Comments on: SGI's Unix variant fading into history

Silicon Graphics to stop selling computers using MIPS chips and Irix operating system at end of 2006.

[katamari]

IRIX - good riddance.

Any decent, security-aware system administrator knows what I mean by this...

[rsfinn]

You forgot one...

"The three major versions of Unix that still are under active
development are HP's HP-UX, Sun Microsystems' Solaris and
IBM's AIX; of those, only Solaris runs on x86 chips."

You left out Apple's Mac OS X, which also runs on x86 chips.
The only grounds under which I can imagine not including it as a
"major version of Unix ... under active development" is (a) it's not
System V derived (so what) or (b) it's not UNIX certified -- but
"Apple intends to submit Leopard Server to The Open Group for
certification against the UNIX 03 product standard" <http://
www.apple.com/server/macosx/leopard/more.html>.

[Penguinisto]

Two, actually...

FreeBSD still slogs along very nicely on x86. Though it's not SysV derived either (no ****, right? :) ), it (as with most BSD variants) still counts as *nix no matter how you slice it.

Also, SCO is dying not because of its *nix products, but because their leadership managed to anger the vast majority of sysadmins on this planet with what has to be the world's most boneheaded lawsuit. That should've been made a bit more clear IMHO (instead of being tossed out there as an afterthought).

[pinniped]

What have the Romans ever done for us?

This story will run a bit like that scene from Monty Python's "Life of Brian" - What have the Romans ever done for us?

NEC supercomputer clusters run on a few UNIX operating systems but most installations will run on NEC's own UNIX which is still very much active.

There is a lot of serious research equipment out there which is still running on SGI boxes with IRIX. I honestly don't know why (for research gear)SGI/IRIX seemed more popular than Solaris, but that's the way things were from the mid 1980s through the late 1990s.

As for CPUs, one chip I do miss is the Alpha - I wish AMD would put an Alpha-like math unit into their processors - that will be one frighteningly fast beast. Since the AMD64 family have a 32-bit math unit, I really don't see any advantage to this "64-bit" chip except for the addressable memory. (But whether you use 'em in 32-bit or 64-bit mode, they're still superior to Intel chips.)

[Penguinisto]

Itanium... *puke*

Gad; I really wish the *nix industry would've settled on an x64 that wasn't so poorly executed...

[pinniped]

Shhh...

It's sacriledge to poo-poo the Itanic.

[John Kuzak]

*nix industry

http://www.analogstereo.com/vauxhall_omega_owners_manual.htm

[ThePenguin]

Elegant IRIX

IRIX straight out of the box was less secure than say SunOS, because SGI promoted the idea of integrating IRIX in with MacOS and Windows clients and servers.

Once properly configured by a competent sysadmin, IRIX was more secure than a similiarly configured SunOS box.

Past the security "issues", nothing could touch IRIX's reliability, uptime and throughput.

Sun wishes they had the interconnect (craylink, later NUMAlink) technology that SGI had. Latency across those links from the 2 farthest nodes are on an order of magnitude less than the best that Sun could ever muster.

IRIX was stable, fast, reliable, secure and elegant. I have used IRIX since the original Indy workstation, all the way upto Origin 3000 servers.

I only wish SGI had played it's financial and technology cards better.

If superior technology was the only factor to commercial success, then SGI would be a household name.

[wbenton]

The enevitable!

As more and more variants disappear... a singular standard will begin to appear.

That's the point which Microsoft needs to be wary of because it WILL replace them!!!

Walt

[chuck_whealton]

Probably not the wrong move...

I guess the worst part about this is that there will be less competition on the CPU market.

As for Irix versus Linux, at least maybe SGI users won't get socked having to shovel out boatloads of dollars for a software development system that should be included in any version of the UNIX operating system.

Other than that, I hope SGI can turn it around. We need competition among vendors, even if we'll probably be stuck using systems that all contain similar CPUs.

Charles R. Whealton
Charles Whealton @ pleasedontspam.com