public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Uros Bizjak <ubizjak@gmail.com>
To: Mingjie Xing <mingjie.xing@gmail.com>
Cc: rdsandiford@googlemail.com, richard.guenther@gmail.com,
		gcc-patches@gcc.gnu.org
Subject: Re: [vect] Ask for review and approving the patch about vect and loongson
Date: Tue, 17 Aug 2010 09:44:00 -0000	[thread overview]
Message-ID: <AANLkTimtUwrwJ75inYTjPF5kHaR6_=8Q76Ot5ZE25Xi+@mail.gmail.com> (raw)
In-Reply-To: <AANLkTinXMM1GispaJ-AWu_Fadk8Q2oTkbCgYknfoLSFW@mail.gmail.com>

On Tue, Aug 17, 2010 at 8:37 AM, Mingjie Xing <mingjie.xing@gmail.com> wrote:

>> There are various copies of vect.exp scattered in the testsuite:
>>
>> $ find . -name vect.exp
>> ./gcc.dg/vect/vect.exp
>> ./gfortran.dg/vect/vect.exp
>> ./g++.dg/vect/vect.exp
>>
>> You will need to change them all. Bonus points if you merge all these
>> files into globally accessible support library.
>>
>> Uros.
>>
>
> Thanks for point it out. I've extracted the common piece of code into
> a function. See the attachment. Tested on Intel Pentium.
>
> One problem is that there is a little difference among these vect.exp's,
>
> ==========gcc.dg/vect/vect.exp===========
> } elseif [is-effective-target arm_neon_ok] {
>    eval lappend DEFAULT_VECTCFLAGS [add_options_for_arm_neon ""]
>    # NEON does not support denormals, so is not used for vectorization by
>    # default to avoid loss of precision.  We must pass -ffast-math to test
>    # vectorization of float operations.
>    lappend DEFAULT_VECTCFLAGS "-ffast-math"
>    if [is-effective-target arm_neon_hw] {
>      set dg-do-what-default run
>    } else {
>      set dg-do-what-default compile
>    }
> } else {
>
> ===========g++.dg/vect/vect.exp, gfortran.dg/vect/vect.exp=========
> } elseif [is-effective-target arm_neon_ok] {
>    eval lappend DEFAULT_VECTCFLAGS [add_options_for_arm_neon ""]
>    if [is-effective-target arm_neon_hw] {
>      set dg-do-what-default run
>    } else {
>      set dg-do-what-default compile
>    }
> } else {
>
> I'm not sure if this is intentional and how to handle this in that case.

The difference is not intentional, it looks that someone changed only
gcc.dg/vect/vect.exp. So, in case of doubt, simply copy the
functionality from gcc.dg/vect, this should be the master copy for
your function.

+# Check whether the vect tests are supported by the target and set additional
+# target-dependent vector flags, which can be overriden by using dg-options
+# in individual tests. Return 1 if the target supports, 0 otherwise.

... whether the vectorizer tests are supported ... Return 1 if
vectorizer tests are supported by target, 0 otherwise.

Thanks,
Uros.

  reply	other threads:[~2010-08-17  9:36 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-08-13 11:37 Uros Bizjak
2010-08-17  7:01 ` Mingjie Xing
2010-08-17  9:44   ` Uros Bizjak [this message]
2010-08-17  9:49     ` Mingjie Xing
2010-08-17 10:53       ` Uros Bizjak
2010-08-19  1:45         ` Mingjie Xing
2010-08-21  8:59         ` Richard Sandiford
2010-08-23 16:50           ` Richard Guenther
2010-08-23 18:51             ` Richard Sandiford
2010-08-24  5:06               ` Mingjie Xing
  -- strict thread matches above, loose matches on Subject: below --
2010-08-13  6:38 Mingjie Xing
2010-08-14 20:02 ` Richard Sandiford
2010-07-09  2:11 Eric Fisher
2010-07-09  8:14 ` Richard Guenther
2010-07-09  8:40   ` Eric Fisher
2010-07-09  9:35     ` Richard Guenther
2010-07-09 10:45       ` Eric Fisher
2010-07-10 14:46     ` Richard Sandiford
2010-07-12  9:50       ` Eric Fisher
2010-07-12 19:21         ` Richard Sandiford
2010-07-13  3:55           ` Eric Fisher
2010-07-13 19:43             ` Richard Sandiford
2010-07-15  6:18               ` Eric Fisher

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='AANLkTimtUwrwJ75inYTjPF5kHaR6_=8Q76Ot5ZE25Xi+@mail.gmail.com' \
    --to=ubizjak@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=mingjie.xing@gmail.com \
    --cc=rdsandiford@googlemail.com \
    --cc=richard.guenther@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).