It has never been smooth sailing on the good ship Itanic, but now the processor is at the centre of a poisonous row, one that's as nasty as any I can recall in years of tech reporting. In case you didn't notice – and you've come to the right place to catch up – HP and Oracle are at loggerheads. HP accuses Oracle, which bought …
Intel's EOL plan?
"...Itanium roadmap is good as far as 2014, and says it is exploring what succeeds it."
It's not what's said there that's interesting, it's what's not said. Regardless of what a successor chip might look like or be called, if there were any plans to continue with EPIC beyond 2014, then a commitment to EPIC would have been an obvious point to include here to steady the ship.
The fact that they are "exploring" this suggests that there will likely not be an Itanium successor per se, but a migration path. Probably to something x86ish.....
RE: Intel's EOL plan?
".....it's what's not said...." Which basically translates to "Regardless of what Intel say, I want to think Itanium is dead." The Itanium roadmap is longer and more detailed than IBM's Power one, and a darn sight more believable than the fairytale Snoreacle one for CMT! Intel and hp have both committed to Itanium and have plans past 2014, they have made public statements of that. Larry wilfully mistook a lack of Itanium content in a presentation from Intel's Xeon team to mean Itanium was due to be EOL'd, simply because he wants to believe that. Someone should have reminded him the Sunshiners tried that for years, and it was Sun that got EOL'd instead.
I had an IBM team in not too long ago talking about their x64 products - since they didn't mention Pee-series, should I start announcing that Pee8 is a dodo? Or how about that EMC storage team that didn't mention VMware, is ESX a goner too? And the Oracle Developers presentation we endured last year was all about Oracle SOFTWARE! Panic! Man the boats! By Larry's criteria that means we're all DOOMED!!!!
Itanium roadmap vs. Power roadmap
Let's look at this statement from Matty. Power and Itanium have roadmaps with two future chips.
IBM Power roadmap has real details for Power7+ and details Power8. While the Itanium roadmap might be "longer" because Kittson will come out long after Power8 does not make it better...just makes it late. There are no details for Kittson and Intel will not even confirm what nm it will be.
Come on Henrik..you can do better.
The perils of closed source
A lesson for HP, and by extension for everyone. Don't bet the shop on something over which you have no control! If they'd stuck to open source for their core needs, they wouldn't be at risk of being held to ransom: at worst they'd need to go hiring!
I recollect hearing the same lesson upwards of a decade ago from an unusually tech-literate and forward-thinking GP in our NHS ...
And they are getting paid how much for this?
Most businesses succeed not because of their top leadership but in spite of it and this just goes to show that none of them are worth the amounts of money they typically get paid. They aren't even competent enough to bribe a sex hound like Hurd well enough to avoid future problems. And Oracle letting Hurd do this is just stupid on their part when they could quite easily have slowly strangled the HP releases by reducing the number of people working on it and increasing support costs due to "low volumes".
But hey no surprise that Hurd thinks with his penis is there? But, at least it gives me something to chuckle over while I have my morning cup of coffee.
It's not about chippery, it's about the OS
"such is the value of the big iron business, built around boutique chips and chipsets"
No one with any sense buys an Integrity server because of The Chip Inside (tm).
They buy it because of the OS it runs (HP-UX, NSK, VMS), and what that OS allows them to do that they cannot get from any other OS on any other hardware.
If The Chip Inside (tm) was SPARC or Power or (heaven forbid) x86-64, some folks would still buy Integrity. Because of THE SOFTWARE.
Will this ever sink in?
Did Intel do Itanium to kill Alpha and MIPS on servers and continuing it to promote Xeon and hurt Power PC, ARM and Sparc for servers?
Linux and Windows is dropping support.
HP sort of invented it hence their ire. I think the architecture is a worse dead end than x86.
The dilemma for HP-UX shops.
The dilemma for HP-UX shops is this:
a) Keep HP-UX as your mission critical platform, risking early obsolescence for what should be the most long-lived systems and applications
b) Form a contingency plan. In which case Larry wins
In the midst of this awesome gladiatorial struggle between two titans, the livelihoods of thousands of honest decent IT professionals are "collateral damage". And this is saddening.
"secret death plan"
Only intel *may* know, and even then, to be honest, I suspect, maybe they themselves don't quite know what to do about the problem. Maybe what's really happening is they're really ignoring it and hoping it will go away and paying it attention once in a while just to keep punters happy. They probably don't have much of a plan as most of their resources are probably diverted toward x86 stuff, mobile computing / ARM threat.
What I think is that they won't get many (any?) new adopters probably because by now most people are probably thinking the same thing I am, and probably realise more so that any other chip than ever that it is not just chip design that is important here, it is *compiler* design, and I somehow doubt there are too many people really interested in this architecture, not with so much happening on other platforms at present.
Pity, it did sound different and interesting as a chip, when it first came out. Heck anything sounded more interesting at the time than IA-32.
I'm sure I've eaten this in a Viennese cake shop.
I had one
but the wheels came off.
Splashback on oracle
Very interesting article, let's hope that oracle see sense and realise that this could backfire on them. The consolidation of the Unix platforms strongly contrasts the proliferation of choices when it comes to databases. Oracle need to tread a bit more carefully.
MS and RH
Why should Oracle see sense and reverse this decision, but such hate and ire is not directed towards Microsoft and Red Hat, who have already stopped support for Itanium?
Re: MS and RH
Microsoft and Red Hat had very few (a few percent) total customers on the Itanium platform. Oracle develops the killer apps for all three HP operating systems (Rdb, Oracle, and Tuxedo.) The impact is orders of magnitude greater.
I'm HP-UX certified and a specialist in their products...
am I on the fast track to retraining (and all the associated costs and apprenticeships?)
No Pity for you....
Skill up in other OS's then, a Unix SA who only knows one flavour is vulnerable, go learn Solaris/AIX/Red Hat. If you are any good it won't take you long..
Ha ha ha ha ha.....
Satan to Beelzebub - you are a cheating lying creep. Beelzebub to Satan - you are a dishonest untruthful reptile. All other devils - help, the hellfire furnace is beached on an iceberg and freezing over fast!
If Intel are really planning to "Sink the Itanic!" as soon as they think they can get away with it, what on earth could they replace it with? Makes me wonder what would have happened if Intel could have bought the Alpha IP in 1999 and made *that* their basis for a next generation processor...
what on earth could they replace it with?
They kept one ARM SoC in the sell off to Marvell
They still are ARM licencee
Apple wants an ARM iMac
Intel owns Infineon which is ARM SoC based.
Intel will say they are not doing an ARM till the day they do.
Which is easier to progam? EPIC x 4 cores or RISC x 16 cores?
Which lets you have chip set onboard with more RAM? 3 billion transistor EPIC or 500 thousand transistor RISC?
Which is cheaper to buy and run? 10 Virtual servers on one EPIC or 10x ARM SoC each with own RAM in a smaller ox using less power?
would have to be arm designs
Considering what a fail design atom is (hello Intel, think mW not W) and how poor Intel has always been in the ultra low power category looks like joining them instead of beating them is the only option (and spare the quotes of future wizbang vaporware chips Intel FUDS the market with, believe it when you see it).
"Intel will say they are not doing an ARM till the day they do"
ARM? I understand they're thinking of making a 64-bit extension to the ARM architecture, but (a) it will be an *extension* and (b) other than Itanium, ironically, Intel design their own CPUs.
Realistically, to displace high-end x86 in the server market, not to mention have any chance of maybe someday replacing x86 on the desktop, Intel would have to come up with a clean traditional 64-bit architecture with few or no programmer-visible quirks. Either something like the Alpha itself or go the other way and have implicitly parallel CPU with out-of-order and speculative execution, multithreading and so forth.
"other than Itanium, ironically, Intel design their own CPUs."
Depends what you mean by design.
Where did Intel get the idea for a 64bit x86?
Where did Intel get the idea that a 64bit x86 might be profitable?
Where did Intel get the idea that "industry standard 64bit" wasn't going to be IA64?
OK Intel built their own circuitry for x86 once they'd been forced to admit that not only was it possible but might even be profitable, but it was hardly Intel that designed the architecture.
In fact, what have Intel successfully designed and produced in the last ten years or so, other than x86s?
Sorry, still thinking of Intel as innovators there ;)
I was thinking in terms of i860, i960, revisions of x86..., but yes fair nuff they are all old designs. I for one would have loved to see a 64-bit from-the-ground-up version (not simply extension) of the i960HX and aimed at the high-end rather than embedded.
You can still find a PDF of the programmers reference manual for the BiiN (the failed project chip that became the i960) via Google, shall enjoy reading that. Many people appear to regard the original i960 general purpose CPU, before Intel re-targeted it as an embedded chip, as "432 done right" and if it hadn't been for the i860 coming out and resistance from the i386 team, who knows what might have been?
Maybe HP should just buy Oracle
Perhaps they could partner up with Google to do it.
I can't see the IT industry being any worse off if Larry Ellison is given the boot, and they would have the opportunity of firing Hurd twice.
Failing that it might be worth spending a bit of time looking into some of Oracle's dirty laundry. IMHO, Oracle has never been at the forefront when it comes to business ethics.
chop oracle into parts
Nice idea. Maybe they can even partner with SAP and chop the Oracle into parts - database, ERP software, java and hardware ;-)
"Intel owns Infineon which is ARM SoC based."
Intel owns Infineon? Are you sure?
The rest makes much more sense though.
9/10. Careless. See me after class.