy/transfer of this region with
/// potentially multiple striding levels from outermost to innermost. For an
/// n-dimensional region, there can be at most n-1 levels of striding
/// successively nested.
// TODO: make this work with non-identity layout maps.
static void getMultiLevelStrides(const MemRefRegion &region,
ArrayRef<int64_t> bufferShape,
SmallVectorImpl<StrideInfo> *strideInfos) {
if (bufferShape.size() <= 1)
return;
int64_t numEltPerStride = 1;
int64_t stride = 1;
for (int d = bufferShape.size() - 1; d >= 1; d--) {
int64_t dimSize = region.memref.getType().cast<MemRefType>().getDimSize(d);
stride *= dimSize;
numEltPerStride *= bufferShape[d];
// A stride is needed only if the region has a shorter extent than the
// memref along the dimension *and* has an extent greater than one along the
// next major dimension.
if (bufferShape[d] < dimSize && bufferShape[d - 1] > 1) {
strideInfos->push_back({stride, numEltPerStride});
}
}
}
/// Generates a point-wise copy from/to `memref' to/from `fastMemRef' and
/// returns the outermost AffineForOp of the copy loop nest. `lbMaps` and
/// `ubMaps` along with `lbOperands` and `ubOperands` hold the lower and upper
/// bound information for the copy loop nest. `fastBufOffsets` contain the
/// expressions to be subtracted out from the respective copy loop iterators in
/// order to index the fast buffer. If `copyOut' is true, generates a copy-out;
/// otherwise a copy-in. Builder `b` should be set to the point the copy nest is
/// inserted.
//
/// The copy-in nest is generated as follows as an example for a 2-d region:
/// for x = ...
/// for y = ...
/// fast_buf[x - offset_x][y - offset_y] = memref[x][y]
///
static AffineForOp
generatePointWiseCopy(Location loc, Value memref, Value fastMemRef,
ArrayRef<AffineMap> lbMaps, ArrayRef<Value> lbOperands,
ArrayRef<AffineMap> ubMaps, ArrayRef<Value> ubOperands,
ArrayRef<AffineExpr> fastBufOffsets, bool isCopyOut,
OpBuilder b) {
assert(llvm::all_of(lbMaps, [&](AffineMap lbMap) {
return lbMap.getNumInputs() == lbOperands.size();
}));
assert(llvm::all_of(ubMaps, [&](AffineMap ubMap) {
return ubMap.getNumInputs() == ubOperands.size();
}));
unsigned rank = memref.getType().cast<MemRefType>().getRank();
assert(lbMaps.size() == rank && "wrong number of lb maps");
assert(ubMaps.size() == rank && "wrong number of ub maps");
SmallVector<Value, 4> memIndices;
SmallVector<AffineExpr, 4> fastBufExprs;
SmallVector<Value, 4> fastBufMapOperands;
AffineForOp copyNestRoot;
SmallVector<AffineApplyOp, 4> mayBeDeadApplys;
for (unsigned d = 0; d < rank; ++d) {
auto forOp = createCanonicalizedAffineForOp(b, loc, lbOperands, lbMaps[d],
ubOperands, ubMaps[d]);
if (d == 0)
copyNestRoot = forOp;
b = OpBuilder::atBlockTerminator(forOp.getBody());
auto fastBufOffsetMap =
AffineMap::get(lbOperands.size(), 0, fastBufOffsets[d]);
auto offset = b.create<AffineApplyOp>(loc, fastBufOffsetMap, lbOperands);
// Construct the subscript for the fast memref being copied into/from:
// x - offset_x.
fastBufExprs.push_back(b.getAffineDimExpr(2 * d + 1) -
b.getAffineDimExpr(2 * d));
fastBufMapOperands.push_back(offset);
fastBufMapOperands.push_back(forOp.getInductionVar());
mayBeDeadApplys.push_back(offset);
// Subscript for the slow memref being copied.
memIndices.push_back(forOp.getInductionVar());
}
auto fastBufMap =
AffineMap::get(2 * rank, /*symbolCount=*/0, fastBufExprs, b.getContext());
fullyComposeAffineMapAndOperands(&fastBufMap, &fastBufMapOperands);
fastBufMap = simplifyAffineMap(fastBufMap);
canonicalizeMapAndOperands(&fastBufMap, &fastBufMapOperands);
// Drop any dead affine.applys.
for (auto applyOp : mayBeDeadApplys)
if (applyOp.use_empty())
applyOp.erase();
if (!isCopyOut) {
// Copy in.
auto load = b.create<AffineLoadOp>(loc, memref, memIndices);
b.create<AffineStoreOp>(loc, load, fastMemRef, fastBufMap,
fastBufMapOperands);
return copyNestRoot;
}
// Copy out.
auto load =
b.create<AffineLoadOp>(loc, fastMemRef, fastBufMap, fastBufMapOperands);
b.create<AffineStoreOp>(loc, load, memref, memIndices);
return copyNestRoot;
}
static InFlightDiagnostic LLVM_ATTRIBUTE_UNUSED
emitRemarkForBlock(Block &block) {
return block.getParentOp()->emitRemark();
}
/// Creates a buffer in the faster memory space for the specified memref region;
/// generates a copy from the lower memory space to this one, and replaces all
/// loads/stores in the block range [`begin', `end') of `block' to load/store
/// from that buffer. Returns failure if copies could not be generated due to
/// yet unimplemented cases. `copyInPlacementStart` and `copyOutPlacementStart`
/// in copyPlacementBlock specify the insertion points where the incoming copies
/// and outgoing copies, respectively, should be inserted (the insertion happens
/// right before the insertion point). Since `begin` can itself be invalidated
/// due to the memref rewriting done from this method, the output argument
/// `nBegin` is set to its replacement (set to `begin` if no invalidation
/// happens). Since outgoing copies could have been inserted at `end`, the
/// output argument `nEnd` is set to the new end. `sizeInBytes` is set to the
/// size of the fast buffer allocated.
static LogicalResult generateCopy(
const MemRefRegion &region, Block *block, Block::iterator begin,
Block::iterator end, Block *copyPlacementBlock,
Block::iterator copyInPlacementStart, Block::iterator copyOutPlacementStart,
AffineCopyOptions copyOptions, DenseMap<Value, Value> &fastBufferMap,
DenseSet<Operation *> &copyNests, uint64_t *sizeInBytes,
Block::iterator *nBegin, Block::iterator *nEnd) {
*nBegin = begin;
*nEnd = end;
FuncOp f = begin->getParentOfType<FuncOp>();
OpBuilder topBuilder(f.getBody());
Value zeroIndex = topBuilder.create<arith::ConstantIndexOp>(f.getLoc(), 0);
if (begin == end)
return success();
// Is the copy out point at the end of the block where we are doing
// explicit copying.
bool isCopyOutAtEndOfBlock = (end == copyOutPlacementStart);
// Copies for read regions are going to be inserted at 'begin'.
OpBuilder prologue(copyPlacementBlock, copyInPlacementStart);
// Copies for write regions are going to be inserted at 'end'.
OpBuilder epilogue(copyPlacementBlock, copyOutPlacementStart);
OpBuilder &b = region.isWrite() ? epilogue : prologue;
// Builder to create constants at the top level.
auto func = copyPlacementBlock->getParent()->getParentOfType<FuncOp>();
OpBuilder top(func.getBody());
auto loc = region.loc;
auto memref = region.memref;
auto memRefType = memref.getType().cast<MemRefType>();
if (!memRefType.getLayout().isIdentity()) {
LLVM_DEBUG(llvm::dbgs() << "Non-identity layout map not yet supported\n");
return failure();
}
// Indices to use for the copying.
// Indices for the original memref being copied from/to.
SmallVector<Value, 4> memIndices;
// Indices for the faster buffer being copied into/from.
SmallVector<Value, 4> bufIndices;
unsigned rank = memRefType.getRank();
SmallVector<int64_t, 4> fastBufferShape;
// Compute the extents of the buffer.
std::vector<SmallVector<int64_t, 4>> lbs;
SmallVector<int64_t, 8> lbDivisors;
lbs.reserve(rank);
Optional<int64_t> numElements = region.getConstantBoundingSizeAndShape(
&fastBufferShape, &lbs, &lbDivisors);
if (!numElements.hasValue()) {
LLVM_DEBUG(llvm::dbgs() << "Non-constant region size not supported\n");
return failure();
}
if (numElements.getValue() == 0) {
LLVM_DEBUG(llvm::dbgs() << "Nothing to copy\n");
*sizeInBytes = 0;
return success();
}
SmallVector<AffineMap, 4> lbMaps(rank), ubMaps(rank);
for (unsigned i = 0; i < rank; ++i)
region.getLowerAndUpperBound(i, lbMaps[i], ubMaps[i]);
const FlatAffineValueConstraints *cst = region.getConstraints();
// 'regionSymbols' hold values that this memory region is symbolic/parametric
// on; these typically include loop IVs surrounding the level at which the
// copy generation is being done or other valid symbols in MLIR.
SmallVector<Value, 8> regionSymbols;
cst->getValues(rank, cst->getNumIds(), &regionSymbols);
// Construct the index expressions for the fast memory buffer. The index
// expression for a particular dimension of the fast buffer is obtained by
// subtracting out the lower bound on the original memref's data region
// along the corresponding dimension.
// Index start offsets for faster memory buffer relative to the original.
SmallVector<AffineExpr, 4> fastBufOffsets;
fastBufOffsets.reserve(rank);
for (unsigned d = 0; d < rank; d++) {
assert(lbs[d].size() == cst->getNumCols() - rank && "incorrect bound size");
AffineExpr offset = top.getAffineConstantExpr(0);
for (unsigned j = 0, e = cst->getNumCols() - rank - 1; j < e; j++)
offset = offset + lbs[d][j] * top.getAffineDimExpr(j);
assert(lbDivisors[d] > 0);
offset =
(offset + lbs[d][cst->getNumCols() - 1 - rank]).floorDiv(lbDivisors[d]);
// Set copy start location for this dimension in the lower memory space
// memref.
if (auto caf = offset.dyn_cast<AffineConstantExpr>()) {
auto indexVal = caf.getValue();
if (indexVal == 0) {
memIndices.push_back(zeroIndex);
} else {
memIndices.push_back(
top.create<arith::ConstantIndexOp>(loc, indexVal).getResult());
}
} else {
// The coordinate for the start location is just the lower bound along the
// corresponding dimension on the memory region (stored in 'offset').
auto map = AffineMap::get(
cst->getNumDimIds() + cst->getNumSymbolIds() - rank, 0, offset);
memIndices.push_back(b.create<AffineApplyOp>(loc, map, regionSymbols));
}
// The fast buffer is copied into at location zero; addressing is relative.
bufIndices.push_back(zeroIndex);
// Record the offsets since they are needed to remap the memory accesses of
// the original memref further below.
fastBufOffsets.push_back(offset);
}
// The faster memory space buffer.
Value fastMemRef;
// Check if a buffer was already created.
bool existingBuf = fastBufferMap.count(memref) > 0;
if (!existingBuf) {
AffineMap fastBufferLayout = b.getMultiDimIdentityMap(rank);
auto fastMemRefType =
MemRefType::get(fastBufferShape, memRefType.getElementType(),
fastBufferLayout, copyOptions.fastMemorySpace);
// Create the fast memory space buffer just before the 'affine.for'
// operation.
fastMemRef =
prologue.create<memref::AllocOp>(loc, fastMemRefType).getResult();
// Record it.
fastBufferMap[memref] = fastMemRef;
// fastMemRefType is a constant shaped memref.
*sizeInBytes = getMemRefSizeInBytes(fastMemRefType).getValue();
LLVM_DEBUG(emitRemarkForBlock(*block)
<< "Creating fast buffer of type " << fastMemRefType
<< " and size " << llvm::divideCeil(*sizeInBytes, 1024)
<< " KiB\n");
} else {
// Reuse the one already created.
fastMemRef = fastBufferMap[memref];
*sizeInBytes = 0;
}
auto numElementsSSA =
top.create<arith::ConstantIndexOp>(loc, numElements.getValue());
Value dmaStride = nullptr;
Value numEltPerDmaStride = nullptr;
if (copyOptions.generateDma) {
SmallVector<StrideInfo, 4> dmaStrideInfos;
getMultiLevelStrides(region, fastBufferShape, &dmaStrideInfos);
// TODO: use all stride levels once DmaStartOp is extended for
// multi-level strides.
if (dmaStrideInfos.size() > 1) {
LLVM_DEBUG(llvm::dbgs() << "Only up to one level of stride supported\n");
return failure();
}
if (!dmaStrideInfos.empty()) {
dmaStride =
top.create<arith::ConstantIndexOp>(loc, dmaStrideInfos[0].stride);
numEltPerDmaStride = top.create<arith::ConstantIndexOp>(
loc, dmaStrideInfos[0].numEltPerStride);
}
}
// Record the last operation where we want the memref replacement to end. We
// later do the memref replacement only in [begin, postDomFilter] so
// that the original memref's used in the data movement code themselves don't
// get replaced.
auto postDomFilter = std::prev(end);
// Create fully composed affine maps for each memref.
auto memAffineMap = b.getMultiDimIdentityMap(memIndices.size());
fullyComposeAffineMapAndOperands(&memAffineMap, &memIndices);
auto bufAffineMap = b.getMultiDimIdentityMap(bufIndices.size());
fullyComposeAffineMapAndOperands(&bufAffineMap, &bufIndices);
if (!copyOptions.generateDma) {
// Point-wise copy generation.
auto copyNest =
generatePointWiseCopy(loc, memref, fastMemRef, lbMaps,
/*lbOperands=*/regionSymbols, ubMaps,
/*ubOperands=*/regionSymbols, fastBufOffsets,
/*isCopyOut=*/region.isWrite(), b);
// Record this so that we can skip it from yet another copy.
copyNests.insert(copyNest);
// Since new ops are being appended (for copy out's), adjust the end to
// mark end of block range being processed if necessary.
if (region.isWrite() && isCopyOutAtEndOfBlock)
*nEnd = Block::iterator(copyNest.getOperation());
} else {
// DMA generation.
// Create a tag (single element 1-d memref) for the DMA.
auto tagMemRefType = MemRefType::get({1}, top.getIntegerType(32), {},
copyOptions.tagMemorySpace);
auto tagMemRef = prologue.create<memref::AllocOp>(loc, tagMemRefType);
SmallVector<Value, 4> tagIndices({zeroIndex});
auto tagAffineMap = b.getMultiDimIdentityMap(tagIndices.size());
fullyComposeAffineMapAndOperands(&tagAffineMap, &tagIndices);
if (!region.isWrite()) {
// DMA non-blocking read from original buffer to fast buffer.
b.create<AffineDmaStartOp>(loc, memref, memAffineMap, memIndices,
fastMemRef, bufAffineMap, bufIndices,
tagMemRef, tagAffineMap, tagIndices,
numElementsSSA, dmaStride, numEltPerDmaStride);
} else {
// DMA non-blocking write from fast buffer to the original memref.
auto op = b.create<AffineDmaStartOp>(
loc, fastMemRef, bufAffineMap, bufIndices, memref, memAffineMap,
memIndices, tagMemRef, tagAffineMap, tagIndices, numElementsSSA,
dmaStride, numEltPerDmaStride);
// Since new ops may be appended at 'end' (for outgoing DMAs), adjust the
// end to mark end of block range being processed.
if (isCopyOutAtEndOfBlock)
*nEnd = Block::iterator(op.getOperation());
}
// Matching DMA wait to block on completion; tag always has a 0 index.
b.create<AffineDmaWaitOp>(loc, tagMemRef, tagAffineMap, zeroIndex,
numElementsSSA);
// Generate dealloc for the tag.
auto tagDeallocOp = epilogue.create<memref::DeallocOp>(loc, tagMemRef);
if (*nEnd == end && isCopyOutAtEndOfBlock)
// Since new ops are being appended (for outgoing DMAs), adjust the end to
// mark end of range of the original.
*nEnd = Block::iterator(tagDeallocOp.getOperation());
}
// Generate dealloc for the buffer.
if (!existingBuf) {
auto bufDeallocOp = epilogue.create<memref::DeallocOp>(loc, fastMemRef);
// When generating pointwise copies, `nEnd' has to be set to deallocOp on
// the fast buffer (since it marks the new end insertion point).
if (!copyOptions.generateDma && *nEnd == end && isCopyOutAtEndOfBlock)
*nEnd = Block::iterator(bufDeallocOp.getOperation());
}
// Replace all uses of the old memref with the faster one while remapping
// access indices (subtracting out lower bound offsets for each dimension).
// Ex: to replace load %A[%i, %j] with load %Abuf[%i - %iT, %j - %jT],
// index remap will be (%i, %j) -> (%i - %iT, %j - %jT),
// i.e., affine.apply (d0, d1, d2, d3) -> (d2-d0, d3-d1) (%iT, %jT, %i, %j),
// and (%iT, %jT) will be the 'extraOperands' for 'rep all memref uses with'.
// d2, d3 correspond to the original indices (%i, %j).
SmallVector<AffineExpr, 4> remapExprs;
remapExprs.reserve(rank);
for (unsigned i = 0; i < rank; i++) {
// The starting operands of indexRemap will be regionSymbols (the symbols on
// which the memref region is parametric); then those corresponding to
// the memref's original indices follow.
auto dimExpr = b.getAffineDimExpr(regionSymbols.size() + i);
remapExprs.push_back(dimExpr - fastBufOffsets[i]);
}
auto indexRemap = AffineMap::get(regionSymbols.size() + rank, 0, remapExprs,
b.getContext());
// Record the begin since it may be invalidated by memref replacement.
Block::iterator prevOfBegin;
bool isBeginAtStartOfBlock = (begin == block->begin());
if (!isBeginAtStartOfBlock)
prevOfBegin = std::prev(begin);
// *Only* those uses within the range [begin, end) of 'block' are replaced.
(void)replaceAllMemRefUsesWith(memref, fastMemRef,
/*extraIndices=*/{}, indexRemap,
/*extraOperands=*/regionSymbols,
/*symbolOperands=*/{},
/*domInstFilter=*/&*begin,
/*postDomInstFilter=*/&*postDomFilter);
*nBegin = isBeginAtStartOfBlock ? block->begin() : std::next(prevOfBegin);
return success();
}
/// Construct the memref region to just include the entire memref. Returns false
/// dynamic shaped memref's for now. `numParamLoopIVs` is the number of
/// enclosing loop IVs of `op` (starting from the outermost) that the region
/// is parametric on.
static bool getFullMemRefAsRegion(Operation *op, unsigned numParamLoopIVs,
MemRefRegion *region) {
unsigned rank;
if (auto loadOp = dyn_cast<AffineLoadOp>(op)) {
rank = loadOp.getMemRefType().getRank();
region->memref = loadOp.getMemRef();
region->setWrite(false);
} else if (auto storeOp = dyn_cast<AffineStoreOp>(op)) {
rank = storeOp.getMemRefType().getRank();
region->memref = storeOp.getMemRef();
region->setWrite(true);
} else {
assert(false && "expected load or store op");
return false;
}
auto memRefType = region->memref.getType().cast<MemRefType>();
if (!memRefType.hasStaticShape())
return false;
auto *regionCst = region->getConstraints();
// Just get the first numSymbols IVs, which the memref region is parametric
// on.
SmallVector<AffineForOp, 4> ivs;
getLoopIVs(*op, &ivs);
ivs.resize(numParamLoopIVs);
SmallVector<Value, 4> symbols;
extractForInductionVars(ivs, &symbols);
regionCst->reset(rank, numParamLoopIVs, 0);
regionCst->setValues(rank, rank + numParamLoopIVs, symbols);
// Memref dim sizes provide the bounds.
for (unsigned d = 0; d < rank; d++) {
auto dimSize = memRefType.getDimSize(d);
assert(dimSize > 0 && "filtered dynamic shapes above");
regionCst->addBound(FlatAffineConstraints::LB, d, 0);
regionCst->addBound(FlatAffineConstraints::UB, d, dimSize - 1);
}
return true;
}
/// Performs explicit copying for the contiguous sequence of operations in the
/// block iterator range [`begin', `end'), where `end' can't be past the
/// terminator of the block (since additional operations are potentially
/// inserted right before `end`. Returns the total size of fast memory space
/// buffers used. `copyOptions` provides various parameters, and the output
/// argument `copyNests` is the set of all copy nests inserted, each represented
/// by its root affine.for. Since we generate alloc's and dealloc's for all fast
/// buffers (before and after the range of operations resp. or at a hoisted
/// position), all of the fast memory capacity is assumed to be available for
/// processing this block range. When 'filterMemRef' is specified, copies are
/// only generated for the provided MemRef.
uint64_t mlir::affineDataCopyGenerate(Block::iterator begin,
Block::iterator end,
const AffineCopyOptions &copyOptions,
Optional<Value> filterMemRef,
DenseSet<Operation *> &copyNests) {
if (begin == end)
return 0;
assert(begin->getBlock() == std::prev(end)->getBlock() &&
"Inconsistent block begin/end args");
assert(end != end->getBlock()->end() && "end can't be the block terminator");
Block *block = begin->getBlock();
// Copies will be generated for this depth, i.e., symbolic in all loops
// surrounding the this block range.
unsigned copyDepth = getNestingDepth(&*begin);
LLVM_DEBUG(llvm::dbgs() << "Generating copies at depth " << copyDepth
<< "\n");
LLVM_DEBUG(llvm::dbgs() << "from begin: " << *begin << "\n");
LLVM_DEBUG(llvm::dbgs() << "to inclusive end: " << *std::prev(end) << "\n");
// List of memory regions to copy for. We need a map vector to have a
// guaranteed iteration order to write test cases. CHECK-DAG doesn't help here
// since the alloc's for example are identical except for the SSA id.
SmallMapVector<Value, std::unique_ptr<MemRefRegion>, 4> readRegions;
SmallMapVector<Value, std::unique_ptr<MemRefRegion>, 4> writeRegions;
// Map from original memref's to the fast buffers that their accesses are
// replaced with.
DenseMap<Value, Value> fastBufferMap;
// To check for errors when walking the block.
bool error = false;
// Walk this range of operations to gather all memory regions.
block->walk(begin, end, [&](Operation *opInst) {
// Gather regions to allocate to buffers in faster memory space.
if (auto loadOp = dyn_cast<AffineLoadOp>(opInst)) {
if ((filterMemRef.hasValue() && filterMemRef != loadOp.getMemRef()) ||
(loadOp.getMemRefType().getMemorySpaceAsInt() !=
copyOptions.slowMemorySpace))
return;
} else if (auto storeOp = dyn_cast<AffineStoreOp>(opInst)) {
if ((filterMemRef.hasValue() && filterMemRef != storeOp.getMemRef()) ||
storeOp.getMemRefType().getMemorySpaceAsInt() !=
copyOptions.slowMemorySpace)
return;
} else {
// Neither load nor a store op.
return;
}
// Compute the MemRefRegion accessed.
auto region = std::make_unique<MemRefRegion>(opInst->getLoc());
if (failed(region->compute(opInst, copyDepth, /*sliceState=*/nullptr,
/*addMemRefDimBounds=*/false))) {
LLVM_DEBUG(llvm::dbgs()
<< "Error obtaining memory region: semi-affine maps?\n");
LLVM_DEBUG(llvm::dbgs() << "over-approximating to the entire memref\n");
if (!getFullMemRefAsRegion(opInst, copyDepth, region.get())) {
LLVM_DEBUG(
opInst->emitError("non-constant memref sizes not yet supported"));
error = true;
return;
}
}
// Each memref has a single buffer associated with it irrespective of how
// many load's and store's happen on it.
// TODO: in the future, when regions don't intersect and satisfy
// other properties (based on load/store regions), we could consider
// multiple buffers per memref.
// Add to the appropriate region if it's not already in it, or take a
// bounding box union with the existing one if it's already in there.
// Note that a memref may have both read and write regions - so update the
// region in the other list if one exists (write in case of read and vice
// versa) since there is a single bounding box for a memref across all reads
// and writes that happen on it.
// Attempts to update; returns true if 'region' exists in targetRegions.
auto updateRegion =
[&](const SmallMapVector<Value, std::unique_ptr<MemRefRegion>, 4>
&targetRegions) {
const auto it = targetRegions.find(region->memref);
if (it == targetRegions.end())
return false;
// Perform a union with the existing region.
if (failed(it->second->unionBoundingBox(*region))) {
LLVM_DEBUG(llvm::dbgs()
<< "Memory region bounding box failed; "
"over-approximating to the entire memref\n");
// If the union fails, we will overapproximate.
if (!getFullMemRefAsRegion(opInst, copyDepth, region.get())) {
LLVM_DEBUG(opInst->emitError(
"non-constant memref sizes not yet supported"));
error = true;
return true;
}
it->second->getConstraints()->clearAndCopyFrom(
*region->getConstraints());
} else {
// Union was computed and stored in 'it->second': copy to 'region'.
region->getConstraints()->clearAndCopyFrom(
*it->second->getConstraints());
}
return true;
};
bool existsInRead = updateRegion(readRegions);
if (error)
return;
bool existsInWrite = updateRegion(writeRegions);
if (error)
return;
// Finally add it to the region list.
if (region->isWrite() && !existsInWrite) {
writeRegions[region->memref] = std::move(region);
} else if (!region->isWrite() && !existsInRead) {
readRegions[region->memref] = std::move(region);
}
});
if (error) {
begin->emitError(
"copy generation failed for one or more memref's in this block\n");
return 0;
}
uint64_t totalCopyBuffersSizeInBytes = 0;
bool ret = true;
auto processRegions =
[&](const SmallMapVector<Value, std::unique_ptr<MemRefRegion>, 4>
&regions) {
for (const auto &regionEntry : regions) {
// For each region, hoist copy in/out past all hoistable
// 'affine.for's.
Block::iterator copyInPlacementStart, copyOutPlacementStart;
Block *copyPlacementBlock;
findHighestBlockForPlacement(
*regionEntry.second, *block, begin, end, &copyPlacementBlock,
&copyInPlacementStart, &copyOutPlacementStart);
uint64_t sizeInBytes;
Block::iterator nBegin, nEnd;
LogicalResult iRet = generateCopy(
*regionEntry.second, block, begin, end, copyPlacementBlock,
copyInPlacementStart, copyOutPlacementStart, copyOptions,
fastBufferMap, copyNests, &sizeInBytes, &nBegin, &nEnd);
if (succeeded(iRet)) {
// begin/end could have been invalidated, and need update.
begin = nBegin;
end = nEnd;
totalCopyBuffersSizeInBytes += sizeInBytes;
}
ret = ret & succeeded(iRet);
}
};
processRegions(readRegions);
processRegions(writeRegions);
if (!ret) {
begin->emitError(
"copy generation failed for one or more memref's in this block\n");
return totalCopyBuffersSizeInBytes;
}
// For a range of operations, a note will be emitted at the caller.
AffineForOp forOp;
uint64_t sizeInKib = llvm::divideCeil(totalCopyBuffersSizeInBytes, 1024);
if (llvm::DebugFlag && (forOp = dyn_cast<AffineForOp>(&*begin))) {
forOp.emitRemark()
<< sizeInKib
<< " KiB of copy buffers in fast memory space for this block\n";
}
if (totalCopyBuffersSizeInBytes > copyOptions.fastMemCapacityBytes) {
StringRef str = "Total size of all copy buffers' for this block "
"exceeds fast memory capacity\n";
block->getParentOp()->emitWarning(str);
}
return totalCopyBuffersSizeInBytes;
}
// A convenience version of affineDataCopyGenerate for all ops in the body of
// an AffineForOp.
uint64_t mlir::affineDataCopyGenerate(AffineForOp forOp,
const AffineCopyOptions &copyOptions,
Optional<Value> filterMemRef,
DenseSet<Operation *> &copyNests) {
return affineDataCopyGenerate(forOp.getBody()->begin(),
std::prev(forOp.getBody()->end()), copyOptions,
filterMemRef, copyNests);
}
LogicalResult mlir::generateCopyForMemRegion(
const MemRefRegion &memrefRegion, Operation *analyzedOp,
const AffineCopyOptions &copyOptions, CopyGenerateResult &result) {
Block *block = analyzedOp->getBlock();
auto begin = analyzedOp->getIterator();
auto end = std::next(begin);
DenseMap<Value, Value> fastBufferMap;
DenseSet<Operation *> copyNests;
auto err = generateCopy(memrefRegion, block, begin, end, block, begin, end,
copyOptions, fastBufferMap, copyNests,
&result.sizeInBytes, &begin, &end);
if (failed(err))
return err;
const auto &en = fastBufferMap.find(memrefRegion.memref);
// In some cases (empty loops), no copy generation would have happened.
if (en == fastBufferMap.end())
return failure();
result.alloc = en->second.getDefiningOp();
assert(result.alloc && "fast buffer expected to be locally allocated");
assert(copyNests.size() <= 1 && "At most one copy nest is expected.");
result.copyNest = copyNests.empty() ? nullptr : *copyNests.begin();
return success();
}
/// Gathers all AffineForOps in 'block' at 'currLoopDepth' in 'depthToLoops'.
static void
gatherLoopsInBlock(Block *block, unsigned currLoopDepth,
std::vector<SmallVector<AffineForOp, 2>> &depthToLoops) {
// Add a new empty level to output if it doesn't exist level already.
assert(currLoopDepth <= depthToLoops.size() && "Unexpected currLoopDepth");
if (currLoopDepth == depthToLoops.size())
depthToLoops.push_back(SmallVector<AffineForOp, 2>());
for (auto &op : *block) {
if (auto forOp = dyn_cast<AffineForOp>(op)) {
depthToLoops[currLoopDepth].push_back(forOp);
gatherLoopsInBlock(forOp.getBody(), currLoopDepth + 1, depthToLoops);
}
}
}
/// Gathers all AffineForOps in 'builtin.func' grouped by loop depth.
void mlir::gatherLoops(FuncOp func,
std::vector<SmallVector<AffineForOp, 2>> &depthToLoops) {
for (auto &block : func)
gatherLoopsInBlock(&block, /*currLoopDepth=*/0, depthToLoops);
// Remove last loop level from output since it's empty.
if (!depthToLoops.empty()) {
assert(depthToLoops.back().empty() && "Last loop level is not empty?");
depthToLoops.pop_back();
}
}
// TODO: if necessary, this can be extended to also compose in any
// affine.applys, fold to constant if all result dimensions of the map are
// constant (canonicalizeMapAndOperands below already does this for single
// result bound maps), and use simplifyMap to perform algebraic simplification.
AffineForOp mlir::createCanonicalizedAffineForOp(
OpBuilder b, Location loc, ValueRange lbOperands, AffineMap lbMap,
ValueRange ubOperands, AffineMap ubMap, int64_t step) {
SmallVector<Value, 4> lowerOperands(lbOperands);
SmallVector<Value, 4> upperOperands(ubOperands);
fullyComposeAffineMapAndOperands(&lbMap, &lowerOperands);
canonicalizeMapAndOperands(&lbMap, &lowerOperands);
lbMap = removeDuplicateExprs(lbMap);
fullyComposeAffineMapAndOperands(&ubMap, &upperOperands);
canonicalizeMapAndOperands(&ubMap, &upperOperands);
ubMap = removeDuplicateExprs(ubMap);
return b.create<AffineForOp>(loc, lowerOperands, lbMap, upperOperands, ubMap,
step);
}
/// Creates an AffineIfOp that encodes the conditional to choose between
/// the constant trip count version and an unknown trip count version of this
/// nest of loops. This is used to separate partial and full tiles if `loops`
/// has the intra-tile loops. The affine.if op is inserted at the builder
/// insertion point of `b`.
static AffineIfOp createSeparationCondition(MutableArrayRef<AffineForOp> loops,
OpBuilder b) {
if (loops.empty())
return nullptr;
auto *context = loops[0].getContext();
FlatAffineValueConstraints cst;
SmallVector<Operation *, 8> ops;
ops.reserve(loops.size());
for (AffineForOp forOp : loops)
ops.push_back(forOp);
(void)getIndexSet(ops, &cst);
// Remove constraints that are independent of these loop IVs.
cst.removeIndependentConstraints(/*pos=*/0, /*num=*/loops.size());
// Construct the constraint set representing the guard for full tiles. The
// lower bound (and upper bound) corresponding to the full tile should be
// larger (and resp. smaller) than any other lower (or upper bound).
SmallVector<int64_t, 8> fullTileLb, fullTileUb;
for (auto loop : loops) {
(void)loop;
// TODO: Non-unit stride is not an issue to generalize to.
assert(loop.getStep() == 1 && "point loop step expected to be one");
// Mark everything symbols for the purpose of finding a constant diff pair.
cst.setDimSymbolSeparation(/*newSymbolCount=*/cst.getNumDimAndSymbolIds() -
1);
unsigned fullTileLbPos, fullTileUbPos;
if (!cst.getConstantBoundOnDimSize(0, /*lb=*/nullptr,
/*lbFloorDivisor=*/nullptr,
/*ub=*/nullptr, &fullTileLbPos,
&fullTileUbPos)) {
LLVM_DEBUG(llvm::dbgs() << "Can't get constant diff pair for a loop\n");
return nullptr;
}
SmallVector<unsigned, 4> lbIndices, ubIndices;
cst.getLowerAndUpperBoundIndices(/*pos=*/0, &lbIndices, &ubIndices);
auto fLb = cst.getInequality(fullTileLbPos);
auto fUb = cst.getInequality(fullTileUbPos);
fullTileLb.assign(fLb.begin(), fLb.end());
fullTileUb.assign(fUb.begin(), fUb.end());
// Full tile lower bound should be >= than any other lower bound.
for (auto lbIndex : lbIndices)
for (unsigned i = 0, e = cst.getNumCols(); i < e; ++i)
cst.atIneq(lbIndex, i) = fullTileLb[i] - cst.atIneq(lbIndex, i);
// Full tile upper bound should be <= any other upper bound.
for (auto ubIndex : ubIndices)
for (unsigned i = 0, e = cst.getNumCols(); i < e; ++i)
cst.atIneq(ubIndex, i) -= fullTileUb[i];
cst.removeId(0);
}
// The previous step leads to all zeros for the full tile lb and ub position
// itself; remove those and any other duplicates / trivial redundancies.
cst.removeTrivialRedundancy();
// Turn everything into dims conservatively since we earlier turned all
// trailing ids past point loop IV into symbols. Some of these could be outer
// loop IVs; we'll canonicalize anyway.
cst.setDimSymbolSeparation(0);
IntegerSet ifCondSet = cst.getAsIntegerSet(context);
// ifCondSet can be null if cst was empty -- this can happen if all loops
// in the nest have constant trip counts.
if (!ifCondSet)
return nullptr;
SmallVector<Value, 4> setOperands;
cst.getValues(0, cst.getNumDimAndSymbolIds(), &setOperands);
canonicalizeSetAndOperands(&ifCondSet, &setOperands);
return b.create<AffineIfOp>(loops[0].getLoc(), ifCondSet, setOperands,
/*withElseRegion=*/true);
}
/// Create the full tile loop nest (along with its body).
static LogicalResult
createFullTiles(MutableArrayRef<AffineForOp> inputNest,
SmallVectorImpl<AffineForOp> &fullTileLoops, OpBuilder b) {
fullTileLoops.reserve(inputNest.size());
// For each loop in the original nest identify a lower/upper bound pair such
// that their difference is a constant.
FlatAffineValueConstraints cst;
for (auto loop : inputNest) {
// TODO: straightforward to generalize to a non-unit stride.
if (loop.getStep() != 1) {
LLVM_DEBUG(llvm::dbgs()
<< "[tile separation] non-unit stride not implemented\n");
return failure();
}
SmallVector<Operation *, 1> loopOp{loop.getOperation()};
(void)getIndexSet(loopOp, &cst);
// We will mark everything other than this loop IV as symbol for getting a
// pair of <lb, ub> with a constant difference.
cst.setDimSymbolSeparation(cst.getNumDimAndSymbolIds() - 1);
unsigned lbPos, ubPos;
if (!cst.getConstantBoundOnDimSize(/*pos=*/0, /*lb=*/nullptr,
/*lbDivisor=*/nullptr, /*ub=*/nullptr,
&lbPos, &ubPos) ||
lbPos == ubPos) {
LLVM_DEBUG(llvm::dbgs() << "[tile separation] Can't get constant diff / "
"equalities not yet handled\n");
return failure();
}
// Set all identifiers as dimensions uniformly since some of those marked as
// symbols above could be outer loop IVs (corresponding tile space IVs).
cst.setDimSymbolSeparation(/*newSymbolCount=*/0);
AffineValueMap lbVmap, ubVmap;
cst.getIneqAsAffineValueMap(/*pos=*/0, lbPos, lbVmap, b.getContext());
cst.getIneqAsAffineValueMap(/*pos=*/0, ubPos, ubVmap, b.getContext());
AffineForOp fullTileLoop = createCanonicalizedAffineForOp(
b, loop.getLoc(), lbVmap.getOperands(), lbVmap.getAffineMap(),
ubVmap.getOperands(), ubVmap.getAffineMap());
b = OpBuilder::atBlockTerminator(fullTileLoop.getBody());
fullTileLoops.push_back(fullTileLoop);
}
// Add the body for the full tile loop nest.
BlockAndValueMapping operandMap;
for (auto loopEn : llvm::enumerate(inputNest))
operandMap.map(loopEn.value().getInductionVar(),
fullTileLoops[loopEn.index()].getInductionVar());
b = OpBuilder::atBlockTerminator(fullTileLoops.back().getBody());
for (auto &op : inputNest.back().getBody()->without_terminator())
b.clone(op, operandMap);
return success();
}
LogicalResult
mlir::separateFullTiles(MutableArrayRef<AffineForOp> inputNest,
SmallVectorImpl<AffineForOp> *fullTileNest) {
if (inputNest.empty())
return success();
auto firstLoop = inputNest[0];
// Each successive for op has to be nested in the other.
auto prevLoop = firstLoop;
for (auto loop : inputNest.drop_front(1)) {
assert(loop->getParentOp() == prevLoop && "input not contiguously nested");
prevLoop = loop;
}
// Create the full tile loop nest.
SmallVector<AffineForOp, 4> fullTileLoops;
OpBuilder b(firstLoop);
if (failed(createFullTiles(inputNest, fullTileLoops, b))) {
if (!fullTileLoops.empty())
fullTileLoops.front().erase();
return failure();
}
// Create and insert the version select right before the root of the nest.
b = OpBuilder(firstLoop);
AffineIfOp ifOp = createSeparationCondition(inputNest, b);
if (!ifOp) {
fullTileLoops.front().erase();
LLVM_DEBUG(llvm::dbgs() << "All tiles are full tiles, or failure creating "
"separation condition\n");
return failure();
}
// Move the full tile into the then block.
Block *thenBlock = ifOp.getThenBlock();
AffineForOp outermostFullTileLoop = fullTileLoops[0];
thenBlock->getOperations().splice(
std::prev(thenBlock->end()),
outermostFullTileLoop->getBlock()->getOperations(),
Block::iterator(outermostFullTileLoop));
// Move the partial tile into the else block. The partial tile is the same as
// the original loop nest.
Block *elseBlock = ifOp.getElseBlock();
elseBlock->getOperations().splice(std::prev(elseBlock->end()),
firstLoop->getBlock()->getOperations(),
Block::iterator(firstLoop));
if (fullTileNest)
*fullTileNest = std::move(fullTileLoops);
return success();
}