Bob Wilson 3992feb075 Adjust the heuristics used to decide when SROA is likely to be profitable. ... The SRThreshold value makes perfect sense for checking if an entire aggregate should be promoted to a scalar integer, but it is not so good for splitting an aggregate into its separate elements. A struct may contain a large embedded array along with some scalar fields that would benefit from being split apart by SROA. Even if the total aggregate size is large, it may still be good to perform SROA. Thus, the most important piece of this patch is simply moving the aggregate size comparison vs. SRThreshold so that it guards only the aggregate promotion. We have also been checking the number of elements to decide if an aggregate should be split up. The limit of "SRThreshold/4" seemed rather arbitrary, and I don't think it's very useful to derive this limit from SRThreshold anyway. I've collected some data showing that the current default limit of 32 (since SRThreshold defaults to 128) is a reasonable cutoff for struct types. One thing suggested by the data is that distinguishing between structs and arrays might be useful. There are (obviously) a lot more large arrays than large structs (as measured by the number of elements and not the total size -- a large array inside a struct still counts as a single element given the way we do SROA right now). Out of 8377 arrays where we successfully performed SROA while compiling a large set of benchmarks, only 16 of them had more than 8 elements. And, for those 16 arrays, it's not at all clear that SROA was actually beneficial. So, to offset the compile time cost of investigating more large structs for SROA, the patch lowers the limit on array elements to 8. This fixes Apple Radar 7563690. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@95224 91177308-0d34-0410-b5e6-96231b3b80d8		2010-02-03 17:23:56 +00:00
..
ABCD.cpp	Change errs() to dbgs().	2010-01-05 01:27:39 +00:00
ADCE.cpp	Use do+while instead of while for loops which obviously have a	2010-01-05 16:27:25 +00:00
BasicBlockPlacement.cpp	eliminate VISIBILITY_HIDDEN from Transforms/Scalar. PR4861	2009-09-02 06:11:42 +00:00
CMakeLists.txt	move instcombine to its own library, it's past time.	2010-01-04 06:23:24 +00:00
CodeGenPrepare.cpp	When doing address-mode sinking, expand the base register first, rather	2010-01-19 22:45:06 +00:00
ConstantProp.cpp	remove a bunch of extraneous LLVMContext arguments	2009-11-06 04:27:31 +00:00
DCE.cpp	eliminate VISIBILITY_HIDDEN from Transforms/Scalar. PR4861	2009-09-02 06:11:42 +00:00
DeadStoreElimination.cpp	Use do+while instead of while for loops which obviously have a	2010-01-05 16:27:25 +00:00
GEPSplitter.cpp	Check in the experimental GEP splitter pass. This pass splits complex	2009-10-26 19:12:14 +00:00
GVN.cpp	Fix some comment typos.	2010-02-03 00:33:21 +00:00
IndVarSimplify.cpp	Re-implement the main strength-reduction portion of LoopStrengthReduction.	2010-01-21 02:09:26 +00:00
JumpThreading.cpp	third bug from PR6119: the xor dupe extension allows	2010-01-23 19:21:31 +00:00
LICM.cpp	Change errs() to dbgs().	2010-01-05 01:27:30 +00:00
LoopDeletion.cpp	LoopDeletion depends on loops having dedicated exits.	2009-11-05 21:47:04 +00:00
LoopIndexSplit.cpp	Use do+while instead of while for loops which obviously have a	2010-01-05 16:27:25 +00:00
LoopRotation.cpp	Update various Loop optimization passes to cope with the possibility that	2009-11-05 21:11:53 +00:00
LoopStrengthReduce.cpp	Generic reformatting and comment fixing. No functionality change.	2010-01-29 00:52:43 +00:00
LoopUnrollPass.cpp	Change errs() to dbgs().	2010-01-05 01:27:44 +00:00
LoopUnswitch.cpp	fix a crash in loop unswitch on a loop invariant vector condition.	2010-02-02 02:26:54 +00:00
Makefile	make -fno-rtti the default unless a directory builds with REQUIRES_RTTI.	2010-01-24 20:43:08 +00:00
MemCpyOptimizer.cpp	Convert a ton of simple integer type equality tests to the new predicate.	2010-01-05 20:07:06 +00:00
Reassociate.cpp	only factor from expressions whose uses are empty and whose	2010-01-09 06:01:36 +00:00
Reg2Mem.cpp	clean up this code a bit.	2009-09-02 06:15:37 +00:00
Scalar.cpp	remove the now dead condprop pass, PR3906.	2009-11-11 05:56:35 +00:00
ScalarReplAggregates.cpp	Adjust the heuristics used to decide when SROA is likely to be profitable.	2010-02-03 17:23:56 +00:00
SCCP.cpp	Change errs() to dbgs().	2010-01-05 01:27:15 +00:00
SCCVN.cpp	Use do+while instead of while for loops which obviously have a	2010-01-05 16:27:25 +00:00
SimplifyCFGPass.cpp	Use do+while instead of while for loops which obviously have a	2010-01-05 16:27:25 +00:00
SimplifyHalfPowrLibCalls.cpp	improve portability to avoid conflicting with std::next in c++'0x.	2009-12-03 00:50:42 +00:00
SimplifyLibCalls.cpp	Recommit this, looks like it wasn't the cause.	2010-02-03 00:21:58 +00:00
TailDuplication.cpp	Change errs() to dbgs().	2010-01-05 01:27:33 +00:00
TailRecursionElimination.cpp	Revert 94937 and move the noreturn check to codegen.	2010-02-03 03:55:59 +00:00