Slashdot Mirror
HP's NonStop Servers Go x86, Countdown To Itanium Extinction Begins
An anonymous reader writes "HP has been the sole holdout on the Itanium, mostly because so much of the PA-RISC architecture lives on in that chip. However, the company recently began migration of Integrity Superdome servers from Itanium to Xeon, and now it has announced that the top of its server line, the NonStop series, will migrate to x86 as well, presumably the 15-core E7 V2 Intel will release next year. So while no one has said it, this likely seems the end of the Itanium experiment, one that went on a lot longer than it should have, given its failure out of the gate."
Not a single major hardware or device maker seems ready to support Linux on non-Intel architectures. Intel, MS, HP, Cisco etc. are part of the TCPA alliance; even Linux on ARM based servers have taken a very long time to arrive.
If you keep throwing chairs, one day you'll break windows....
given its failure out of the gate.
For a multibillion dollar industry, "failure" is a rather strong term. It may be declining, but it topped over $4.4bn a year at one point. That's probably bigger than AMD.
SJW n. One who posts facts.
Earlier this year HP announced the end of the line for VMS. That was certainly connected with the Itanium retirement as well.
"Almost every wise saying has an opposite one, no less wise, to balance it." - George Santayana
I work on a product that supports Itanium, and we have a few customer that are still using Itanium servers, who knows why. We just discovered that unless you get the top-tier developer subscription to Microsoft Visual Studio, you don't get Itanium compilers.
PHEM - party like it's 1997-2003!
Let me count the ways
And there was much rejoicing!
And nothing of value was lost.
For those saying it wasn't a failure you must look at what Intel intended Itanium for. If they had succeeded Intel would have owned the 64 bit CPU realm on the desktop with a proprietary architecture effectively eliminating any competition in the space. To succeed they had to get all popular software including Windows to be rewritten for the new processor. This was a daunting task and few were ready at the time to make the switch to 64 bit. AMD introduced the Opteron in 2003 with their 64 bit extensions for the existing x86 architecture which allowed the reuse of the 32 bit code in existence. AMD's x86-64 was well received and Intel ultimately adopted the architecture in their own processors. So yes the Itinaium failed to succeed in its intended task despite lingering for over a decade.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
Unfortunately it became a legend for all of the wrong reasons. Billions of dollars have been sunk into it over the years and many lawsuits have been filed over it demise by vendors desperate to get out of it or force another vendor to stay in it.
http://www.eweek.com/servers/hp-to-seek-4-billion-in-damages-from-oracle-over-itanium/
http://news.cnet.com/Allies-pledge-10-billion-to-boost-Itanium/2100-1006_3-6031773.html
http://www.masslive.com/news/index.ssf/2013/09/hudson_intel_plant_closing_wil.html
Unfortunately sales never came close to the billions of dollars that have been sunk into it, and it has been that way for years:
http://www.theregister.co.uk/2005/02/28/itanium_04_sales/
http://www.wired.com/wiredenterprise/2012/02/hpearnings/
http://www.zdnet.com/photos/charts-mining-itanium/21115
I'm sure someone has a comparison of how much money has been invested compared to how much money has been made in sales. I might be mistaken, but from what I've been reading from the beginning Itanium has never come close to breaking even for hardware or software sales. Certainly companies like HP and Oracle spent millions of dollars on their lawsuit trying to get out Itanium.
Itanium has always been nothing more than a desperate multi-billion dollar effort to break free from the chains of x86.
Until there is a supported COBOL environment in Linux, HP-UX on Itanium will be around for a long time.
I work in the power industry, and we use some very specific applications that are only available on HP-UX and AIX. HP-UX is by far their largest install base.
These apps are used by the power plants/coal mines for everything. As you'd expect, there are very few applications that are certified for use by the power industry that meet the regulations. The one we use will begin supporting LDAP instead of NIS next year.
There's no incentive for new players in this software market due to the small number of potential customers and the massive trust curve they'd have to meet to make somebody switch.
We're one of the reasons there's a pretty long road map for Itaniums and HP-UX.
My mom says I'm cool.
It's just a grid of cores on the chip layout. Nothing wrong with a grid that's 3x5. A 2-dimensional grid does not force the number of cores to be a power of 2.
IA64 started as an HP Labs project to be a new instruction set to replace HP's PA-RISC. VLIW has a hot topic around 1995. HP Labs was always proposing stuff and the development groups (those making chips/systems) ignored it, but for some reason this had legs.
The HP executive culture is: HP hired mid-level executives from outside. They would then do something big to get a bigger job in another company. A lot of HP's poor decisions in the last 20 years can be directly traced to this culture. And there was no downside--if you failed, you'd go to an equivalent job at another company to try again.
So enterprising HP executives turned HP's VLIW project into a partnership with Intel, and in return HP got access to Intel's fabs. This was not done for technical reasons. Intel wanted a 64-bit architecture with patents to lock out AMD, and would never buy PA-RISC. So it had to be new. HP was behind the CPU performance curve by 1995 due to its own internal fab not keeping up with the industry due to HP not wanting to spend money. So HP could save billions in fab costs if Intel would fab HP's PA-RISC CPU chips until IA64 took off. So, for these non-technical reasons, IA64 was born, and enough executives at both companies became committed enough to guarantee it would ship.
For a while, this worked well for HP. The HP CPUs went from 360MHz to 550MHz in one generation, then pretty quickly up to 750MHz. I thought IA64 would be canceled many times, but then it became clear that Intel was fully committed, and they did get Merced out the door only 2 years late. IA64 was a power struggle inside Intel, with the IA64 group trying to wrest control from the x86 group. That's where the "IA64 will replace x86" was coming from--but even inside Intel many people knew that was unlikely. Large companies easily can do two things at once--try something, but have a backup plan in case it doesn't work.
But IA64 as an architecture is a huge mess. It became full of every performance idea anyone ever had. This just meant there was a lot of complexity to get right, and many of the first implementations made poor implementation choices. It was a bad time for a new architecture--designed for performance, IA64 missed out on the power wall about to hit the industry hard. It also bet too heavily on compiler technology, which again all the engineers knew would be a problem. But see the above non-technical reasons--IA64 was going to happen, performance features had to be put in to make it crush the competition and be successful. The powerpoint presentations looked impressive. It didn't work out--performance features ended up lowering the clock speed and delaying the projects, and hurting overall performance.
Itanium would have allowed Intel to dump all the x86 baggage and move the world to a Brave New Shinier CPU that was 64-bit and appeared to offer substantially better performance.
And it would have made them sole supplier for the mainstream CPU market, taking out AMD and the other clone x86 makers.
Unfortunately, the early compilers sucked and x86 emulation really, really, really sucked, so no-one with a big investment in x86 software was going to make the switch. If I remember correctly, it was also years late, so performance that would have been impressive at the initial release date had become 'meh' by the time it actually hit the market.
The thing I loathe most about the Itanic is how it sunk 2 or 3 better CPUs before it - PA-RISC, MIPS V and Alpha. Compaq/HP should have left NonStop on MIPS and VAX on Alpha, and never gone the Itanic route w/ them. Heck, even Linux largely (except Debian) has abandoned the Itanic, and only FreeBSD, of the BSDs, did an Itanic port (I'm not sure that even the much ported NetBSD or OpenBSD has been ported to it). There is even less reason for VMS or NonStop to have gone Itanic.
Two things the Itanic can be good for - supercomputing, and a test bed for Inferno/Plan 9.
Ditto that comment. I really hate the fact that HP gets any credit for NON-STOP, Tandems baby, or VAX and Alpha, 2 of the 3 great things that Digital brought forth into this world, the other WAS Alta-Vista. The only thing HP ever did well was printers and that time has LONG since passed...
Note : I spent several years supporting all of the above machines and may be a bit biased.
errr....umm...*whooosh* *whoosh* Is this thing on ?
Once this line goes x86, how are they different from anybody else - their own ProLiant line, or top end servers from IBM or Dell? Also, are they planning to migrate HP/UX to x64, or will they simply migrate HP/UX customers to Lintel (Linux on x64)? If it's the latter, their customers have probably beaten them to it by probably more than a decade
So once this line is x64, there will be nobody who's making Itanium servers, will there? Or will it be a China only CPU - w/ customers like Huawei? So does that mean that Intel will finally put the Itanic out of its misery?
Your forgetting RPN calculators and instrumentation
No, it didn't. Reason HP went w/ it was the perception that since RISC was faster than CISC, which got proven by Intel building in all sort of RISC based concepts into the Pentium, which included moving some of the complexity of a CPU to compilers, VLIW could be faster as a result of moving all of the complexity of a CPU - even a RISC CPU - into the compiler.
Main issue w/ that, even before the project started - was the fact that in VLIW, since everything - register renaming, speculative execution - is moved from the CPU to the compiler, w/ every CPU generation, unless all you do is bump up the GHz as well as the cache, any change to a CPU would necessitate recompilation of existing software to get any performance improvement: running existing binaries would show no improvements whatsoever. That alone should have killed the idea.
As it turned out, RISC was an optimal spot b/w CISC & VLIW. While Pentium adapted a lot of RISC concepts like superscalar execution, expanded register sets and branch prediction, Itanium too smuggled in RISC practices like Register Renaming, which is supposed to be gotten rid of in VLIW. In the meantime, RISC CPUs such as Alpha 21364 & POWER adapted some VLIW concepts such as smarter compilers that extract more parallelism, MIMD operations and so on. So HP would have done better in selling Intel the PA-RISC and Intel improving that rather than starting out w/ this VLIW concept.
No, Wikipedia is right... as Section 1.1 of the MIPS R10000 Microprocessor User’s Manual says, "MIPS has defined an instruction set architecture (ISA), implemented in the following sets of CPU designs: MIPS IV, implemented in the R8000 and R10000"
The R10000 is MIPS IV, not MIPS V
or will they simply migrate HP/UX customers to Lintel (Linux on x64)?
One can only hope, HP-UX has been walking dead for years. 11.31 is six years old, 11.00 was released in 1997. It is at best an edge on an edge case, I want both the Itanium and HP-UX to go away. We have enough choice with Linux, Solaris, and AIX.