[llvm] r199015 - LoopVectorizer: Enable strided memory accesses versioning per default

[Arnold]

One benchmark in TSVC (kernels that test vectorization) and spec/dealII.

Looking into dealII it is most likely noise. The functions which we now vectorize don't show up on a timed profile.

Sent from my iPhone

> On Jan 11, 2014, at 12:52 PM, Eric Christopher <echristo at gmail.com> wrote:
> 
> Did you see any benefits?
> 
> -eric
> 
> On Sat, Jan 11, 2014 at 12:40 PM, Arnold Schwaighofer
> <aschwaighofer at apple.com> wrote:
>> Author: arnolds
>> Date: Sat Jan 11 14:40:34 2014
>> New Revision: 199015
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=199015&view=rev
>> Log:
>> LoopVectorizer: Enable strided memory accesses versioning per default
>> 
>> I saw no compile or execution time regressions on x86_64 -mavx -O3.
>> 
>> radar://13075509
>> 
>> Modified:
>>    llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
>> 
>> Modified: llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp?rev=199015&r1=199014&r2=199015&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp (original)
>> +++ llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp Sat Jan 11 14:40:34 2014
>> @@ -126,7 +126,7 @@ TinyTripCountVectorThreshold("vectorizer
>> ///    } else
>> ///      ...
>> static cl::opt<bool> EnableMemAccessVersioning(
>> -    "enable-mem-access-versioning", cl::init(false), cl::Hidden,
>> +    "enable-mem-access-versioning", cl::init(true), cl::Hidden,
>>     cl::desc("Enable symblic stride memory access versioning"));
>> 
>> /// We don't unroll loops with a known constant trip count below this number.
>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits