[Mlir-commits] [mlir] [mlir][TilingInterface] Allow controlling what fusion is done within tile and fuse (PR #76871)
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Wed Jan 3 14:29:42 PST 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-mlir
@llvm/pr-subscribers-mlir-scf
Author: None (MaheshRavishankar)
<details>
<summary>Changes</summary>
Currently the `tileConsumerAndFuseProducerGreedilyUsingSCFFor` method greedily fuses through all slices that are generated during the tile and fuse flow. That is not the normal use case. Ideally the caller would like to control which slices get fused and which dont. This patch introduces a new field to the `SCFTileAndFuseOptions` to specify this control.
The contol function also allows the caller to specify if the replacement for the fused producer needs to be yielded from within the tiled computation. This allows replacing the fused producers in case they have other uses. Without this the original producers still survive negating the utility of the fusion.
The change here also means that the name of the function `tileConsumerAndFuseProducerGreedily...` can be updated. Defering that to a later stage to reduce the churn of API changes.
---
Full diff: https://github.com/llvm/llvm-project/pull/76871.diff
3 Files Affected:
- (modified) mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h (+17)
- (modified) mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp (+47-18)
- (modified) mlir/test/lib/Interfaces/TilingInterface/TestTilingInterface.cpp (+34-64)
``````````diff
diff --git a/mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h b/mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h
index 2f8f337bb8057c..0571b4aede4885 100644
--- a/mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h
+++ b/mlir/include/mlir/Dialect/SCF/Transforms/TileUsingInterface.h
@@ -97,6 +97,23 @@ struct SCFTileAndFuseOptions {
tilingOptions = options;
return *this;
}
+
+ /// Control function to check if a slice needs to be fused or not,
+ /// The control function recieves
+ /// 1) the slice along which fusion is to be done,
+ /// 2) the producer value that is to be fused
+ /// 3) a boolean value set to `true` if the fusion is from
+ /// a destination operand.
+ using ControlFnTy = std::function<std::tuple<bool, bool>(
+ tensor::ExtractSliceOp candidateSliceOp, OpResult originalProducer,
+ bool isDestinationOperand)>;
+ ControlFnTy fusionControlFn = [](tensor::ExtractSliceOp, OpResult, bool) {
+ return std::make_tuple(true, false);
+ };
+ SCFTileAndFuseOptions &setFusionControlFn(ControlFnTy controlFn) {
+ fusionControlFn = controlFn;
+ return *this;
+ }
};
/// Fuse the producer of the source of `candidateSliceOp` by computing the
diff --git a/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp b/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
index 1b6b4db9d20907..22826cababe779 100644
--- a/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
+++ b/mlir/lib/Dialect/SCF/Transforms/TileUsingInterface.cpp
@@ -728,32 +728,36 @@ mlir::scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
}
// 1. First tile the consumer.
- SmallVector<scf::ForOp> forLoops;
SetVector<Operation *> fusedProducers, tiledAndFusedOps;
- DenseMap<Value, Value> replacements;
- llvm::SmallDenseMap<Value, int64_t> yieldedValueToResultNumber;
- {
- FailureOr<scf::SCFTilingResult> tilingResult =
- tileUsingSCFForOp(rewriter, consumer, options.tilingOptions);
- if (failed(tilingResult))
- return rewriter.notifyMatchFailure(consumer, "failed to tile consumer");
- for (auto *tiledOp : tilingResult->tiledOps)
- tiledAndFusedOps.insert(tiledOp);
- forLoops = castToTypedOperations<scf::ForOp>(tilingResult->loops);
- for (auto [index, origValue, replacement] :
- llvm::enumerate(consumer->getResults(), tilingResult->replacements)) {
- replacements[origValue] = replacement;
- yieldedValueToResultNumber[tilingResult->tiledOps.back()->getResult(
- index)] = index;
- }
- }
+ llvm::SmallDenseMap<Value, size_t> origProducerToLoopResultNum;
+ FailureOr<scf::SCFTilingResult> tilingResult =
+ tileUsingSCFForOp(rewriter, consumer, options.tilingOptions);
+ if (failed(tilingResult))
+ return rewriter.notifyMatchFailure(consumer, "failed to tile consumer");
+ for (auto *tiledOp : tilingResult->tiledOps)
+ tiledAndFusedOps.insert(tiledOp);
+ SmallVector<scf::ForOp> forLoops =
+ castToTypedOperations<scf::ForOp>(tilingResult->loops);
// If there are no loops generated, fusion is immaterial.
if (forLoops.empty()) {
+ DenseMap<Value, Value> replacements;
+ for (auto [origVal, replacement] :
+ llvm::zip_equal(consumer->getResults(), tilingResult->replacements)) {
+ replacements[origVal] = replacement;
+ }
return scf::SCFTileAndFuseResult{fusedProducers, tiledAndFusedOps,
getAsOperations(forLoops), replacements};
}
+ // To keep track of replacements for now just record the map from the original
+ // untiled value to the result number of the for loop. Since the loop gets
+ // potentially replaced during fusion, keeping the value directly wont work.
+ DenseMap<Value, size_t> origValToResultNumber;
+ for (auto [index, result] : llvm::enumerate(consumer->getResults())) {
+ origValToResultNumber[result] = index;
+ }
+
// 2. Typically, the operands of the tiled operation are slices of the
// operands of the untiled operation. These are expressed in IR using
// `tensor.extract_slice` operations with source being the operands of the
@@ -776,6 +780,18 @@ mlir::scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
tensor::ExtractSliceOp candidateSliceOp = candidates.front();
candidates.pop_front();
+ // Find the original producer of the slice.
+ auto [fusableProducer, destinationInitArg] =
+ getUntiledProducerFromSliceSource(&candidateSliceOp.getSourceMutable(),
+ forLoops);
+ if (!fusableProducer)
+ continue;
+
+ auto [fuseSlice, yieldReplacement] = options.fusionControlFn(
+ candidateSliceOp, fusableProducer, destinationInitArg.has_value());
+ if (!fuseSlice)
+ continue;
+
// The operands of the fused producer might themselved be slices of
// values produced by operations that implement the `TilingInterface`.
// Add these operations to the worklist.
@@ -784,6 +800,13 @@ mlir::scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
if (!fusedResult)
continue;
+ if (yieldReplacement) {
+ yieldReplacementForFusedProducer(rewriter, candidateSliceOp,
+ fusedResult.value(), forLoops);
+ origValToResultNumber[fusableProducer] =
+ forLoops.front().getNumResults() - 1;
+ }
+
if (Operation *tiledAndFusedOp =
fusedResult->tiledAndFusedProducer.getDefiningOp()) {
fusedProducers.insert(fusedResult->origProducer.getDefiningOp());
@@ -791,6 +814,12 @@ mlir::scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
addCandidateSlices(tiledAndFusedOp, candidates);
}
}
+
+ DenseMap<Value, Value> replacements;
+ for (auto [origVal, resultNumber] : origValToResultNumber) {
+ replacements[origVal] = forLoops.front()->getResult(resultNumber);
+ }
+
return scf::SCFTileAndFuseResult{fusedProducers, tiledAndFusedOps,
getAsOperations(forLoops), replacements};
}
diff --git a/mlir/test/lib/Interfaces/TilingInterface/TestTilingInterface.cpp b/mlir/test/lib/Interfaces/TilingInterface/TestTilingInterface.cpp
index 112ad6cbde8589..798293bc1327e1 100644
--- a/mlir/test/lib/Interfaces/TilingInterface/TestTilingInterface.cpp
+++ b/mlir/test/lib/Interfaces/TilingInterface/TestTilingInterface.cpp
@@ -311,80 +311,50 @@ struct TestTileConsumerFuseAndYieldProducerUsingSCFForOp
// Collect list of operations that can be tiled and fused.
llvm::SmallDenseSet<Operation *> tiledAndFusedOps =
collectTiledAndFusedOps(rootOp);
- auto isIgnoredUser = [&](Operation *user, scf::ForOp outerMostTiledLoop) {
- return tiledAndFusedOps.count(user) || isa<tensor::DimOp>(user) ||
- outerMostTiledLoop->isAncestor(user);
+ llvm::SmallDenseMap<Operation *, bool> yielded;
+ auto isIgnoredUser = [&](Operation *user) {
+ return tiledAndFusedOps.count(user) || isa<tensor::DimOp>(user);
};
-
- // The rest of this method is similar to
- // scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp, except that also
- // yields replacements for values of the fused producer.
-
- // 1. Tile the consumer.
- SmallVector<OpResult> yieldedValuesToOrigValues;
- FailureOr<scf::SCFTilingResult> tilingResult =
- scf::tileUsingSCFForOp(rewriter, rootOp, options);
- if (failed(tilingResult)) {
- return rewriter.notifyMatchFailure(rootOp,
- "failed to tile base operation");
+ for (Operation *op : tiledAndFusedOps) {
+ yielded[op] = llvm::any_of(op->getUsers(), [&](Operation *user) {
+ return !isIgnoredUser(user);
+ });
}
- yieldedValuesToOrigValues.append(rootOp->result_begin(),
- rootOp->result_end());
-
- // 2. Tiling each operation results in generation of slices. The source of
- // these slices could be producers that can be fused into the tiled loops by
- // computing the slices of these producers in-place. This results in more
- // slices created for operands of the "fused producer". This open up more
- // opportunities for fusion. Use a worklist to fuse greedily.
- auto addCandidateSlices =
- [](Operation *fusedOp, std::deque<tensor::ExtractSliceOp> &candidates) {
- for (Value operand : fusedOp->getOperands())
- if (auto sliceOp = operand.getDefiningOp<tensor::ExtractSliceOp>())
- candidates.push_back(sliceOp);
- };
- std::deque<tensor::ExtractSliceOp> candidates;
- addCandidateSlices(tilingResult->tiledOps.back(), candidates);
- OpBuilder::InsertionGuard g(rewriter);
- auto forLoops = llvm::to_vector(llvm::map_range(
- tilingResult->loops, [](auto op) { return cast<scf::ForOp>(op); }));
- while (!candidates.empty()) {
- // Traverse the slices in BFS fashion.
- tensor::ExtractSliceOp candidateSliceOp = candidates.front();
- candidates.pop_front();
-
- // Materialize the slice of the producer in place.
- std::optional<scf::SCFFuseProducerOfSliceResult> fusedProducer =
- tileAndFuseProducerOfSlice(rewriter, candidateSliceOp, forLoops);
- if (!fusedProducer)
- continue;
-
- // Check if the fused producer has other uses that require the value
- // to be yielded from within the tiled loop.
- OpResult untiledProducer = fusedProducer->origProducer;
- if (llvm::any_of(untiledProducer.getUsers(), [&](Operation *user) {
- return !isIgnoredUser(user, forLoops.front());
- })) {
- yieldReplacementForFusedProducer(rewriter, candidateSliceOp,
- fusedProducer.value(), forLoops);
- yieldedValuesToOrigValues.push_back(untiledProducer);
- }
+ scf::SCFTileAndFuseOptions tileAndFuseOptions;
+ tileAndFuseOptions.setTilingOptions(options);
+ scf::SCFTileAndFuseOptions::ControlFnTy controlFn =
+ [&](tensor::ExtractSliceOp candidateSliceOp, OpResult originalProducer,
+ bool isDestinationOperand) {
+ Operation *owner = originalProducer.getOwner();
+ return std::make_tuple(true,
+ yielded.contains(owner) && yielded[owner]);
+ };
+ tileAndFuseOptions.setFusionControlFn(controlFn);
- // Add more fusion candidates to the worklist.
- if (auto fusedProducerOp =
- fusedProducer->tiledAndFusedProducer.getDefiningOp())
- addCandidateSlices(fusedProducerOp, candidates);
+ FailureOr<scf::SCFTileAndFuseResult> tileAndFuseResult =
+ scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp(
+ rewriter, rootOp, tileAndFuseOptions);
+ if (failed(tileAndFuseResult)) {
+ return rewriter.notifyMatchFailure(
+ rootOp, "failed to tile and fuse with op as root");
}
- scf::ForOp outermostLoop = forLoops.front();
- for (auto [index, origVal] : llvm::enumerate(yieldedValuesToOrigValues)) {
- Value replacement = outermostLoop.getResult(index);
+ for (auto it : tileAndFuseResult->replacements) {
+ Value origVal = it.first;
+ Value replacement = it.second;
rewriter.replaceUsesWithIf(origVal, replacement, [&](OpOperand &use) {
- return !isIgnoredUser(use.getOwner(), outermostLoop);
+ Operation *user = use.getOwner();
+ return !isIgnoredUser(user) &&
+ !tileAndFuseResult->loops.front()->isAncestor(user);
});
}
+
rewriter.eraseOp(rootOp);
- filter.replaceTransformationFilter(rewriter, tilingResult->tiledOps.back());
+ for (auto tiledAndFusedOp : tileAndFuseResult->tiledAndFusedOps)
+ if (tiledAndFusedOp->hasAttr(kTransformMarker))
+ filter.replaceTransformationFilter(rewriter, tiledAndFusedOp);
+
return success();
}
``````````
</details>
https://github.com/llvm/llvm-project/pull/76871
More information about the Mlir-commits
mailing list