Re: Using i386 by mistake on 64-bit hardware [was Re: i386 in the future 32-bit archs: a proposal)]
On Sat, May 20, 2023 at 09:15:00AM +0200, Josh Triplett wrote:
> How easily could we add 64-bit system detection to the i386 installer,
> and a message saying something like:
>
> "You're installing the i386 architecture on a 64-bit system. While this
> will work, this is the last release it'll be supported. We recommend
> installing the 64-bit amd64 architecture instead.
This is not a valid use for i386. Running the i386 kernel on _modern_
hardware is insecure, slower (esp. if you have a non-tiny amount of
memory), etc. We should put a big fat warnings for _that_.
On the other hand, it's okayish to run 32-bit userland -- in fact, in
some cases it might be quite a bit faster¹, as long as you use the
appropriate kernel. Which is 32-bit for ancient hardware (including
some first 64-bit-capable models), and 64-bit today.
Meow!
[¹]. Sometimes amd64 wins by a lot due to more wider registers and SSE,
sometimes i386 win hugely due to halved pointers and better code density,
which allows using a higher-tier cache. We could have been using x32
to get both, but oh well...
--
⢀⣴⠾⠻⢶⣦⠀ The ill-thought conversion to time64_t will make us suffer from
⣾⠁⢠⠒⠀⣿⡁ the Y292B problem. So let's move the Epoch by 435451400064000000
⢿⡄⠘⠷⠚⠋⠀ (plus a safety margin in case of bad physicists) and make it
⠈⠳⣄⠀⠀⠀⠀ unsigned -- that'll almost double the range.
Reply to: