[SCEV] Split collecting and applying rewrite info from loop guards (NFC) #97316
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-llvm-analysis Author: Florian Hahn (fhahn) ChangesSplit off collecting rewrite info from loop guards to collectRewriteInfoFromLoopGuards. This allows users of applyLoopGuards to collect rewrite info once in cases where the same loop guards are applied multiple times. This is used to collect rewrite info once in howFarToZero, which saves a bit of compile-time: Notably this improves mafft by -0.9% with -O3, -0.11% with LTO and -0.12% with stage2-O3. Full diff: https://github.com/llvm/llvm-project/pull/97316.diff 2 Files Affected:
diff --git a/llvm/include/llvm/Analysis/ScalarEvolution.h b/llvm/include/llvm/Analysis/ScalarEvolution.h
index 97b30daf4427a..d9173e8745688 100644
--- a/llvm/include/llvm/Analysis/ScalarEvolution.h
+++ b/llvm/include/llvm/Analysis/ScalarEvolution.h
@@ -1299,8 +1299,17 @@ class ScalarEvolution {
/// sharpen it.
void setNoWrapFlags(SCEVAddRecExpr *AddRec, SCEV::NoWrapFlags Flags);
+ /// Collect rewrite map for loop guards for loop \p L, together with flags
+ /// indidcating if NUW and NSW can be preserved during rewriting.
+ std::tuple<DenseMap<const SCEV *, const SCEV *>, bool, bool>
+ collectRewriteInfoFromLoopGuards(const Loop *L);
+
/// Try to apply information from loop guards for \p L to \p Expr.
const SCEV *applyLoopGuards(const SCEV *Expr, const Loop *L);
+ const SCEV *
+ applyLoopGuards(const SCEV *Expr, const Loop *L,
+ const DenseMap<const SCEV *, const SCEV *> &RewriteMap,
+ bool PreserveNUW, bool PreserveNSW);
/// Return true if the loop has no abnormal exits. That is, if the loop
/// is not infinite, it must exit through an explicit edge in the CFG.
diff --git a/llvm/lib/Analysis/ScalarEvolution.cpp b/llvm/lib/Analysis/ScalarEvolution.cpp
index e998fe9452ad7..9e39ceb6526e1 100644
--- a/llvm/lib/Analysis/ScalarEvolution.cpp
+++ b/llvm/lib/Analysis/ScalarEvolution.cpp
@@ -10490,8 +10490,11 @@ ScalarEvolution::ExitLimit ScalarEvolution::howFarToZero(const SCEV *V,
if (!isLoopInvariant(Step, L))
return getCouldNotCompute();
+ const auto &[RewriteMap, PreserveNUW, PreserveNSW] =
+ collectRewriteInfoFromLoopGuards(L);
// Specialize step for this loop so we get context sensitive facts below.
- const SCEV *StepWLG = applyLoopGuards(Step, L);
+ const SCEV *StepWLG =
+ applyLoopGuards(Step, L, RewriteMap, PreserveNUW, PreserveNSW);
// For positive steps (counting up until unsigned overflow):
// N = -Start/Step (as unsigned)
@@ -10508,7 +10511,8 @@ ScalarEvolution::ExitLimit ScalarEvolution::howFarToZero(const SCEV *V,
// N = Distance (as unsigned)
if (StepC &&
(StepC->getValue()->isOne() || StepC->getValue()->isMinusOne())) {
- APInt MaxBECount = getUnsignedRangeMax(applyLoopGuards(Distance, L));
+ APInt MaxBECount = getUnsignedRangeMax(
+ applyLoopGuards(Distance, L, RewriteMap, PreserveNUW, PreserveNSW));
MaxBECount = APIntOps::umin(MaxBECount, getUnsignedRangeMax(Distance));
// When a loop like "for (int i = 0; i != n; ++i) { /* body */ }" is rotated,
@@ -10549,7 +10553,8 @@ ScalarEvolution::ExitLimit ScalarEvolution::howFarToZero(const SCEV *V,
getUDivExpr(Distance, CountDown ? getNegativeSCEV(Step) : Step);
const SCEV *ConstantMax = getCouldNotCompute();
if (Exact != getCouldNotCompute()) {
- APInt MaxInt = getUnsignedRangeMax(applyLoopGuards(Exact, L));
+ APInt MaxInt = getUnsignedRangeMax(
+ applyLoopGuards(Exact, L, RewriteMap, PreserveNUW, PreserveNSW));
ConstantMax =
getConstant(APIntOps::umin(MaxInt, getUnsignedRangeMax(Exact)));
}
@@ -10566,7 +10571,8 @@ ScalarEvolution::ExitLimit ScalarEvolution::howFarToZero(const SCEV *V,
const SCEV *M = E;
if (E != getCouldNotCompute()) {
- APInt MaxWithGuards = getUnsignedRangeMax(applyLoopGuards(E, L));
+ APInt MaxWithGuards = getUnsignedRangeMax(
+ applyLoopGuards(E, L, RewriteMap, PreserveNUW, PreserveNSW));
M = getConstant(APIntOps::umin(MaxWithGuards, getUnsignedRangeMax(E)));
}
auto *S = isa<SCEVCouldNotCompute>(E) ? M : E;
@@ -15096,7 +15102,7 @@ class SCEVLoopGuardRewriter : public SCEVRewriteVisitor<SCEVLoopGuardRewriter> {
public:
SCEVLoopGuardRewriter(ScalarEvolution &SE,
- DenseMap<const SCEV *, const SCEV *> &M,
+ const DenseMap<const SCEV *, const SCEV *> &M,
bool PreserveNUW, bool PreserveNSW)
: SCEVRewriteVisitor(SE), Map(M) {
if (PreserveNUW)
@@ -15191,7 +15197,8 @@ class SCEVLoopGuardRewriter : public SCEVRewriteVisitor<SCEVLoopGuardRewriter> {
}
};
-const SCEV *ScalarEvolution::applyLoopGuards(const SCEV *Expr, const Loop *L) {
+std::tuple<DenseMap<const SCEV *, const SCEV *>, bool, bool>
+ScalarEvolution::collectRewriteInfoFromLoopGuards(const Loop *L) {
SmallVector<const SCEV *> ExprsToRewrite;
auto CollectCondition = [&](ICmpInst::Predicate Predicate, const SCEV *LHS,
const SCEV *RHS,
@@ -15600,9 +15607,6 @@ const SCEV *ScalarEvolution::applyLoopGuards(const SCEV *Expr, const Loop *L) {
}
}
- if (RewriteMap.empty())
- return Expr;
-
// Let the rewriter preserve NUW/NSW flags if the unsigned/signed ranges of
// the replacement expressions are contained in the ranges of the replaced
// expressions.
@@ -15626,6 +15630,22 @@ const SCEV *ScalarEvolution::applyLoopGuards(const SCEV *Expr, const Loop *L) {
RewriteMap.insert({Expr, Rewriter.visit(RewriteTo)});
}
}
+ return {RewriteMap, PreserveNUW, PreserveNSW};
+}
+
+const SCEV *ScalarEvolution::applyLoopGuards(const SCEV *Expr, const Loop *L) {
+ const auto &[RewriteMap, PreserveNUW, PreserveNSW] =
+ collectRewriteInfoFromLoopGuards(L);
+ return applyLoopGuards(Expr, L, RewriteMap, PreserveNUW, PreserveNSW);
+}
+
+const SCEV *ScalarEvolution::applyLoopGuards(
+ const SCEV *Expr, const Loop *L,
+ const DenseMap<const SCEV *, const SCEV *> &RewriteMap, bool PreserveNUW,
+ bool PreserveNSW) {
+ if (RewriteMap.empty())
+ return Expr;
+
SCEVLoopGuardRewriter Rewriter(*this, RewriteMap, PreserveNUW, PreserveNSW);
return Rewriter.visit(Expr);
}
|
fhahn
changed the title
nikic
reviewed
Jul 2, 2024
| @@ -1299,8 +1299,17 @@ class ScalarEvolution { | |||
| /// sharpen it. | |||
| void setNoWrapFlags(SCEVAddRecExpr *AddRec, SCEV::NoWrapFlags Flags); | |||
|
|
|||
| /// Collect rewrite map for loop guards for loop \p L, together with flags | |||
| /// indidcating if NUW and NSW can be preserved during rewriting. | |||
| std::tuple<DenseMap<const SCEV *, const SCEV *>, bool, bool> | |||
There was a problem hiding this comment.
Can we make this a class LoopGuards instead?
There was a problem hiding this comment.
Moved collection and rewriting to a LoopGuards class, thanks!
Split off collecting rewrite info from loop guards to collectRewriteInfoFromLoopGuards. This allows users of applyLoopGuards to collect rewrite info once in cases where the same loop guards are applied multiple times. This is used to collect rewrite info once in howFarToZero, which saves a bit of compile-time: stage1-O3: -0.04% stage1-ReleaseThinLTO: -0.02% stage1-ReleaseLTO-g: -0.04% stage2-O3: -0.02% https://llvm-compile-time-tracker.com/compare.php?from=117b53ae38428ca66eaa886fb432e6f09db88fe4&to=4ffb7b2e1c99081ccebe6f236c48a0be2f64b6ff&stat=instructions:u Notably this improves mafft by -0.9% with -O3, -0.11% with LTO and -0.12% with stage2-O3.
fhahn
force-pushed
the
scev-cache-loop-guards
branch
from
a16eeca to
78f6143
Compare
nikic
approved these changes
Jul 2, 2024
|
|
||
| public: | ||
| /// Collect rewrite map for loop guards for loop \p L, together with flags | ||
| /// indidcating if NUW and NSW can be preserved during rewriting. |
There was a problem hiding this comment.
Suggested change
| /// indidcating if NUW and NSW can be preserved during rewriting. | |
| /// indidcating if NUW and NSW can be preserved during rewriting. |
Suggested change
| /// indidcating if NUW and NSW can be preserved during rewriting. | |
| /// indicating if NUW and NSW can be preserved during rewriting. |
| /// indidcating if NUW and NSW can be preserved during rewriting. | ||
| static LoopGuards collect(const Loop *L, ScalarEvolution &SE); | ||
|
|
||
| /// Try to apply the collected loop guards to \p Expr |
There was a problem hiding this comment.
Suggested change
| /// Try to apply the collected loop guards to \p Expr | |
| /// Try to apply the collected loop guards to \p Expr. |
| /// Try to apply information from loop guards for \p L to \p Expr. | ||
| const SCEV *applyLoopGuards(const SCEV *Expr, const Loop *L); | ||
| const SCEV *applyLoopGuards(const SCEV *Expr, const LoopGuards &Gards); |
There was a problem hiding this comment.
Suggested change
| const SCEV *applyLoopGuards(const SCEV *Expr, const LoopGuards &Gards); | |
| const SCEV *applyLoopGuards(const SCEV *Expr, const LoopGuards &Guards); |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/66/builds/920 Here is the relevant piece of the build log for the reference: |
lravenclaw
pushed a commit
to lravenclaw/llvm-project
that referenced
this pull request
Jul 3, 2024
…FC) (llvm#97316) Introduce a new LoopGuards class to track info from loop guards and split off collecting rewrite info to LoopGuards::collect. This allows users of applyLoopGuards to collect rewrite info once in cases where the same loop guards are applied multiple times. This is used to collect rewrite info once in howFarToZero, which saves a bit of compile-time: stage1-O3: -0.04% stage1-ReleaseThinLTO: -0.02% stage1-ReleaseLTO-g: -0.04% stage2-O3: -0.02% https://llvm-compile-time-tracker.com/compare.php?from=117b53ae38428ca66eaa886fb432e6f09db88fe4&to=4ffb7b2e1c99081ccebe6f236c48a0be2f64b6ff&stat=instructions:u Notably this improves mafft by -0.9% with -O3, -0.11% with LTO and -0.12% with stage2-O3. PR: llvm#97316
kbluck
pushed a commit
to kbluck/llvm-project
that referenced
this pull request
Jul 6, 2024
…FC) (llvm#97316) Introduce a new LoopGuards class to track info from loop guards and split off collecting rewrite info to LoopGuards::collect. This allows users of applyLoopGuards to collect rewrite info once in cases where the same loop guards are applied multiple times. This is used to collect rewrite info once in howFarToZero, which saves a bit of compile-time: stage1-O3: -0.04% stage1-ReleaseThinLTO: -0.02% stage1-ReleaseLTO-g: -0.04% stage2-O3: -0.02% https://llvm-compile-time-tracker.com/compare.php?from=117b53ae38428ca66eaa886fb432e6f09db88fe4&to=4ffb7b2e1c99081ccebe6f236c48a0be2f64b6ff&stat=instructions:u Notably this improves mafft by -0.9% with -O3, -0.11% with LTO and -0.12% with stage2-O3. PR: llvm#97316
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Split off collecting rewrite info from loop guards to collectRewriteInfoFromLoopGuards. This allows users of applyLoopGuards to collect rewrite info once in cases where the same loop guards are applied multiple times.
This is used to collect rewrite info once in howFarToZero, which saves a bit of compile-time:
stage1-O3: -0.04%
stage1-ReleaseThinLTO: -0.02%
stage1-ReleaseLTO-g: -0.04%
stage2-O3: -0.02%
https://llvm-compile-time-tracker.com/compare.php?from=117b53ae38428ca66eaa886fb432e6f09db88fe4&to=4ffb7b2e1c99081ccebe6f236c48a0be2f64b6ff&stat=instructions:u
Notably this improves mafft by -0.9% with -O3, -0.11% with LTO and -0.12% with stage2-O3.