Re: [boost] Default x86 instruction set

Subject: Re: [boost] Default x86 instruction set
From: Jonathan Wakely (jwakely.boost_at_[hidden])
Date: 2013-05-07 13:06:55

On 7 May 2013 17:59, Andrey Semashev wrote:
> On Tuesday 07 May 2013 17:34:17 Jonathan Wakely wrote:
>> On 7 May 2013 17:17, Andrey Semashev wrote:
>> > Noone uses pure 386 these days, so the default in gcc.jam looks
>> > inadequate.
>>
>> I would have chosen a different word :)
>>
>> It's even been suggested to deprecate 80386 support in GCC, see
>> http://gcc.gnu.org/ml/gcc/2012-12/msg00079.html
>>
>> It hasn't happened yet, but if/when it does the -march=i386 switch
>> will look even more silly.
>
> Thanks for pointing to that discussion. Turns out 386 actually didn't support
> atomic ops. :)

Right, that's why GCC's std::string and std::shared_ptr (and
tr1::shared_ptr) use a mutex when compiled for i386.  You really don't
want that if you're running on 486 or better, and *everyone* is
running on better.