<taps> So long, Pentium III, Athlon XP, and older...

[eccerr0r]

Re: &lt;taps&gt; So long, Pentium III, Athlon XP, and older...

Actually I probably should have said Coppermine and older, forgot about those... then again not sure if Coppermine has SSE2 or not. I know my Tualatins have SSE2 among the P4s.

It's unfortunate this is even Linux, and even for a source based distribution (like Gentoo) if the source stops maintaining it, still screwed...

I actually still have 32-bit machines in active duty, one is a VM, other is an Atom. I think all my other 32-bit setups aren't in primary service any more. My old 32-bit laptop is a Pentium-M so at least that has a little more life with SSE2.

[linuxguru]

Re: &lt;taps&gt; So long, Pentium III, Athlon XP, and older...

While I do have legacy systems with PIIIs in sporadic use, they're all Coppermine or Tualatin (more recent than Katmai). I have Athlon 32 T'bird and Thoroughbred XP boxes, but running older OSes (RH9 and so on, with no net connectivity). The direct impact to me is limited for now, but I suppose the writing is on the wall for i686 and IA32 in general - it's headed for EOL, and migration to AMD64 is inevitable.

Edit: I think all the (very few) Core Solo SKUs were x86-64 - Merom/Conroe single cores, but I could be mistaken. The last IA-32 core was probably Yonah, a dual-core, and some LGA775 P4s without x86-64 enabled.

[eccerr0r]

Well, looks like they did it.

Rust, the "secure" language used for firefox and librsvg, no longer work on AthlonXP, Athlon, Pentium III (Katmai and older), Pentium II, and anything older. Because of rust's decision to call "i686" at least a Pentium 4 (with SSE2), these older machines have been left out to pasture. Not sure how hard it is to remove SSE2 instructions from rust's instruction scheduler and still produce a working binary...

I've been depending on these old machines to keep on working as they still share the same instruction set...or seemingly do, until people actually force one to have SSE2.

librsvg is significant as it's a component of gtk+ -- which is used by xfce4 and gnome desktop environments. Looks like KDE or lxde may need to be looked at, and even KDE with its dependency on QT5 and QT5's insistence on having SSE2 instructions, though that can be worked around.

So I guess the original P4, Pentium-M, Pentium III (Tualatin), Core Solo, or perhaps all 32-bit machines are next on the chopping block?