[3.6 Release] RC3 has been tagged

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
67 messages Options
1234
Reply | Threaded
Open this post in threaded view
|

[3.6 Release] RC3 has been tagged

Hans Wennborg-2
Hello testers,

Start your engines, RC3 has just been tagged (at r229050 on the
branch). If this one looks good, it will become the release.

There has been quite a bit of activity on the branch since RC2; let's
hope it's all goodness :-)

Please let me know how it looks, and upload binaries to the sftp as usual.

Thanks for all your efforts so far!

 - Hans
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Ben Pope
On Friday, February 13, 2015 11:46 AM, Hans Wennborg wrote:

> Hello testers,
>
> Start your engines, RC3 has just been tagged (at r229050 on the
> branch). If this one looks good, it will become the release.
>
> There has been quite a bit of activity on the branch since RC2; let's
> hope it's all goodness :-)
>
> Please let me know how it looks, and upload binaries to the sftp as usual.
>
> Thanks for all your efforts so far!

All looking good here, uploaded
clang+llvm-3.6.0-rc3-x86_64-linux-gnu-ubuntu-14.04.tar.xz

Ben
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

sebastian.dressler
Hi,

Testing on OS X succeeded and looks good. Uploaded to SFTP, hash: MD5
(clang+llvm-3.6.0-rc3-x86_64-apple-darwin.tar.xz) =
e57b8eb53c10abb1829a7ce5c5113863

Cheers,
Sebastian

2015-02-13 6:14 GMT+01:00 Ben Pope <[hidden email]>:

> On Friday, February 13, 2015 11:46 AM, Hans Wennborg wrote:
>>
>> Hello testers,
>>
>> Start your engines, RC3 has just been tagged (at r229050 on the
>> branch). If this one looks good, it will become the release.
>>
>> There has been quite a bit of activity on the branch since RC2; let's
>> hope it's all goodness :-)
>>
>> Please let me know how it looks, and upload binaries to the sftp as usual.
>>
>> Thanks for all your efforts so far!
>
>
> All looking good here, uploaded
> clang+llvm-3.6.0-rc3-x86_64-linux-gnu-ubuntu-14.04.tar.xz
>
> Ben
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Dimitry Andric
After some manual prodding due to ulimit timeouts, all tests succeeded on FreeBSD 10 too.  I have uploaded these files to SFTP:

SHA256 (clang+llvm-3.6.0-rc3-i386-unknown-freebsd10.tar.xz) = aa62b0252a529d38647e9eac8a74d0e0e677bb27c8e1aff4f66bf64bd9743ecc
SHA256 (clang+llvm-3.6.0-rc3-amd64-unknown-freebsd10.tar.xz) = f0ef8a78576b78bddace6ad92f70f120233e033ae4d833039155095bb5f35d16

-Dimitry

> On 13 Feb 2015, at 14:49, Sebastian Dreßler <[hidden email]> wrote:
>
> Hi,
>
> Testing on OS X succeeded and looks good. Uploaded to SFTP, hash: MD5
> (clang+llvm-3.6.0-rc3-x86_64-apple-darwin.tar.xz) =
> e57b8eb53c10abb1829a7ce5c5113863
>
> Cheers,
> Sebastian
>
> 2015-02-13 6:14 GMT+01:00 Ben Pope <[hidden email]>:
>> On Friday, February 13, 2015 11:46 AM, Hans Wennborg wrote:
>>>
>>> Hello testers,
>>>
>>> Start your engines, RC3 has just been tagged (at r229050 on the
>>> branch). If this one looks good, it will become the release.
>>>
>>> There has been quite a bit of activity on the branch since RC2; let's
>>> hope it's all goodness :-)
>>>
>>> Please let me know how it looks, and upload binaries to the sftp as usual.
>>>
>>> Thanks for all your efforts so far!
>>
>>
>> All looking good here, uploaded
>> clang+llvm-3.6.0-rc3-x86_64-linux-gnu-ubuntu-14.04.tar.xz
>>
>> Ben

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Sylvestre Ledru-6
In reply to this post by Hans Wennborg-2
Hello,

On 13/02/2015 04:46, Hans Wennborg wrote:
> Hello testers,
>
> Start your engines, RC3 has just been tagged (at r229050 on the
> branch). If this one looks good, it will become the release.
>
I uploaded in Debian and it seems to build fine (thanks for landing the
KFreebsd & hurd patches).

However, there are some old regressions (so, they should not block 3.6)
but they can be confusing.
Some OrcJIT tests are failing (GNU/Linux Debian)
http://llvm.org/bugs/show_bug.cgi?id=20586#c3
(I wrote a patch to silent them).

There is also a critical issue on LLDB under GNU/Linux:
http://lists.cs.uiuc.edu/pipermail/lldb-dev/2015-February/006575.html
but it is not new either (but I think it is causing the lldb code
coverage to be super low [1]).
I don't know if it is specific to Debian or my options but it might
affect others.

Cheers,
Sylvestre
[1] http://llvm.org/reports/coverage/


_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Daniel Sanders
In reply to this post by Hans Wennborg-2
Uploaded clang+llvm-3.6.0-rc3-mips-linux-gnu.tar.xz and clang+llvm-3.6.0-rc3-mipsel-linux-gnu.tar.xz.

clang+llvm-3.6.0-rc3-mips-linux-gnu.tar.xz:
    All good.

clang+llvm-3.6.0-rc3-mipsel-linux-gnu.tar.xz:
    Still running test-suite.

clang+llvm-3.6.0-rc3-x86_64-linux-gnu-ubuntu-14.04.tar.xz (cross compiling for Mips):
    All good.
    microMIPS has two expected failures (smallpt and bullet) but microMIPS is new to this release and is considered experimental because of these failures.
________________________________________
From: [hidden email] [[hidden email]] on behalf of Hans Wennborg [[hidden email]]
Sent: 13 February 2015 03:46
To: Ben Pope; Renato Golin; Sylvestre Ledru; Dimitry Andric; Sebastian Dreßler; Daniel Sanders; Nikola Smiljanić
Cc: llvmdev; cfe-dev
Subject: [3.6 Release] RC3 has been tagged

Hello testers,

Start your engines, RC3 has just been tagged (at r229050 on the
branch). If this one looks good, it will become the release.

There has been quite a bit of activity on the branch since RC2; let's
hope it's all goodness :-)

Please let me know how it looks, and upload binaries to the sftp as usual.

Thanks for all your efforts so far!

 - Hans

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Jack Howarth-2
In reply to this post by Hans Wennborg-2
   What are the acceptable performance regressions in the generated
code for a llvm release? We seem to be badly regressed in some
benchmarks (which I first noticed from the review of 3.6-rc1 at
http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1).
This same issue has also been reported in
http://llvm.org/bugs/show_bug.cgi?id=22058. In the case of the 22%
performance degradation in SciMark2's Sparse matmult benchmark, I have
identified both commits that contribute equally to this regression in
http://llvm.org/bugs/show_bug.cgi?id=22589...

Author: mcrosier
Date: Wed Sep 17 09:10:33 2014
New Revision: 217953

URL: http://llvm.org/viewvc/llvm-project?rev=217953&view=rev
Log:
[IndVarSimplify] Widen loop compare instructions.

This improves other optimizations such as LSR.  A sext may be added to the
compare's other operand, but this can often be hoisted outside of the loop.

Added:
    llvm/trunk/test/Transforms/IndVarSimplify/widen-loop-comp.ll
Modified:
    llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp
    llvm/trunk/test/Transforms/IndVarSimplify/no-iv-rewrite.ll
    llvm/trunk/test/Transforms/IndVarSimplify/verify-scev.ll

and

Author: mzolotukhin
Date: Thu Nov 20 14:19:55 2014
New Revision: 222451

URL: http://llvm.org/viewvc/llvm-project?rev=222451&view=rev
Log:
Fix a trip-count overflow issue in LoopUnroll.

Currently LoopUnroll generates a prologue loop before the main loop
body to execute first N%UnrollFactor iterations. Also, this loop is
used if trip-count can overflow - it's determined by a runtime check.

However, we've been mistakenly optimizing this loop to a linear code for
UnrollFactor = 2, not taking into account that it also serves as a safe
version of the loop if its trip-count overflows.

Added:
    llvm/trunk/test/Transforms/LoopUnroll/tripcount-overflow.ll
Modified:
    llvm/trunk/lib/Transforms/Utils/LoopUnrollRuntime.cpp
    llvm/trunk/test/Transforms/LoopUnroll/runtime-loop1.ll

Hopefully these issues can be triaged before the final release Better
a delayed release than one that is badly regressed in performance.
              Jack

On Thu, Feb 12, 2015 at 10:46 PM, Hans Wennborg <[hidden email]> wrote:

> Hello testers,
>
> Start your engines, RC3 has just been tagged (at r229050 on the
> branch). If this one looks good, it will become the release.
>
> There has been quite a bit of activity on the branch since RC2; let's
> hope it's all goodness :-)
>
> Please let me know how it looks, and upload binaries to the sftp as usual.
>
> Thanks for all your efforts so far!
>
>  - Hans
> _______________________________________________
> LLVM Developers mailing list
> [hidden email]         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Renato Golin-2
In reply to this post by Hans Wennborg-2

ARM binaries uploaded, seem to be working. I'll get the AArch64 on Monday, after I reset my board.

Cheers,
Renato

On 13 Feb 2015 03:47, "Hans Wennborg" <[hidden email]> wrote:
Hello testers,

Start your engines, RC3 has just been tagged (at r229050 on the
branch). If this one looks good, it will become the release.

There has been quite a bit of activity on the branch since RC2; let's
hope it's all goodness :-)

Please let me know how it looks, and upload binaries to the sftp as usual.

Thanks for all your efforts so far!

 - Hans

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Nikola Smiljanic
Fedora and OpenSUSE binaries uploaded, all looking good.

On Mon, Feb 16, 2015 at 3:17 AM, Renato Golin <[hidden email]> wrote:

ARM binaries uploaded, seem to be working. I'll get the AArch64 on Monday, after I reset my board.

Cheers,
Renato

On 13 Feb 2015 03:47, "Hans Wennborg" <[hidden email]> wrote:
Hello testers,

Start your engines, RC3 has just been tagged (at r229050 on the
branch). If this one looks good, it will become the release.

There has been quite a bit of activity on the branch since RC2; let's
hope it's all goodness :-)

Please let me know how it looks, and upload binaries to the sftp as usual.

Thanks for all your efforts so far!

 - Hans


_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Renato Golin-2
In reply to this post by Renato Golin-2
On 15 February 2015 at 16:17, Renato Golin <[hidden email]> wrote:
> ARM binaries uploaded, seem to be working. I'll get the AArch64 on Monday,
> after I reset my board.

AArhc64 binaries good. Uploaded.

cheers,
--renato
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Hans Wennborg-2
In reply to this post by Jack Howarth-2
Hi Jack,

On Sun, Feb 15, 2015 at 6:05 AM, Jack Howarth
<[hidden email]> wrote:
>    What are the acceptable performance regressions in the generated
> code for a llvm release? We seem to be badly regressed in some
> benchmarks (which I first noticed from the review of 3.6-rc1 at
> http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1).
> This same issue has also been reported in
> http://llvm.org/bugs/show_bug.cgi?id=22058. In the case of the 22%
> performance degradation in SciMark2's Sparse matmult benchmark, I have
> identified both commits that contribute equally to this regression in
> http://llvm.org/bugs/show_bug.cgi?id=22589...

Thank you very much for trying out the release candidate.

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.

Having said that, we will be doing an rc4, and Hal said that he or
Sanjoy might be able to squeeze in a patch for the issue you pointed
out before that. Otherwise, it will have to wait to 3.6.1 or 3.7.

Thanks,
Hans
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Hans Wennborg-2
In reply to this post by Hans Wennborg-2
On Thu, Feb 12, 2015 at 7:46 PM, Hans Wennborg <[hidden email]> wrote:
> Hello testers,
>
> Start your engines, RC3 has just been tagged (at r229050 on the
> branch). If this one looks good, it will become the release.
>
> There has been quite a bit of activity on the branch since RC2; let's
> hope it's all goodness :-)
>
> Please let me know how it looks, and upload binaries to the sftp as usual.

Windows binary uploaded:

bd2001cf66c04a6fc2722cfeaf7b76c4b9c9c1ff  LLVM-3.6.0-rc3-win32.exe

It was built with the attached script.

 - Hans

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

build_llvm_360._bat_ (3K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Hayden Livingston
In reply to this post by Hans Wennborg-2
In the interest of educating others, can someone explain what the regression is? Is it understood?

I'm new to this area and would love to understand how we identified what is causing this regression? An optimization gone rogue? Some optimization missing, etc.

On Tue, Feb 17, 2015 at 11:12 AM, Hans Wennborg <[hidden email]> wrote:
Hi Jack,

On Sun, Feb 15, 2015 at 6:05 AM, Jack Howarth
<[hidden email]> wrote:
>    What are the acceptable performance regressions in the generated
> code for a llvm release? We seem to be badly regressed in some
> benchmarks (which I first noticed from the review of 3.6-rc1 at
> http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1).
> This same issue has also been reported in
> http://llvm.org/bugs/show_bug.cgi?id=22058. In the case of the 22%
> performance degradation in SciMark2's Sparse matmult benchmark, I have
> identified both commits that contribute equally to this regression in
> http://llvm.org/bugs/show_bug.cgi?id=22589...

Thank you very much for trying out the release candidate.

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.

Having said that, we will be doing an rc4, and Hal said that he or
Sanjoy might be able to squeeze in a patch for the issue you pointed
out before that. Otherwise, it will have to wait to 3.6.1 or 3.7.

Thanks,
Hans
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev


_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Jack Howarth-2
In reply to this post by Hans Wennborg-2
On Tue, Feb 17, 2015 at 2:12 PM, Hans Wennborg <[hidden email]> wrote:

> Hi Jack,
>
> On Sun, Feb 15, 2015 at 6:05 AM, Jack Howarth
> <[hidden email]> wrote:
>>    What are the acceptable performance regressions in the generated
>> code for a llvm release? We seem to be badly regressed in some
>> benchmarks (which I first noticed from the review of 3.6-rc1 at
>> http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1).
>> This same issue has also been reported in
>> http://llvm.org/bugs/show_bug.cgi?id=22058. In the case of the 22%
>> performance degradation in SciMark2's Sparse matmult benchmark, I have
>> identified both commits that contribute equally to this regression in
>> http://llvm.org/bugs/show_bug.cgi?id=22589...
>
> Thank you very much for trying out the release candidate.
>
> 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.
>
> Having said that, we will be doing an rc4, and Hal said that he or
> Sanjoy might be able to squeeze in a patch for the issue you pointed
> out before that. Otherwise, it will have to wait to 3.6.1 or 3.7.
>
> Thanks,
> Hans

Hans,
    We probably ought to open a meta-bug for all of these new
performance regressions introduced in 3.6. The 20% performance
regression observed in the "8 queens puzzle" solver example
(http://llvm.org/bugs/show_bug.cgi?id=22058) doesn't seem to be
introduced by either the r217953 or r222451 commits. I'll try to do a
regression hunt later this week to pin down the offending commit for
that regression. Likewise, the reported regression in the -m32
performance of the sudoku solver example (reported in
http://llvm.org/bugs/show_bug.cgi?id=22589) doesn't match up to those
two commits either (but I suspect a register pressure issue since it
only shows up at 32-bit).
     Jack
ps Do we have a complete set of benchmark results for current 3.6svn
vs 3.5.1 posted anywhere? Hopefully some these are showing up the SPEC
benchmarks and should be added to the meta-bug by someone who has
access to the sources for those benchmarks.
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Jack Howarth-2
In reply to this post by Hayden Livingston
On Tue, Feb 17, 2015 at 2:44 PM, Hayden Livingston
<[hidden email]> wrote:
> In the interest of educating others, can someone explain what the regression
> is? Is it understood?
>
> I'm new to this area and would love to understand how we identified what is
> causing this regression? An optimization gone rogue? Some optimization
> missing, etc.
>

Hayden,
     See...

http://llvm.org/bugs/show_bug.cgi?id=22589#c5
http://llvm.org/bugs/show_bug.cgi?id=22589#c7
http://llvm.org/bugs/show_bug.cgi?id=22589#c9
http://llvm.org/bugs/show_bug.cgi?id=22589#c26
http://llvm.org/bugs/show_bug.cgi?id=22589#c27
http://llvm.org/bugs/show_bug.cgi?id=22589#c33
http://llvm.org/bugs/show_bug.cgi?id=22589#c36

which are the key postings on this bug. I would describe it as an
optimization no longer taken. As for identifying cause of regression,
the process is a brutal walk back through prior svn trunk until the
offending commit which cause the loss of optimization is pin-pointed.
             Jack

> On Tue, Feb 17, 2015 at 11:12 AM, Hans Wennborg <[hidden email]> wrote:
>>
>> Hi Jack,
>>
>> On Sun, Feb 15, 2015 at 6:05 AM, Jack Howarth
>> <[hidden email]> wrote:
>> >    What are the acceptable performance regressions in the generated
>> > code for a llvm release? We seem to be badly regressed in some
>> > benchmarks (which I first noticed from the review of 3.6-rc1 at
>> >
>> > http://www.phoronix.com/scan.php?page=article&item=llvm-clang-3.5-3.6-rc1).
>> > This same issue has also been reported in
>> > http://llvm.org/bugs/show_bug.cgi?id=22058. In the case of the 22%
>> > performance degradation in SciMark2's Sparse matmult benchmark, I have
>> > identified both commits that contribute equally to this regression in
>> > http://llvm.org/bugs/show_bug.cgi?id=22589...
>>
>> Thank you very much for trying out the release candidate.
>>
>> 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.
>>
>> Having said that, we will be doing an rc4, and Hal said that he or
>> Sanjoy might be able to squeeze in a patch for the issue you pointed
>> out before that. Otherwise, it will have to wait to 3.6.1 or 3.7.
>>
>> Thanks,
>> Hans
>> _______________________________________________
>> LLVM Developers mailing list
>> [hidden email]         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Hans Wennborg-2
In reply to this post by Jack Howarth-2
On Tue, Feb 17, 2015 at 11:47 AM, Jack Howarth
<[hidden email]> wrote:
>     We probably ought to open a meta-bug for all of these new
> performance regressions introduced in 3.6.

Feel free to open one. The important thing is whether anyone is
investigating the bugs, though.

> The 20% performance
> regression observed in the "8 queens puzzle" solver example
> (http://llvm.org/bugs/show_bug.cgi?id=22058) doesn't seem to be
> introduced by either the r217953 or r222451 commits. I'll try to do a
> regression hunt later this week to pin down the offending commit for
> that regression. Likewise, the reported regression in the -m32
> performance of the sudoku solver example (reported in
> http://llvm.org/bugs/show_bug.cgi?id=22589) doesn't match up to those
> two commits either (but I suspect a register pressure issue since it
> only shows up at 32-bit).
>      Jack
> ps Do we have a complete set of benchmark results for current 3.6svn
> vs 3.5.1 posted anywhere?

Not that I know. There are test results going into the LNT tool at
http://llvm.org/perf, but I'm not really familiar with that and also
don't think it makes it easy to do comparisons between releases.

 - Hans
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [3.6 Release] RC3 has been tagged

Mehmet Erol Sanliturk


On Tue, Feb 17, 2015 at 4:23 PM, Hans Wennborg <[hidden email]> wrote:
On Tue, Feb 17, 2015 at 11:47 AM, Jack Howarth
<[hidden email]> wrote:
>     We probably ought to open a meta-bug for all of these new
> performance regressions introduced in 3.6.

Feel free to open one. The important thing is whether anyone is
investigating the bugs, though.

> The 20% performance
> regression observed in the "8 queens puzzle" solver example
> (http://llvm.org/bugs/show_bug.cgi?id=22058) doesn't seem to be
> introduced by either the r217953 or r222451 commits. I'll try to do a
> regression hunt later this week to pin down the offending commit for
> that regression. Likewise, the reported regression in the -m32
> performance of the sudoku solver example (reported in
> http://llvm.org/bugs/show_bug.cgi?id=22589) doesn't match up to those
> two commits either (but I suspect a register pressure issue since it
> only shows up at 32-bit).
>      Jack
> ps Do we have a complete set of benchmark results for current 3.6svn
> vs 3.5.1 posted anywhere?

Not that I know. There are test results going into the LNT tool at
http://llvm.org/perf, but I'm not really familiar with that and also
don't think it makes it easy to do comparisons between releases.

 - Hans



Comparison of Clang Test Results :



Assume Clang compiled some programs , the following data is obtained :


Assume  compilation_time for the Program i is obtained on the SAME computer with similar execution environments .

Computers for Program i and Program j may be different .




                 Release 3.5.0         Release 3.6.0
                 ----------------      ---------------
Program 1       compilation_time      compilation_time
 
 .
 .
 .

Program n       compilation_time      compilation_time


Means            Mean_of_Release_3.5.0    Mean_of_Release_3.6.0


You may apply the following tests :


http://en.wikipedia.org/wiki/Levene%27s_test
( If W is significant : Variability of compilation speeds are different between releases )

http://en.wikipedia.org/wiki/Paired_difference_test
( If D is significant : Average compilation time increased ( or decreased )


A similar analysis may be performed on "execution times" of programs :


                 Release 3.5.0         Release 3.6.0
                 ----------------      ---------------
Program 1       execution_time      execution_time
 
 .
 .
 .

Program n       execution_time      execution_time


Means            Mean_of_Release_3.5.0    Mean_of_Release_3.6.0


If you use more than 15 programs , test results will be "good" .
If you use more than 60 programs , test results will be "better" .


The tests will give at least a rough idea about performance changes .


Thank you very much .

Mehmet Erol Sanliturk







_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [cfe-dev] [3.6 Release] RC3 has been tagged

Joerg Sonnenberger
In reply to this post by Hans Wennborg-2
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?

Joerg
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [cfe-dev] [3.6 Release] RC3 has been tagged

Hal Finkel
----- Original Message -----

> From: "Joerg Sonnenberger" <[hidden email]>
> To: [hidden email], "llvmdev" <[hidden email]>
> 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
> [hidden email]         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: [cfe-dev] [3.6 Release] RC3 has been tagged

Jack Howarth-2
On Wed, Feb 18, 2015 at 9:37 AM, Hal Finkel <[hidden email]> wrote:

> ----- Original Message -----
>> From: "Joerg Sonnenberger" <[hidden email]>
>> To: [hidden email], "llvmdev" <[hidden email]>
>> 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
>> [hidden email]         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
> _______________________________________________
> LLVM Developers mailing list
> [hidden email]         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
1234