• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //===- AffineDataCopyGeneration.cpp - Explicit memref copying pass ------*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file implements a pass to automatically promote accessed memref regions
10 // to buffers in a faster memory space that is explicitly managed, with the
11 // necessary data movement operations performed through either regular
12 // point-wise load/store's or DMAs. Such explicit copying (also referred to as
13 // array packing/unpacking in the literature), when done on arrays that exhibit
14 // reuse, results in near elimination of conflict misses, TLB misses, reduced
15 // use of hardware prefetch streams, and reduced false sharing. It is also
16 // necessary for hardware that explicitly managed levels in the memory
17 // hierarchy, and where DMAs may have to be used. This optimization is often
18 // performed on already tiled code.
19 //
20 //===----------------------------------------------------------------------===//
21 
22 #include "PassDetail.h"
23 #include "mlir/Analysis/Utils.h"
24 #include "mlir/Dialect/Affine/IR/AffineOps.h"
25 #include "mlir/Dialect/Affine/Passes.h"
26 #include "mlir/Dialect/StandardOps/IR/Ops.h"
27 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
28 #include "mlir/Transforms/LoopUtils.h"
29 #include "llvm/ADT/MapVector.h"
30 #include "llvm/Support/CommandLine.h"
31 #include "llvm/Support/Debug.h"
32 #include <algorithm>
33 
34 #define DEBUG_TYPE "affine-data-copy-generate"
35 
36 using namespace mlir;
37 
38 namespace {
39 
40 /// Replaces all loads and stores on memref's living in 'slowMemorySpace' by
41 /// introducing copy operations to transfer data into `fastMemorySpace` and
42 /// rewriting the original load's/store's to instead load/store from the
43 /// allocated fast memory buffers. Additional options specify the identifier
44 /// corresponding to the fast memory space and the amount of fast memory space
45 /// available. The pass traverses through the nesting structure, recursing to
46 /// inner levels if necessary to determine at what depth copies need to be
47 /// placed so that the allocated buffers fit within the memory capacity
48 /// provided.
49 // TODO: We currently can't generate copies correctly when stores
50 // are strided. Check for strided stores.
51 struct AffineDataCopyGeneration
52     : public AffineDataCopyGenerationBase<AffineDataCopyGeneration> {
53   AffineDataCopyGeneration() = default;
AffineDataCopyGeneration__anonb238fc050111::AffineDataCopyGeneration54   explicit AffineDataCopyGeneration(unsigned slowMemorySpace,
55                                     unsigned fastMemorySpace,
56                                     unsigned tagMemorySpace,
57                                     int minDmaTransferSize,
58                                     uint64_t fastMemCapacityBytes) {
59     this->slowMemorySpace = slowMemorySpace;
60     this->fastMemorySpace = fastMemorySpace;
61     this->tagMemorySpace = tagMemorySpace;
62     this->minDmaTransferSize = minDmaTransferSize;
63     this->fastMemoryCapacity = fastMemCapacityBytes / 1024;
64   }
65 
66   void runOnFunction() override;
67   LogicalResult runOnBlock(Block *block, DenseSet<Operation *> &copyNests);
68 
69   // Constant zero index to avoid too many duplicates.
70   Value zeroIndex = nullptr;
71 };
72 
73 } // end anonymous namespace
74 
75 /// Generates copies for memref's living in 'slowMemorySpace' into newly created
76 /// buffers in 'fastMemorySpace', and replaces memory operations to the former
77 /// by the latter. Only load op's handled for now.
78 /// TODO: extend this to store op's.
createAffineDataCopyGenerationPass(unsigned slowMemorySpace,unsigned fastMemorySpace,unsigned tagMemorySpace,int minDmaTransferSize,uint64_t fastMemCapacityBytes)79 std::unique_ptr<OperationPass<FuncOp>> mlir::createAffineDataCopyGenerationPass(
80     unsigned slowMemorySpace, unsigned fastMemorySpace, unsigned tagMemorySpace,
81     int minDmaTransferSize, uint64_t fastMemCapacityBytes) {
82   return std::make_unique<AffineDataCopyGeneration>(
83       slowMemorySpace, fastMemorySpace, tagMemorySpace, minDmaTransferSize,
84       fastMemCapacityBytes);
85 }
86 std::unique_ptr<OperationPass<FuncOp>>
createAffineDataCopyGenerationPass()87 mlir::createAffineDataCopyGenerationPass() {
88   return std::make_unique<AffineDataCopyGeneration>();
89 }
90 
91 /// Generate copies for this block. The block is partitioned into separate
92 /// ranges: each range is either a sequence of one or more operations starting
93 /// and ending with an affine load or store op, or just an affine.forop (which
94 /// could have other affine for op's nested within).
95 LogicalResult
runOnBlock(Block * block,DenseSet<Operation * > & copyNests)96 AffineDataCopyGeneration::runOnBlock(Block *block,
97                                      DenseSet<Operation *> &copyNests) {
98   if (block->empty())
99     return success();
100 
101   uint64_t fastMemCapacityBytes =
102       fastMemoryCapacity != std::numeric_limits<uint64_t>::max()
103           ? fastMemoryCapacity * 1024
104           : fastMemoryCapacity;
105   AffineCopyOptions copyOptions = {generateDma, slowMemorySpace,
106                                    fastMemorySpace, tagMemorySpace,
107                                    fastMemCapacityBytes};
108 
109   // Every affine.forop in the block starts and ends a block range for copying;
110   // in addition, a contiguous sequence of operations starting with a
111   // load/store op but not including any copy nests themselves is also
112   // identified as a copy block range. Straightline code (a contiguous chunk of
113   // operations excluding AffineForOp's) are always assumed to not exhaust
114   // memory. As a result, this approach is conservative in some cases at the
115   // moment; we do a check later and report an error with location info.
116   // TODO: An 'affine.if' operation is being treated similar to an
117   // operation. 'affine.if''s could have 'affine.for's in them;
118   // treat them separately.
119 
120   // Get to the first load, store, or for op (that is not a copy nest itself).
121   auto curBegin =
122       std::find_if(block->begin(), block->end(), [&](Operation &op) {
123         return isa<AffineLoadOp, AffineStoreOp, AffineForOp>(op) &&
124                copyNests.count(&op) == 0;
125       });
126 
127   // Create [begin, end) ranges.
128   auto it = curBegin;
129   while (it != block->end()) {
130     AffineForOp forOp;
131     // If you hit a non-copy for loop, we will split there.
132     if ((forOp = dyn_cast<AffineForOp>(&*it)) && copyNests.count(forOp) == 0) {
133       // Perform the copying up unti this 'for' op first.
134       affineDataCopyGenerate(/*begin=*/curBegin, /*end=*/it, copyOptions,
135                              /*filterMemRef=*/llvm::None, copyNests);
136 
137       // Returns true if the footprint is known to exceed capacity.
138       auto exceedsCapacity = [&](AffineForOp forOp) {
139         Optional<int64_t> footprint =
140             getMemoryFootprintBytes(forOp,
141                                     /*memorySpace=*/0);
142         return (footprint.hasValue() &&
143                 static_cast<uint64_t>(footprint.getValue()) >
144                     fastMemCapacityBytes);
145       };
146 
147       // If the memory footprint of the 'affine.for' loop is higher than fast
148       // memory capacity (when provided), we recurse to copy at an inner level
149       // until we find a depth at which footprint fits in fast mem capacity. If
150       // the footprint can't be calculated, we assume for now it fits. Recurse
151       // inside if footprint for 'forOp' exceeds capacity, or when
152       // skipNonUnitStrideLoops is set and the step size is not one.
153       bool recurseInner = skipNonUnitStrideLoops ? forOp.getStep() != 1
154                                                  : exceedsCapacity(forOp);
155       if (recurseInner) {
156         // We'll recurse and do the copies at an inner level for 'forInst'.
157         // Recurse onto the body of this loop.
158         runOnBlock(forOp.getBody(), copyNests);
159       } else {
160         // We have enough capacity, i.e., copies will be computed for the
161         // portion of the block until 'it', and for 'it', which is 'forOp'. Note
162         // that for the latter, the copies are placed just before this loop (for
163         // incoming copies) and right after (for outgoing ones).
164 
165         // Inner loop copies have their own scope - we don't thus update
166         // consumed capacity. The footprint check above guarantees this inner
167         // loop's footprint fits.
168         affineDataCopyGenerate(/*begin=*/it, /*end=*/std::next(it), copyOptions,
169                                /*filterMemRef=*/llvm::None, copyNests);
170       }
171       // Get to the next load or store op after 'forOp'.
172       curBegin = std::find_if(std::next(it), block->end(), [&](Operation &op) {
173         return isa<AffineLoadOp, AffineStoreOp, AffineForOp>(op) &&
174                copyNests.count(&op) == 0;
175       });
176       it = curBegin;
177     } else {
178       assert(copyNests.count(&*it) == 0 &&
179              "all copy nests generated should have been skipped above");
180       // We simply include this op in the current range and continue for more.
181       ++it;
182     }
183   }
184 
185   // Generate the copy for the final block range.
186   if (curBegin != block->end()) {
187     // Can't be a terminator because it would have been skipped above.
188     assert(!curBegin->isKnownTerminator() && "can't be a terminator");
189     // Exclude the affine.yield - hence, the std::prev.
190     affineDataCopyGenerate(/*begin=*/curBegin, /*end=*/std::prev(block->end()),
191                            copyOptions, /*filterMemRef=*/llvm::None, copyNests);
192   }
193 
194   return success();
195 }
196 
runOnFunction()197 void AffineDataCopyGeneration::runOnFunction() {
198   FuncOp f = getFunction();
199   OpBuilder topBuilder(f.getBody());
200   zeroIndex = topBuilder.create<ConstantIndexOp>(f.getLoc(), 0);
201 
202   // Nests that are copy-in's or copy-out's; the root AffineForOps of those
203   // nests are stored herein.
204   DenseSet<Operation *> copyNests;
205 
206   // Clear recorded copy nests.
207   copyNests.clear();
208 
209   for (auto &block : f)
210     runOnBlock(&block, copyNests);
211 
212   // Promote any single iteration loops in the copy nests and collect
213   // load/stores to simplify.
214   SmallVector<Operation *, 4> copyOps;
215   for (Operation *nest : copyNests)
216     // With a post order walk, the erasure of loops does not affect
217     // continuation of the walk or the collection of load/store ops.
218     nest->walk([&](Operation *op) {
219       if (auto forOp = dyn_cast<AffineForOp>(op))
220         promoteIfSingleIteration(forOp);
221       else if (isa<AffineLoadOp, AffineStoreOp>(op))
222         copyOps.push_back(op);
223     });
224 
225   // Promoting single iteration loops could lead to simplification of
226   // contained load's/store's, and the latter could anyway also be
227   // canonicalized.
228   OwningRewritePatternList patterns;
229   AffineLoadOp::getCanonicalizationPatterns(patterns, &getContext());
230   AffineStoreOp::getCanonicalizationPatterns(patterns, &getContext());
231   FrozenRewritePatternList frozenPatterns(std::move(patterns));
232   for (Operation *op : copyOps)
233     applyOpPatternsAndFold(op, frozenPatterns);
234 }
235