Rietveld Code Review Tool
Help | Bug tracker | Discussion group | Source code | Sign in
(2032)

Issue 5490054: [google][4.6]Compiler Directed Multiversioning with new -mvarch option

Can't Edit
Can't Publish+Mail
Start Review
Created:
12 years, 4 months ago by Sriraman
Modified:
12 years, 4 months ago
Reviewers:
triegel, davidxl
CC:
gcc-patches_gcc.gnu.org, tejohnson
Base URL:
svn+ssh://gcc.gnu.org/svn/gcc/branches/google/gcc-4_6/gcc/
Visibility:
Public.

Patch Set 1 #

Patch Set 2 : [google][4.6]Compiler Directed Multiversioning with new -mvarch option #

Total comments: 4

Patch Set 3 : [google][4.6]Compiler Directed Multiversioning with new -mvarch option #

Unified diffs Side-by-side diffs Delta from patch set Stats (+1318 lines, -24 lines) Patch
M config/i386/i386.c View 6 chunks +523 lines, -0 lines 0 comments Download
M config/i386/i386.opt View 1 chunk +4 lines, -0 lines 0 comments Download
M doc/tm.texi View 1 chunk +5 lines, -0 lines 0 comments Download
M doc/tm.texi.in View 1 chunk +5 lines, -0 lines 0 comments Download
M mversn-dispatch.c View 1 2 9 chunks +726 lines, -9 lines 0 comments Download
M params.def View 1 chunk +5 lines, -0 lines 0 comments Download
M passes.c View 1 chunk +1 line, -0 lines 0 comments Download
M target.def View 1 chunk +6 lines, -0 lines 0 comments Download
M testsuite/g++.dg/mversn10.C View 1 2 1 chunk +2 lines, -2 lines 0 comments Download
M testsuite/g++.dg/mversn10a.C View 1 2 1 chunk +2 lines, -2 lines 0 comments Download
M testsuite/g++.dg/mversn16.C View 1 2 1 chunk +1 line, -1 line 0 comments Download
M testsuite/g++.dg/mversn8.C View 1 2 1 chunk +4 lines, -4 lines 0 comments Download
M testsuite/g++.dg/tree-prof/mversn15.C View 1 2 1 chunk +1 line, -1 line 0 comments Download
A testsuite/gcc.dg/automversn_1.c View 1 chunk +27 lines, -0 lines 0 comments Download
M testsuite/gcc.dg/mversn2.c View 1 2 1 chunk +5 lines, -5 lines 0 comments Download
M tree-pass.h View 1 chunk +1 line, -0 lines 0 comments Download

Messages

Total messages: 8
Sriraman
Hi, I am working on user-directed and compiler-directed function multiversioning which has been discussed in ...
12 years, 4 months ago (2011-12-16 06:53:54 UTC) #1
triegel_redhat.com
For Transactional Memory support, we also create versions of functions (see code in trunk, e.g., ...
12 years, 4 months ago (2011-12-16 11:35:34 UTC) #2
davidxl
On Fri, Dec 16, 2011 at 3:35 AM, Torvald Riegel <triegel@redhat.com> wrote: > For Transactional ...
12 years, 4 months ago (2011-12-16 19:05:13 UTC) #3
davidxl
http://codereview.appspot.com/5490054/diff/1011/config/i386/i386.c File config/i386/i386.c (right): http://codereview.appspot.com/5490054/diff/1011/config/i386/i386.c#newcode26569 config/i386/i386.c:26569: +mversion_for_core2 (tree *optimization_node, -> mversionable_for_core2_p ? http://codereview.appspot.com/5490054/diff/1011/mversn-dispatch.c File mversn-dispatch.c ...
12 years, 4 months ago (2011-12-16 19:39:47 UTC) #4
Sriraman
I have uploaded a new patch set with all the mentioned changes made. If a ...
12 years, 4 months ago (2011-12-16 22:05:53 UTC) #5
triegel_redhat.com
On Fri, 2011-12-16 at 11:05 -0800, Xinliang David Li wrote: > On Fri, Dec 16, ...
12 years, 4 months ago (2011-12-16 22:11:42 UTC) #6
davidxl
ok for google branches. David On Fri, Dec 16, 2011 at 2:05 PM, <tmsriram@google.com> wrote: ...
12 years, 4 months ago (2011-12-17 08:25:54 UTC) #7
Sriraman
12 years, 4 months ago (2011-12-18 00:14:51 UTC) #8
Committed to google 4_6 branch.

Thanks,
-Sri.

On Sat, Dec 17, 2011 at 12:25 AM, Xinliang David Li <davidxl@google.com> wrote:
> ok for google branches.
>
> David
>
> On Fri, Dec 16, 2011 at 2:05 PM,  <tmsriram@google.com> wrote:
>> I have uploaded a new patch set with all the mentioned changes made. If
>> a function has the target attribute it will not be touched by the
>> autoclone pass.
>>
>> Also fixed some test cases which were broken because the clone names
>> used '_' instead of '.' for suffixing.
>>
>>
>> On 2011/12/16 19:39:47, davidxl wrote:
>>>
>>> http://codereview.appspot.com/5490054/diff/1011/config/i386/i386.c
>>> File config/i386/i386.c (right):
>>
>>
>>
>>
http://codereview.appspot.com/5490054/diff/1011/config/i386/i386.c#newcode26569
>>>
>>> config/i386/i386.c:26569: +mversion_for_core2 (tree
>>
>> *optimization_node,
>>>
>>> -> mversionable_for_core2_p ?
>>
>>
>>> http://codereview.appspot.com/5490054/diff/1011/mversn-dispatch.c
>>> File mversn-dispatch.c (right):
>>
>>
>>
>> http://codereview.appspot.com/5490054/diff/1011/mversn-dispatch.c#newcode931
>>>
>>> mversn-dispatch.c:931: DECL_STATIC_DESTRUCTOR (new_decl) = 0;
>>> Should you assert it instead? Should not clone ctor/dtors.
>>
>>
>>
>> http://codereview.appspot.com/5490054/diff/1011/mversn-dispatch.c#newcode2221
>>>
>>> mversn-dispatch.c:2221: VEC_truncate (edge, EXIT_BLOCK_PTR->preds, 0);
>>> {} --> remove
>>
>>
>>
>> http://codereview.appspot.com/5490054/diff/1011/mversn-dispatch.c#newcode2389
>>>
>>> mversn-dispatch.c:2389:
>>> How does it interact with manual multi-versioning from user? You
>>
>> probably don't
>>>
>>> want to clone functions that are marked with target attributes
>>
>> (explicitly --
>>>
>>> not implied from command line).
>>
>>
>>
>>
>> http://codereview.appspot.com/5490054/
Sign in to reply to this message.

Powered by Google App Engine
RSS Feeds Recent Issues | This issue
This is Rietveld f62528b