llvm-6502/lib/Analysis
Jakob Stoklund Olesen f7477470d3 Try to keep the cached inliner costs around for a bit longer for big functions.
The Caller cost info would be reset everytime a callee was inlined. If the
caller has lots of calls and there is some mutual recursion going on, the
caller cost info could be calculated many times.

This patch reduces inliner runtime from 240s to 0.5s for a function with 20000
small function calls.

This is a more conservative version of r98089 that doesn't break the clang
test CodeGenCXX/temp-order.cpp. That test relies on rather extreme inlining
for constant folding.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@98099 91177308-0d34-0410-b5e6-96231b3b80d8
2010-03-09 23:02:17 +00:00
..
IPA Update CMake build. 2010-03-01 19:42:47 +00:00
AliasAnalysis.cpp
AliasAnalysisCounter.cpp Fix a missing newline in debug output. 2010-03-01 17:42:55 +00:00
AliasAnalysisEvaluator.cpp There are two ways of checking for a given type, for example isa<PointerType>(T) 2010-02-16 11:11:14 +00:00
AliasDebugger.cpp
AliasSetTracker.cpp
Analysis.cpp
BasicAliasAnalysis.cpp There are two ways of checking for a given type, for example isa<PointerType>(T) 2010-02-16 11:11:14 +00:00
CaptureTracking.cpp There are two ways of checking for a given type, for example isa<PointerType>(T) 2010-02-16 11:11:14 +00:00
CFGPrinter.cpp
CMakeLists.txt
ConstantFolding.cpp Remove the code which constant-folded ptrtoint(inttoptr(x)+c) to 2010-02-23 16:35:41 +00:00
DbgInfoPrinter.cpp
DebugInfo.cpp Start using DIFile. See updated SourceLevelDebugging.html for more information. 2010-03-09 00:44:10 +00:00
DomPrinter.cpp
InlineCost.cpp Try to keep the cached inliner costs around for a bit longer for big functions. 2010-03-09 23:02:17 +00:00
InstCount.cpp
InstructionSimplify.cpp fix incorrect folding of icmp with undef, PR6481. 2010-03-03 19:46:03 +00:00
Interval.cpp
IntervalPartition.cpp
IVUsers.cpp Spelling fixes. 2010-03-01 17:49:51 +00:00
LazyValueInfo.cpp
LibCallAliasAnalysis.cpp
LibCallSemantics.cpp Remove Value::getNameLen 2009-07-26 08:34:35 +00:00
LiveValues.cpp Fix "the the" and similar typos. 2010-02-10 16:03:48 +00:00
LoopDependenceAnalysis.cpp
LoopInfo.cpp Make isLCSSA ignore uses in blocks not reachable from the entry block, 2010-03-09 01:53:33 +00:00
LoopPass.cpp
Makefile
MemoryBuiltins.cpp Fix "the the" and similar typos. 2010-02-10 16:03:48 +00:00
MemoryDependenceAnalysis.cpp Reapply r97010, the speculative revert failed. 2010-02-24 08:48:04 +00:00
PHITransAddr.cpp Reapply r97010, the speculative revert failed. 2010-02-24 08:48:04 +00:00
PointerTracking.cpp There are two ways of checking for a given type, for example isa<PointerType>(T) 2010-02-16 11:11:14 +00:00
PostDominators.cpp
ProfileEstimatorPass.cpp
ProfileInfo.cpp
ProfileInfoLoader.cpp
ProfileInfoLoaderPass.cpp
ProfileVerifierPass.cpp
README.txt
ScalarEvolution.cpp Spelling fixes. 2010-03-01 17:49:51 +00:00
ScalarEvolutionAliasAnalysis.cpp Add a comment. 2010-03-01 17:56:04 +00:00
ScalarEvolutionExpander.cpp Another place where debug info affected codegen. 2010-03-09 01:08:11 +00:00
SparsePropagation.cpp
Trace.cpp
ValueTracking.cpp Move GetStringLength and helper from SimplifyLibCalls to ValueTracking. 2010-03-05 06:58:57 +00:00

Analysis Opportunities:

//===---------------------------------------------------------------------===//

In test/Transforms/LoopStrengthReduce/quadradic-exit-value.ll, the
ScalarEvolution expression for %r is this:

  {1,+,3,+,2}<loop>

Outside the loop, this could be evaluated simply as (%n * %n), however
ScalarEvolution currently evaluates it as

  (-2 + (2 * (trunc i65 (((zext i64 (-2 + %n) to i65) * (zext i64 (-1 + %n) to i65)) /u 2) to i64)) + (3 * %n))

In addition to being much more complicated, it involves i65 arithmetic,
which is very inefficient when expanded into code.

//===---------------------------------------------------------------------===//