[llvm-dev] [GlobalISel][AArch64] Toward flipping the switch for O0: Please give it a try!

Quentin Colombet via llvm-dev llvm-dev at lists.llvm.org
Thu May 11 09:40:15 PDT 2017


Hi Diana,

Thanks for the summary.

> On May 11, 2017, at 4:01 AM, Diana Picus <diana.picus at linaro.org> wrote:
> 
> Hi all,
> 
> I'm still running some validation on this, I'll send an email when
> it's done. If that goes well I don't have anything against making the
> switch.
> 
> For the record, here's a summary of issues that were deferred for
> later on (some of which are optimization-ish and we might decide to
> never do at -O0 at all):
> * Crash in RegBankSelect for half fp types:
> https://bugs.llvm.org/show_bug.cgi?id=32560 <https://bugs.llvm.org/show_bug.cgi?id=32560>

I’ll have a look.

> * Improving constant placement: http://bugs.llvm.org/show_bug.cgi?id=32561 <http://bugs.llvm.org/show_bug.cgi?id=32561>

I’ve commented in the PR to mention the localizer technic I was playing with, if someone wants to give it a try.

> * Fancy switch lowering
> * Transforming division-by-constant-power-of-2 into right shift
> 
> AFAICT all the other issues that were brought up were fixed (yay!).
> 
> Cheers,
> Diana

Cheers,
-Quentin
> 
> 
> On 11 May 2017 at 08:44, Kristof Beyls via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
>> 
>> On 10 May 2017, at 17:36, Quentin Colombet <qcolombet at apple.com> wrote:
>> 
>> 
>> MultiSource/Benchmarks/mediabench/mpeg2/mpeg2dec/mpeg2decode (46%): Function
>> Reference_IDCT: Probably due to creating all constants in the entry BB +
>> spilling floating point data through an X register:
>> 
>> FastISel:
>> fadd d0, d1, d0
>> str d0, [sp,#528]
>> GlobalISel:
>> fadd d0, d1, d0
>> fmov x9, d0
>> stur x9, [x29,#-48]
>> 
>> 
>> Good finding, I forgot to do stores in my previous fix. I’ll do them
>> shortly.
>> 
>> 
>> Should be fixed by r302679
>> 
>> 
>> Thanks Quentin,
>> 
>> That reduces the slow-down when enabling globalisel at -O0 from 13% (on
>> r302453)  to 9.5% (on r302679) in my experiments.
>> The code size increase also reduces from just over 3% to 2.8%.
>> 
>> Kristof
>> 
>> 
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170511/d4b409be/attachment.html>


More information about the llvm-dev mailing list