This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [autovect] If-conversion related patches
- From: Dorit Naishlos <DORIT at il dot ibm dot com>
- To: Devang Patel <dpatel at apple dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 15 Nov 2004 21:50:59 +0200
- Subject: Re: [autovect] If-conversion related patches
> I wanted to ask: Do I remove that loop OR Do I update test to expect
> one vectorized loop ?
Lets go with the second option. (by the way - it will be the second loop
vectorized in this testcase).
dorit
Devang Patel
<dpatel@apple.com> To: Dorit Naishlos/Haifa/IBM@IBMIL
Sent by: cc: GCC Patches <gcc-patches@gcc.gnu.org>
gcc-patches-owner@ Subject: Re: [autovect] If-conversion related patches
gcc.gnu.org
15/11/2004 21:03
On Nov 14, 2004, at 2:05 PM, Dorit Naishlos wrote:
>> 2004-11-12 Devang Patel <dpatel@apple.com>
>>
>> * gcc.dg/vect/vect-20040915-1.c: New test.
>> * gcc.dg/vect/vect-20040915-2.c: New test.
>> * gcc.dg/vect/vect-20040915-3.c: New test.
>> * gcc.dg/vect/vect-20040915-4.c: New test.
>> * gcc.dg/vect/vect-20040915-5.c: New test.
>> * gcc.dg/vect/vect-20040915-6.c: New test.
>> * gcc.dg/vect/vect-20040915-7.c: New test.
>
> I guess we should enable these tests on all relevant platforms, and
> xfail
> on those that don't support conditional vector code yet.
Right now only powerpc has support for conditional vector. I will xfail
these tests on other platforms.
> Also, one of the loops in the testcase vect-none.c now gets vectorized
> (the
> loop in line 145 - it has a condition), so we have a failure now cause
> we
> don't expect to vectorize it.
I wanted to ask: Do I remove that loop OR Do I update test to expect
one vectorized loop ?
> Lastly - how about naming these tests vect-ifcvt-[1,2,3,4,5,6].c
> instead?
fine with me :)
-
Devang