This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: opt-info message change for vectorizer
- From: Xinliang David Li <davidxl at google dot com>
- To: Teresa Johnson <tejohnson at google dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, Richard Guenther <richard dot guenther at gmail dot com>
- Date: Tue, 27 Aug 2013 11:40:59 -0700
- Subject: Re: opt-info message change for vectorizer
- Authentication-results: sourceware.org; auth=none
- References: <CAAkRFZK60Yer3SOds_=KbXzvqfz3EBfcxLFJEZdpKVSXiVFMvw at mail dot gmail dot com> <CAAkRFZ+JC55X5YhLC30XwUrVeDWhM-ff4WC5arwYiv2t3yQZzQ at mail dot gmail dot com> <CAAe5K+XPXipPX-0_rnLzDMqnxQsA74gWBHg=+Sy1MLwXoY9p8Q at mail dot gmail dot com>
yes -- the long unmangled names can be annoying -- that is why I chose
to dump the short form of the function names -- combined with line
numbers, it should be enough to get the full context.
David
On Tue, Aug 27, 2013 at 11:36 AM, Teresa Johnson <tejohnson@google.com> wrote:
> My only concern is whether the dump messages will get too long with
> the full function name on the same line. The infrastructure that emits
> inform() notes ensures that the function name is printed before each
> block of messages related to that function (via an "In function foo:"
> type message), but I had found before that the new dump infrastructure
> doesn't do that. OTOH, your approach will make it much easier to grep
> the output of a large build. Personally, I use grep on this type of
> output enough to make the longer lines worth it. Either that or the
> new dump infrastructure needs to be fixed to emit the function name
> before each block of messages, a la inform().
>
> Thanks,
> Teresa
>
> On Tue, Aug 27, 2013 at 11:22 AM, Xinliang David Li <davidxl@google.com> wrote:
>> Does this one look ok?
>>
>> thanks,
>>
>> David
>>
>> On Thu, Aug 22, 2013 at 4:20 PM, Xinliang David Li <davidxl@google.com> wrote:
>>> Hi, In this patch, loop alignment peeling and loop versioning
>>> transformation will be reported via -fopt-info by default. This will
>>> help vectorizer size tuning.
>>>
>>> It also enhances the opt-info dump to include current function name as
>>> context. Otherwise, we may see duplicate messeages from inline/cloned
>>> instances.
>>>
>>> An example opt report:
>>>
>>>
>>>
>>> b.c:16:A::foo: note: Loop is vectorized
>>>
>>> b.c:16:A::foo: note: Loop is versioned to remove aliases for vectorization
>>>
>>> b.c:16:A::foo: note: Loop is peeled to enhance alignment for vectorization
>>>
>>> b.c:16:A::foo: note: Completely unroll loop 6 times
>>>
>>> b.c:12:A::foo: note: Completely unroll loop 6 times
>>>
>>>
>>> Ok after testing?
>>>
>>> thanks,
>>>
>>> David
>
>
>
> --
> Teresa Johnson | Software Engineer | tejohnson@google.com | 408-460-2413