The future of i486 arch in Th

Tomasz Pala gotar at polanet.pl
Tue Dec 30 22:33:11 CET 2014


On Tue, Dec 30, 2014 at 20:22:01 +0100, Jan Rękorajski wrote:

> If we have i686 arch we need to support i686 CPUs, not a random subset.

We don't "need" to do anything. SSE2 is 13 years old now (and available
for 11 years in AMD), I haven't seen unsupported CPUs in years.

Hardware produced in that time is already dead (capacitors on mainboard
were dead after 2-3 years of running in heat produced by power hungry
CPUs and graphic cards). I've got more fully-functional i586- hardware
in my drawer than early-2000 crapware.
Pre-SSE2 hardware is worth less than energy it consumes, supporting this
is unrational - unless you prefer proper terminology over pragmatic
approach.

>> So how about replacing i486 with i786?
> 
> Creating such artificial entity would be a waste of resources.

Running i686 without modern extensions is a waste of resources.
Creating x32 is a waste of resources (this won't ever be generic purpose arch).
Let's drop i686 as well and call this i686+, i686_with_SSE2 or whatever
you like if i686 name is bogus for SSE2 superset and
i786 is not appropriate (e.g. pentium4 in gcc way).

2012: https://bbs.archlinux.org/viewtopic.php?pid=1153414

Or let's keep ancient. After all there might be 2 or 3 PLD users not
subscribed to pld-devel, one of them might still use 32-bit AMD.

-- 
Tomasz Pala <gotar at pld-linux.org>


More information about the pld-devel-en mailing list