Following the removal of IA-64 (Itanium) support in the Linux kernel and glibc, and subsequent discussions on our mailing list, as well as a vote by the Gentoo Council, Gentoo will discontinue all ia64 profiles and keywords. The primary reason for this decision is the inability of the Gentoo IA-64 team to support this architecture without kernel support, glibc support, and a functional development box (or even a well-established emulator). In addition, there have been only very few users interested in this type of hardware.
↫ Gentoo website
Et tu, Gentoo?
Linux removing Itanium I can understand; the Freemason corporate overlords who pull the strings of Linux kernel development are terrified of just how powerful Itanium really is. GCC removing Itanium makes sense too, as the unwashed communists at the FSF just don’t understand the capitalist greatness that is Itanium. But Gentoo?
Now I know how Jesus felt when Judas betrayed him; how Caeser felt when he gazed upon Brutus’ face. I feel empty inside.
I could do without the blasphemous commentary…. jesting at the Itannic is fair game though.
Bye bye Itanium, you were an attempt by Intel to build a monopoly on CPUs (since they had to share x86 with AMD but didn’t have to share IA-64 with anyone), so in a sense, Itanium sucking saved the entire industry.
Probably DEC dying also contributed to the situation has Alpha DNA and brains ended up in AMD’s platform.
I like to think of things as many issues converging to a point, rather than reducing every point in history to a single cause.
Alpha’s teams, as well as PA-RISC, ended up at intel as well.
Both Nehalem and Opteron were basically what Alpha and whatever PA-RISC 2.x would have looked like if they had been used in platforms with a viable market scale.
Itanium was technically the heir to PA-RISC 2.x of all things. It’s original name was PA-WIDE (or PA-RISC 3.0)
We’ll yes I assume so but that is much later… DEC guys (like Jim Keller ended up at AMD working on Athlon in 1998 a decade earlier than Nehalem).
I dont’ know all the guys names but they were involved in designing hypertransport also… which in it’s later iterations is still relevant today.
Intel took over DEC’s fabs the StrongARM team and a big chunk of their high performance uArch personnel as well. I think part of it was some sort of settlement.
Keller is a great asset in terms of engineering leadership. Also DEC and AMD were also sort of working on some shared architecture projects as well. Since there had a couple projects before DEC imploded where the goal was to offer a common system platform where DEC AXP or AMD x86 parts could be drop in. And AMD themselves did not have as good out-of-order uArch in the mid 90s as Intel did. So those DEC guys had a much bigger and direct impact @ AMD.
Incidentally a lot of the research uArchs @ intel during most of the 00s started life as AXP cores, since the software simulator used for early architecture exploration came from the DEC team 😉 Later on during the design process, the x86 front end would be bolted in. But for a brief moment during the initial phase of the design Nehalem, Sandy Bridge et al started as simulations running AXP binaries 😉
LOL, the hyperbole in internet forums.
FWIW HP shared IA-64 with Intel initially. Itanium was as much HP as intel. Lots of people forget that…
I think Itanium was mostly HP. Intel did some architecture work, and fabbed the thing, but HP did most of the legwork regarding ISA
Yeah. Itanium had a much stronger backwards compatibility with PA-RISC at the ISA level. And a lot of the compiler team came from HP. And basically the last remaining customer was HP, which was the only reason Intel didn’t can Itanium earlier.
A lot of people have it backwards, in terms of thinking that Intel hoodwinked the rest of the industry into adopting itanium in lieu of their own high performance RISC designs. When it was initially HP who hoodwinked Intel into adopting their internal PA-WIDE architecture as the future of Intel’s 64bit strategy.
It was HP who put a lot of effort into convincing Intel to give LIW another chance. Right after the bad taste that the i860 left in Intel’s mouth regarding exposing a lot of the scalar ordering to the programmer.
So in a sense HP was responsible for convincing a big chunk of the industry to discard their own RISC designs for PA-WIDE 😉
On a politics side, it’s indeed interesting on how around 20 years back GNU/Linux might still have been perceived as a communist idea, to some, and now each and every big IT company uses it extensively, regardless of the economic system it operates in. It just show on how wrong a simple thing as a perception can be. Mesmerizing.
It’d be cool with a special post about the corporate freemasonry, communism (and what should we call Gentoo? Alchemists?) in the Linux world 🙂
The idea behind Itanium, EPIC, may return. There is an outfit calling themselves “Mill” (horrible name) who are working on a processor that has a resemblance to the architecture behind itanium. Hopefully the paradigm was just a bit before it’s time, and maybe deserves a story here?
https://millcomputing.com/
1. EPIC is a sub variant of a VLIW (very long instruction word) architecture
2. VLIW is an idea from the early 80s – long before EPIC.
3. The “Belt” by Mill-computing is using some form of VLIW but not(!!) EPIC
4. The “Belt” is more related to stack machines, than to register machines – it does in no way resemble the architecture of Itanium
5. Sadly the “Belt” CPU is absolutely dead since a couple of years now
6. The Russian “Elbrus” CPU is VLIW
FWIW EPIC was more like LIW (the instruction bundle in IA64 is just 3 ops wide at most I believe, which is not particularly “very large”).
In fact Intel/HP made a whole noise about Itanium architecture type being referred to as EPIC not VLIW.
Not that it helped any, because once the whole Itanic caught on…
Thom,
Sorry to be that guy, but please:
s/Caeser/Caesar/
I can’t unsee it 🙁
There’s still HP-UX and uhh.. yeah.
And OpenVMS!
Itanium works fine with HPUX. Will have support longer than Windows 10 on x86-64. Ill be dead by the time Itanium is out of support by HP in 2046.
Without kernel and glibc, Gentoo just cannot support Itanium. How could they? Linux distributions are just glorified repackagers. If they had skills/time/resources to maintain IA64 support, they would do it in kernel or in-glibc.