llvm dev - Feb 2015 - [LLVMdev] [cfe-dev] [3.6 Release] RC3 has been tagged

----- Original Message -----
> From: "Joerg Sonnenberger" <joerg at britannica.bec.de>
> To: cfe-dev at cs.uiuc.edu, "llvmdev" <llvmdev at
cs.uiuc.edu>
> Sent: Wednesday, February 18, 2015 7:42:11 AM
> Subject: Re: [LLVMdev] [cfe-dev]  [3.6 Release] RC3 has been tagged
> 
> On Tue, Feb 17, 2015 at 11:12:20AM -0800, Hans Wennborg wrote:
> > I asked a few of the other developers, and the consensus was that
> > while unfortunate, we won't block the release on a perf regression
> > like this, at least not at this stage in the release process.
> 
> Wasn't the result of the discussion much stronger -- the performance
> regression seems to be more a result of hypersensitivity to specific
> code generation choices and less a fundamental issue?
There are several commits involved, and I don't believe we understand
whether the issues are all hypersensitivity, or some other more systematic
interactions. There is one issue, which involved a correctness fix for runtime
unrolling that was done in a sub-optimal way (causing it to introduce an
unnecessary comparison) that I think is well understood, and for which we have a
fix (http://reviews.llvm.org/D7715). We should pick up this in the release. I
think that, unfortunately, for the others, there's not sufficient time to
investigate before the release.

 -Hal
> 
> Joerg
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

On Wed, Feb 18, 2015 at 9:37 AM, Hal Finkel <hfinkel at anl.gov>
wrote:
> ----- Original Message -----
>> From: "Joerg Sonnenberger" <joerg at britannica.bec.de>
>> To: cfe-dev at cs.uiuc.edu, "llvmdev" <llvmdev at
cs.uiuc.edu>
>> Sent: Wednesday, February 18, 2015 7:42:11 AM
>> Subject: Re: [LLVMdev] [cfe-dev]  [3.6 Release] RC3 has been tagged
>>
>> On Tue, Feb 17, 2015 at 11:12:20AM -0800, Hans Wennborg wrote:
>> > I asked a few of the other developers, and the consensus was that
>> > while unfortunate, we won't block the release on a perf
regression
>> > like this, at least not at this stage in the release process.
>>
>> Wasn't the result of the discussion much stronger -- the
performance
>> regression seems to be more a result of hypersensitivity to specific
>> code generation choices and less a fundamental issue?
>
> There are several commits involved, and I don't believe we understand
whether the issues are all hypersensitivity, or some other more systematic
interactions. There is one issue, which involved a correctness fix for runtime
unrolling that was done in a sub-optimal way (causing it to introduce an
unnecessary comparison) that I think is well understood, and for which we have a
fix (http://reviews.llvm.org/D7715). We should pick up this in the release. I
think that, unfortunately, for the others, there's not sufficient time to
investigate before the release.
>
>  -Hal
Hal,
     In my testing, we only had two commits that triggered the 22%
performance regression (r217953 and r222451) with each contributing to
half of the performance lost. The proposed fix in
http://reviews.llvm.org/D7715 entirely eliminates the regression due
to  r222451. Committing this fix will at least mitigate the regression
down to 11%.
             Jack
>
>>
>> Joerg
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

On 18 February 2015 at 14:37, Hal Finkel <hfinkel at anl.gov>
wrote:
> I think that, unfortunately, for the others, there's not sufficient
time to investigate before the release.
This looks like a serious case for 3.6.1, not RC5.

cheers,
--renato

I finally got around to testing this on a Bloomfield processor (Early
2009 MacPro 2x2.66 GHz dual-quad core) and the regressions from
http://llvm.org/bugs/show_bug.cgi?id=22589 are even more severe. For
10 runs of scimark2_1c built with "-O3 -march=native"...

llvm 3.5.1   1204.16+/-2.66 Mflops
3.6 branch   866.49+/-1.26 Mflops

Do you seriously want to ship with a 39% performance regression in a
major benchmark?
              Jack

On Wed, Feb 18, 2015 at 10:33 AM, Renato Golin <renato.golin at
linaro.org> wrote:
> On 18 February 2015 at 14:37, Hal Finkel <hfinkel at anl.gov> wrote:
>> I think that, unfortunately, for the others, there's not sufficient
time to investigate before the release.
>
> This looks like a serious case for 3.6.1, not RC5.
>
> cheers,
> --renato
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev