[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: i686 ---> i586, i386 ---> i586, x86_64 stays the same

On Mon, Apr 06, 2009 at 11:30:47AM +0100, psmith wrote:
> Will Woods wrote:
>> On Sun, 2009-04-05 at 23:01 +0100, psmith wrote:
>>> i completely understand the reasoning to move i365 up to i586, but 
>>> why on earth restrict i686 capable processors to the i586 instruction 
>>> sets?
>> Again: the only difference between i586 and i686 is the CMOV family of
>> instructions, which are generally *slower* than the i586 version.
>> You can read Linus' detailed explanation here:
>> http://ondioline.org/mail/cmov-a-bad-idea-on-out-of-order-cpus
>> Building for i686 gives *no* real performance benefit[1], but breaks
>> support for i586 machines - Via C7-based netbooks, AMD Geode (e.g. the
>> OLPC XO-1), and so on.
>> -w
>> [1] except on in-order CPUs like Atom and the original Pentium
> well i understand why fedora stuck with downgrading to i586 now, saving  
> work by only having to compile 1 kernel version for the x86 arch (plus  
> to keep the olpc folks happy i suppose) and while linus is undoubtably a  
> clever guy i think intel know their arch better than anyone, and if cmov  
> was a performance hit it would be removed ;) looks like i'll need to  
> compile my own i686 for my aspire one as the i586 kernel in F11 beta is  
> showing a noticeable performance hit over F10 i686 :(

You can use the i686-PAE kernel on your Atom-based netbook.


Dominik 'Rathann' Mierzejewski <rathann*at*icm.edu.pl>  |  LAN  Staff
Interdisciplinary Centre for Mathematical and Computational Modelling
Warsaw University  |  http://www.icm.edu.pl  |  tel. +48 (22) 5540810

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]