Original message:
Performance optimizations:
- SwitchInst: case values stored separately from Operands List. It allows to make faster access to individual case value numbers or ranges.
- Optimized IntItem, added APInt value caching.
- Optimized IntegersSubsetGeneric: added optimizations for cases when subset is single number or when subset consists from single numbers only.
llvm-svn: 158997
fail. Original commit message:
Performance optimizations:
- SwitchInst: case values stored separately from Operands List. It allows to make faster access to individual case value numbers or ranges.
- Optimized IntItem, added APInt value caching.
- Optimized IntegersSubsetGeneric: added optimizations for cases when subset is single number or when subset consists from single numbers only.
On my machine these optimizations gave about 4-6% of compile-time improvement.
llvm-svn: 158986
* Primarily fixed \param commands with names not matching any actual
parameters of the documented functions. In many cases this consists
just of fixing up the parameter name in the \param to match the code,
in some it means deleting obsolete documentation and occasionally it
means documenting the parameter that has replaced the older one that
was documented, which sometimes means some simple reverse-engineering
of the docs from the implementation;
* Fixed \param ParamName [out] to the correct format with [out] before
the parameter name;
* Fixed some \brief summaries.
llvm-svn: 158980
- SwitchInst: case values stored separately from Operands List. It allows to make faster access to individual case value numbers or ranges.
- Optimized IntItem, added APInt value caching.
- Optimized IntegersSubsetGeneric: added optimizations for cases when subset is single number or when subset consists from single numbers only.
On my machine these optimizations gave about 4-6% of compile-time improvement.
llvm-svn: 158979
* Added \file documentation for PPCallbacks.h;
* Added/formated \brief summaries;
* Deleted documentation for parameters that no longer exist;
* Used \param more systematically for documentation of parameters;
* Escaped # characters in Doxygen comments.
llvm-svn: 158978
* Add \file documentation;
* Add \verbatim...\endverbatim markup as needed;
* Add \brief summaries;
* Escaped "::" in Doxygen comments when preceded by space, to avoid a
Doxygen warning where Doxygen takes this as an explicit link request;
* Add \code...\endcode markup to code examples;
* Fix a grammatical glitch in "is this declarator is a".
llvm-svn: 158977
* Escaped # and < characters in Doxygen comments as needed;
* Fixed up some \brief summaries;
* Marked up some parameter references with \p;
* Added \code...\endcode around code examples;
* Used \returns a little more.
llvm-svn: 158966
* Made \brief documentation be brief, mostly by adding a blank line to make
the rest of the text be part of the detailed description only;
* Removed "FunctionOrClassName - " and other redundant text from the start
of Doxygen comments.
llvm-svn: 158964
This makes it explicit when ScoreboardHazardRecognizer will be used.
"GenericItineraries" would only make sense if it contained real
itinerary values and still required ScoreboardHazardRecognizer.
llvm-svn: 158963
The code in X86TargetLowering::LowerEH_RETURN() assumes that a frame
pointer exists, but the frame pointer was forced by the presence of
llvm.eh.unwind.init which isn't guaranteed.
If llvm.eh.unwind.init is actually required in functions calling
eh.return (is it?), we should diagnose that instead of emitting bad
machine code.
This should fix the dragonegg-x86_64-linux-gcc-4.6-test bot.
llvm-svn: 158961
This is a minor drive-by fix with no robust way to unit test.
As an example see neon-div.ll:
SU(16): %Q8<def> = VMOVLsv4i32 %D17, pred:14, pred:%noreg, %Q8<imp-use,kill>
val SU(1): Latency=2 Reg=%Q8
...should be latency=1
llvm-svn: 158960
Minor drive by fix to cleanup latency computation. Calling
getOperandLatency with a deliberately incorrect operand index does not
give you the latency you want.
llvm-svn: 158959
boolean flag to an enum: { Fast, Standard, Strict } (default = Standard).
This option controls the creation by optimizations of fused FP ops that store
intermediate results in higher precision than IEEE allows (E.g. FMAs). The
behavior of this option is intended to match the behaviour specified by a
soon-to-be-introduced frontend flag: '-ffuse-fp-ops'.
Fast mode - allows formation of fused FP ops whenever they're profitable.
Standard mode - allow fusion only for 'blessed' FP ops. At present the only
blessed op is the fmuladd intrinsic. In the future more blessed ops may be
added.
Strict mode - allow fusion only if/when it can be proven that the excess
precision won't effect the result.
Note: This option only controls formation of fused ops by the optimizers. Fused
operations that are explicitly requested (e.g. FMA via the llvm.fma.* intrinsic)
will always be honored, regardless of the value of this option.
Internally TargetOptions::AllowExcessFPPrecision has been replaced by
TargetOptions::AllowFPOpFusion.
llvm-svn: 158956
Kostya Serebryany