This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: instruction ordering
- From: Brian Budge <brian dot budge at gmail dot com>
- To: tprince at computer dot org
- Cc: "John (Eljay) Love-Jensen" <eljay at adobe dot com>, Segher Boessenkool <segher at kernel dot crashing dot org>, "marian at jozep dot com dot au" <marian at jozep dot com dot au>, GCC-help <gcc-help at gcc dot gnu dot org>
- Date: Thu, 7 May 2009 11:06:51 -0700
- Subject: Re: instruction ordering
- References: <C6285F4A.3AEAC%eljay@adobe.com> <4A032196.6000902@sbcglobal.net>
This may be a silly point to bring up, but it hasn't been mentioned
whether the code was compiled with optimizations, and what level. My
understanding is that -march only specifies which instructions are
allowed to be emitted by the compiler. You can compile with
-march=native -O0 and still have slow code.
Brian
On Thu, May 7, 2009 at 10:59 AM, Tim Prince <TimothyPrince@sbcglobal.net> wrote:
> John (Eljay) Love-Jensen wrote:
>>>
>>> You mean -mcpu= .
>>>
>>
>> I thought -mcpu was a deprecated synonym for -mtune.
>>
>> Whereas -march implies -mtune, but is not a synonym for -mtune.
>>
>> My understanding may be out of date.
>>
>>
>
> I wouldn't have much hope for gcc if an understanding more up to date than
> Eljay's were needed.
>