Re: Switching to 64 bit
I'm sorry, but if you're using a 32-bit PAE kernel to address more that
4 GiB of RAM on a *64-bit machine* you are completely and utterly doing
it wrong (In fact, you're doing it stupid.). PAE is slower, can't
address anywhere near as much as native 64-bit can, and isn't as stable.
Heck, even the Linux devs would rather people use native 64-bit than PAE
on 64-bit machines. Unless there's some actual reason, like a 32-bit
application you absolutely can't live without doesn't work with
multilib, there is 100% no reason not to use 64-bit OS on a 64-bit CPU,
and about several dozen good reasons why it's a stupid idea to limit
yourself to 32-bit.
On Thu, 2013-06-27 at 23:15 -0500, Stan Hoeppner wrote:
> On 6/27/2013 10:12 PM, Gary Dale wrote:
> > On 27/06/13 10:56 PM, Stan Hoeppner wrote:
>
> >> You don't have to change anything. 32 bit Sid will run just fine on an
> >> x86-64 CPU. Switching to 64 bit software is a choice, not a requirement.
>
> > Not really.
>
> Yes, really. This is fact. Don't disagree with facts Gary, especially
> when I am the one stating them. And don't do it with a flexible spine.
> If you're going to disagree with me, take a firm stance.
>
> > 32bit systems have memory limitations that you don't
> > encounter with 64bit.
>
> There are two such limitations when using a PAE kernel and 32bit user
> space on x86-64, which are the same limitations on P6 class CPUs:
>
> 1. 64GB maximum physical memory
> 2. 2GB per process address space
>
> ~99.9999% of desktop Linux users will never exceed either of these. The
> same number of users have less than 64GB physical memory. For those who
> have trouble with percentages, this is 1 in a million users.
>
> Given your propensity for disagreeing with facts, I'll assume that you
> are also the type of person who infers intentions, or a position, that
> do not in fact exist. So I'll spell it out clearly:
>
> I am not *advocating* that people in masse run a 32bit software platform
> on x86-64 hardware. What I've done is state a valid option that may be
> preferable to the OP who posed the question.
>
> Instead of disagreeing with me, with the facts I presented, you should
> have simply stated your case for a full 64bit platform, in a standalone
> manner directly replying to the OP's post instead of my factual reply.
>
> --
> Stan
>
>
Reply to: