back to article Intel drags feet on Itanium quad-core (again)

Are you one of the IT shops with Itanium-based servers at the heart of your data center expecting a substantial speed boost this year from Intel with the quad-core "Tukwila" Itaniums? You'd better forget about it, because the processor won't be making an appearance until the first quarter of 2010. As we reported earlier this …

COMMENTS

This topic is closed for new posts.
  1. Pierre

    HA-HA

    BS detector went so high, it looks like Amy Winehouse now

    "During final system-level testing, we identified an opportunity to further enhance application scalability best optimized for high-end systems"

    Read: we fired it up and noticed that during the first 5 milliseconds it was running at 8 MHz, after which it burst into flame and burnt the whole facility to the ground. We might have to improve it a bit before release.

  2. Anonymous Coward
    Paris Hilton

    "as they ship to OEM server makers"

    C'mon, get real. Itanium = HP, you even say as much in the rest of the article. And therefore Itanium is in a death spiral of increasing non-manufacturing costs as time goes by (chip development and fab setup costs aren't small and aren't getting smaller) and Itanium sales revenues hardly cover Itanium non-manufacturing costs without subsidy from the rest of Intel, subsidy which cannot continue indefinitely, in a market where Itanium is increasingly irrelevant.

    "100,000 chips being pumped out each quarter and chips ranging in price from between $2,000 and $3,000"

    Let's take that as read. In which case I make that $800M to $1,200M per year. Howcome your article then says "somewhere between $1.5bn and $2.5bn in Itanium processor sales a year"? What did I miss?

    "NonStop servers actually need hardware features relating to system reliability that are on Itanium but are not on Xeon chips."

    HP/Intel PR BS, on the whole. Justify the claim or withdraw it. Also note that there's a world of difference between features of an *architecture* which are usually hard to retrofit if not considered from day 1, and features of an *implementation* which may be optionally implemented in a given CPU and support chipset, depending on constraints such as target time to market and target system cost. What *architectural* features is AMD64 missing? See e.g. the RAS-related comments on your recent Itanium article, and don't forget the evidence in those comments that chip level lockstep in Itanium, and memory resilience via lockstep in Xeon, is pure BS.

    http://www.theregister.co.uk/2009/05/05/itanium_optimism/comments/

    (hello Al)

  3. Anonymous Coward
    Anonymous Coward

    RISC replaces RISC

    Since Itanium is RISC, how is it a "RISC replacement"?????????????????

    Or is this Intel speaking out of both ends again???

  4. Peter Hawkins
    Thumb Down

    Good grief Matt

    HP's delivery of Tukwila based systems is making Sun's Rock timescales look positively speedy.

    Anyway what do you expect from a reseller, it's not like HP actually design anything any more anyway.

    FAIL

  5. Anonymous Coward
    Alert

    HP's Itanium policy

    I think HP had better have a skunk works somewhere reversing the endianship of HP-UX and porting it to Nehalem. I also disagree that Xeon doesn't have availability that matches Itanium. It may not be built into the CPU but that's what a chipset is for. A computer system after all is more than just a CPU. The main reason HP customers have put up with Itanium delays and disappointing performance is that HP have to some extent compensated for it with advanced chipsets and adding capacity to their ccNUMA server designs. These features have gone a long way to make up for the shortcomings in Itanium performance, though the delays in delivery of Tukwila and the expectation that it will only double the performance of the current Montecito and the still oversized fab process (go figure, if you have two Montecito 9100s on a SX2000 motherboard you can blow away the projected performance of Tukwila right now), have led customers looking to upgrade who are running portable sortware such as Oracle to look at other options. At the moment support agreements are largely what keep customers loyal to a given hardware vendor. These delays and the cost of Intel producing two different processor architectures don't bode well for the future of Itanium, however. Nor for HP-UX if HP doesn't port it since they also have Linux as an option. People buy servers to run applications, not operating systems. So far HP has managed to keep the Itanic afloat but if customers start moving to IBM or Fujitsu they may rethink that policy, and much of their chipset technology can be adapted to Nehalem.

  6. Anonymous Coward
    Thumb Down

    Tukwila vs. Common Sense

    Anybody still believing that Tukwila will see its successors, Poulson, Kittson, Whateverson getting reality?

    Tukwila is now something like 4 years delayed.

    It still uses the outdated Madison core.

    "Performance ~ 2x faster than its predecessor".

    Well, yes, sort of. 4 cores on Tukwila vs. 2 on Montvale.

    Which reads "Even if we might overcome the 1.6 GHz barrier, a single core of Tukwila will probably come close to the single core performance of Montecito or Montvale".

    Or "we screwed that thing up several times so badly, we still don't even get the numbers from processors released 4 years ago."

    Tukwila will be released together with the Hurd kernel and Duke Nukem Forever.

  7. Ken Hagan Gold badge

    Re: RISC replaces RISC

    "Since Itanium is RISC, how is it a "RISC replacement"?"

    Whoah! Itanium is about the least RISC-y architecture ever attempted. It is so massively complex that no-one has ever actually built a chip that implements more than a fraction of the original plan. Indeed, it is highly probable that no-one ever will and it may even be truly unimplementable due to internal contradictions in the various features. It took about five years after the initial release before anyone produced a compiler for it that was safe to use with optimisation switched on.

    The acronym you were looking for, btw, is VLIW. Itanium is that, at least for sufficiently short values of "very long".

    As an aside, VLIW is a RISC replacement in the sense that it is the next four-letter acronym to be utterly destroyed by the x86 juggernaut.

  8. Henry Wertz Gold badge
    Joke

    correction

    "But Itanium is not a hellacious money pit in terms of investment and they will have to soldier on."

    I think he might have meant "Itanium IS a hellacious money pit" Zing!

    Actually, the only two Itaniums systems I've seen were not a money pit at all, at least for the purchaser -- HP was pretty desperate to show Itanium sales so they discounted them about 98%. (They were to replace some PA-RISC Superdomes, but initially ran slower than the Superdomes until the Itanium compiler improved.)

  9. Macka

    Just what HP didn't need

    So bang goes Tukwilla's only real potential advantage - a modern up to date competitive processor for 4+ socket systems for HP to sell before Nehalem EX ships and steals its market. The window of opportunity is lost. Today's Montecito 9100s don't cut the mustard any more. AMDs shanghai processor already bests that in both integer and floating point performance at less cost, and Istanbul will just leave it in the dust.

    I believe HP when they say they don't have a skunkworks project to port HP-UX to Xeon. It makes no sense. The cost would be enormous: an endian flip, porting and writing new drivers, etc, and would only serve to put HP-UX in direct competition with Linux. it would get eaten alive, what with the expensive cpu licensing costs levied on HP-UX customers. HP would never get the money back.

    But what about a set of HP-UX compatibility libraries for Linux? Just install them, recompile your HP-UX apps and you're good to go. That would be far cheaper and quicker. Or maybe an HP-UX / Itanium run time binary emulator? Apple have done it with Rosetta (from Transitive) to successfully transition off PowerPC to x86 so it's a workable get out of jail card. In fact HP already have a relationship with Transitive to try and tempt customers off their SPARC Solaris systems, bringing their apps with them and onto HP ProLiants using Transitive software. So they already know how to do this. Only time will tell.

  10. Oninoshiko
    Boffin

    replacing "propritary arcitectures" like SPARC?

    How can I be the first person who has caught this bold faced lie. Sparc is not a propriatry arcitecture, it is an open one. Anyone who wants to can call up a endependent fab facility, send them the freely avalible VHDL of a couple of the varients and have a run done.

    try that with Itanium.

  11. Matt Bryant Silver badge
    Happy

    Please, no more obituaries just yet!

    <Yawn> Oh dear, is the same crowd telling us Itanium is dead yet again? Did you forget the current crop of Itaniums are doing just fine owning that high-end enterprise market (you obviously didn't write it on the back of your hands like I told you to)? With Power6+ having turned out to be a non-event and Power7 with no set date, SPARC64 "Venus" an HPC part never likely to appear in a Sun server, and Rock even more uncertain than it was before Larry bought Sun's carcass, the pressure is off the Itanium team. With the current Integrity range selling well, especially in the blades, hp isn't exactly dying for Tukzilla in the same way Sun salesgrunts are for ANYTHING to replace UltraSPANked and SPARC64 kit.

    As a customer I'm disappointed as we did have plans around Tukzilla kit which will now have to be revised to use the current Itaniums instead, but that isn't a problem as we already know the current Itaniums can do the job. We'll just hammer a few more points of the deal to teach hp a lesson!

    Of course, there is a rumour that hp had a contingency plan. The last time Intel upset hp was when they were late getting the first dual-core Itaniums to market. As hp didn't want to wait (and they wanted to steal a march on the other Itanium vendors), they came up with the mx2 module that let you plug two single-core Itaniums into one socket. Now, wouldn't it be fun if they got tired of waiting and made a module to put two dual-core Itaniums in a CSI socket.....

  12. Anonymous Coward
    Happy

    Re: Please, no more obituaries just yet!

    HA! The HP apologist has arrived. Pretty late at that... Hilarious!

    "As a customer I'm disappointed as we did have plans around Tukzilla kit..."

    Wow, your plans must be 4 years behind as Tukwilla is about the most "late" processor in history. You keep hoping and praying that anything after Tukwilla will ever materialize. At least HP can put X64 cores into those overpriced HP boxen. Of course, that leaves you out of luck with HPUX. Lucky that HP OEM's Solaris!!! Larry must have known something, huh.

  13. Anonymous Coward
    Coat

    Naming stuff after stupid places

    Lovely how companies name stuff after stupid or awful places - Intel and Tukwila, Toyota and Tacoma (crackhead capital and armpit of Washington state).

    What next, Microsoft changes name of Windows 7 to Windows Humptulips (yet another town in WA) or Intel names their new chip Boring (town in Oregon).

    Probably related to how WA sometimes names things in deliberate attempts to mock and confuse outsiders and hapless tourists who "can't pronounce" the name such as the Washington State Ferry "Puyallup" (also town, and yet another Indian tribe). Possibly partly related to neighboring Oregon's infamous "visit but don't stay" thing.

    Disclaimer: author is WA naive, er I mean native ;) since 1950-something.

    Note: Only native-born WA residents are allowed to make fun of WA; outsiders should STFU and go back to California to continue ruining that instead. ;)

    </ot>

  14. Anonymous Coward
    Dead Vulture

    re: replacing "propritary arcitectures" like SPARC?

    You are not the first by any means. Every time a technically challenged hack like TPM comments on the "fact" that SPARC is proprietary someone comments on the FACT that SPARC is actually quite open (since 1989 as a matter of fact); hence, the ability for so many SPARC clones in the wild. Sun has taken advantage of the openness of SPARC for years. Now with SPARC64 from Fujitsu, and in the past with HyperSparc (from ROSS). There are many other SPARC manufacturers and OEM's in the embedded market, especially in the government sector.

    To say that SPARC is proprietary just demonstrates the ignorance of the speaker. However, if the intention is to say that x86 is the de facto standard... and since by definition all others are not the de facto standard, all others are therefore non-standard... That could be argued, though not entirely well on the high-end of the market.

  15. asdf
    Thumb Down

    lmao

    Re: Please, no more obituaries just yet!

    Yep Matt you are right. As long as HP can keep finding retards who actually enjoy single vendor lock-in (I guess some people get their jollies being kicked in the nuts too), Itanium will linger on in a coma for another twenty years. Of course, we have all seen how this will go down, slow and painful ala DEC all over again. I guess its a good thing HP can always fall back on their ink sales.

  16. Anonymous Coward
    Stop

    @Matt

    Hey Matt, please tell us what is the planet that you live?

    HP is loosing big chunks of market share since Itanium was released.

    Don't say to us, El reg readers, that HP is not desperate.

    Yes, HP is DESPERATE! 3 years delay on Tukwila, performance lags behind x86, POWER and Sparc64, poor virtualization scaling to just 4 CPU's...

    Customers don't buy Itanium box, nor HP-UX crap for a long time...

    At least in our planet!

    Cheers!

  17. Anonymous Coward
    Dead Vulture

    "the pressure is off the Itanium team"

    "the pressure is off the Itanium team"

    Maybe from outside Intel. But inside Intel (tm), there will be questions being asked by the bean counters. What's our Return on Investment on money spent on Itanium vs RoI on Xeon. Unlike WiMax or whatever other harebrained scheme is fashionable inside Intel, money spent on Itanium competes pretty much directly with money spent on Xeon. That's the competition the Itanium folks at Intel need to worry about.

    Now obviously Intel's Itanium RoI can be increased. They could increase revenue by increasing the sales volume (sorry, not gonna happen in any meaningful way) or increasing the price. HP can pass those cost increases on, at which point it's fortunate for HP that their main Itanium customers don't have any realistic software alternative to go to, at least in the short term.

    Gravestone. Wrong label though.

  18. Anonymous Coward
    Coat

    all these comments prove...

    is there's a lot of Sun (soon to be ex Sun) folks and Sun fanboys with plenty of time on their hands...

    wonder why that is?

    mine's the one with the rock/power7 release dates in the pocket (of course they won't slip will they???)

  19. Macka

    Re: Please, no more obituaries just yet!

    --"The last time Intel upset hp was when they were late getting the first dual-core Itaniums to market. As hp didn't want to wait (and they wanted to steal a march on the other Itanium vendors), they came up with the mx2 module that let you plug two single-core Itaniums into one socket. Now, wouldn't it be fun if they got tired of waiting and made a module to put two dual-core Itaniums in a CSI socket....."--

    You really think so? Double the power consumption, double the heat, and twice the price. Don't you think that completely flies in the face of the direction the industry has been heading in for the past few years? Every new chip released increases horse power, but at the same time incorporates new advanced power management features to keep the power and heat footprint as low as possible. Then there's the technical challenges. Do you really think the current low and mid-range Integrity systems will just take a change like that without having their innards redesigned to create extra space, increase air flow and beef up the power supplies to cater for it? Sounds like a very expensive thumb in the dyke to me. Nah, I don't believe it.

    You seem to have quite a blind spot for the Nehalem NX chips too. You can't keep ignoring this you know. It's a bigger threat to Tukwilla than any of your favourite targets. So is AMD's Magny-Cours for that matter.

  20. Gerard Cannon
    Flame

    Tim, power chips have never slipped their timeline

    Tim, you may want to run a few searches on IBM power releases. I've worked for IBM since 1997 and we have NEVER slipped our timeline on Power chip releases. Every 3.5 years we release a new chip like clockwork. I don't know where you got the info that power 7 doesn't have a release date either. As usual it is slated for its 3.5 year iteration which is end of 2010. Power 6+ wasn't meant to be a speed increase. It was meant to increase capacity of existing servers as it is a quad core design vs. a dual core design for power 6. This increased capacity of our 16 way system to 32. The speed bump was just a byproduct.

    This is straight from wikipedia

    POWER3

    Main article: POWER3

    IBM introduced the POWER3 processor in 1998. It implemented the 64-bit POWER instruction set, including all of the optional instructions of the ISA (at the time), and had two floating-point units, three fixed-point units, and two load-store units. All subsequent POWER processors implemented the full 64-bit PowerPC and POWER instruction sets, so that there were no longer any IBM processors that implemented only POWER or only POWER2.

    [edit] POWER4

    Main article: POWER4

    IBM introduced the POWER4 processor, the first in the GIGA-Series, in 2001. Like the POWER3, it was a full 64-bit processor, implementing the full 64-bit PowerPC instruction set; it also had the AS/400 extensions, and was used in both RS/6000 and AS/400 systems, replacing both POWER3 and the RS64 processors. There was a new ISA release at this point called the PowerPC 2.00 ISA, which added a couple of extensions to the ISA, such as a version of mfcr which also took a field argument.

    [edit] POWER5

    Main article: POWER5

    POWER5 MCM with four processors and four 36 MB external L3 cache modules.

    IBM introduced the POWER5 processor in 2004. It is a dual-core processor with support for simultaneous multithreading with two threads, so it implements 4 logical processors. Using the Virtual Vector Architecture, several POWER5 processors can act together as a single vector processor. The POWER5 added more instructions to the ISA.

    The POWER5+ added even more instructions, bringing the ISA to version 2.02.

    [edit] POWER6

    Main article: POWER6

    POWER6 was announced on May 21, 2007. It adds VMX to the POWER series. It also introduces the second generation of IBM ViVA, ViVA-2, which is the biggest change to the POWER series of processor since the transition from POWER3 to POWER4. It is a dual-core design, reaching 5.0 GHz at 65 nm. It has very advanced interchip communication technology. Its power consumption is nearly the same as the preceding POWER5, whilst offering doubled performance.

    [edit] POWER7

    Main article: POWER7

    Currently in development at IBM, POWER7 will be the first of the Peta-Series. It's projected for release around 2010 and has been selected by DARPA as a potential processor to be used in their Peta-FLOPS SuperComputer. In the early 2000s, IBM submitted their proposal and received $53 million from DARPA to continue to participate in the challenge; in 2006, IBM received $244 million to build a petaFLOPS computer for DARPA.

  21. Matt Bryant Silver badge
    Happy

    RE: Sunshiners

    RE: Re: Please, no more obituaries just yet!

    "..... Wow, your plans must be 4 years behind as Tukwilla is about the most "late" processor in history....." Nope, I hear there are still some Sunshiners that don't remember something called UltraSPARC V. And Rock is later than Tukwila anyway.

    ".... At least HP can put X64 cores into those overpriced HP boxen...." Hp doesn't need to, they have the most popular x64 range of servers. They have had for something like 12 consecutive quarters. Lets not mention the tiny and dying Sun x64 bizz, it would only upset you lot.

    ".....Of course, that leaves you out of luck with HPUX. Lucky that HP OEM's Solaris!!!..." Well, someone has to offer to save those mugs that bought Slowaris on SPARC. And that means more money for hp, not Soreacle.

    "....Larry must have known something, huh." That would be Larry that didn't want the Sun hardware bizz, and still hasn't commited to a roadmap for ANY Sun product? Yeah, Larry knows exactly which Sun products are going to survive, it's just the poor saps like you that bought the Sunshine that have yet to realise it ain't going to be much.

    RE: lmao

    What you forget is that as well as hp-ux, hp offer full solutions for OpenVMS, Linux and Windows. A tad wider and more portable than AIX or Linux on Power, or just several incompatible flavours of Slowaris on different SPARC or x64. So much for vendr lock-in.

    RE: @Matt

    ".....HP is loosing big chunks of market share since Itanium was released....." Oh dear, another Sunshiner with reality issues. Please go read Gartner or IDC marketshare figures before you embarass yourself further. Last year Itanium was the ONLY enterprise CPU to grow whilst Power stayed flat and SPARC declined (again). Check out the high-end enterprise (yes, I'll say it again, the high end where all the lucrative services and support pull-through occurs) and you'll see Integrity is gaining in marketshare at the expense of i/pSeries and especially at the expense of Sun SPARC.

    "....poor virtualization scaling to just 4 CPU's..." You really are a special kind of stupid Sunshiner, aren't you? Please go read up on the hp Partitioning Continuum, they have offered hardware (cell-based) and software (resource-, SLA- or percentage-based) partitioning for years before Sun copied Xen to get zones/containers. And that Integrity partitioning includes for multiple OSs (hp-ux, OpenVMS, Windows and Linux), whereas Sun can't even provide the same level of partitioning in just Slowaris! Do yourself a favour and at least try and read up on your competition before you post anything quite as ignorant again.

    ".....Customers don't buy Itanium box, nor HP-UX crap for a long time...At least in our planet!...." Looking at the Gartner and IDC figures I'll have to surmise that you are posting from Mars and that you have ignored the advice to use a helmet and are breathing the oxygen-lacking Martian atmosphere. With any luck that should hopefully mean no repeat posts!

    RE: "the pressure is off the Itanium team"

    You don't understand Intel's strategy. Intel are very focused on their competition. Xeon has no competition for funds in Intel, it is the undisputed money machine. Itanium is a bonus which also allows a two-pronged startegy. Xeon has always been the attack vector against RISC from below, whilst Itanium has been that attack from above. Combining technologies like CSI and DDR3 make Itanium cheaper for Intel to develop than Power is for IBM or ... oh, hold on, there isn't going to be any more SPARC development! Anyway, it's that two-pronged attack that killed Sun's SPARC business - Sun couldn't make SPARC compete downwards against Xeon, and couldn't get it to perform for even close to the same price as Itanium in the high-end. With SPARC dead, Intel now only have to continue applying the same strategy to Power. As long as Itanium keeps Power focused on the high-end, IBM is forced to accept Xeon, but if Itanium went then IBM could back off on Power development and also look at a Power-Lite or Cell variant to take on Xeon.

    Where this dealy is bad news for hp (not Intel) is that a lot of customers will do what we will probably do - insist hp gives us clauses in new purchases so that we get a generous buy-back if we take more of the current Integrity boxes now rather than waiting for the new ones. Hp will want the server sales. A few years back when we wanted dual-core Itaniums and Intel delayed them, hp gave us the mx2 dual-CPU modules at "two-for-one" price becasue we kicked up a fuss. I'm practicing my stroppy face now! ;)

    /I hear pink is the new Slowaris colour - as in pink slips!

  22. kit
    Alert

    The end in nigh

    So does it mean the end for itanic is nigh, the news for the late delivery as well as the low tech 65nm for the construction of the chip (mind you, it is certain other server chip manufacturers will sure migrate to 45 or even 32nm architect in 2010) gives us a strong hint that Itanium is on the line of live and dead. HP must now be in deep thought as to their next move if itanium supply in the future is in doubt.

  23. Anonymous Coward
    Happy

    Re: RE: Sunshiners

    I agree there are some Sun fanboys commenting here, but I also think there are more X86/X64 fans out there commenting on the demise of Itanium. Now on to your patently false statements:

    "Nope, I hear there are still some Sunshiners that don't remember something called UltraSPARC V. And Rock is later than Tukwila anyway."

    You obviously don't remember Whitefield, and another canceled Itanium I can't remember right now... Tukwila was originally expected in 2006 and will not be here until 2010. That's over 3 years for those keeping track. Rock wasn't even announced as a possible product until 2006, so it's impossible to say that Rock is later than Tukwila. As a matter of fact, Rock will be released in servers before Tukwila is even released as a CPU.

    "...tiny and dying Sun x64 bizz..."

    I don't know about dying, though in comparison to HP it is rather tiny. Sun's X64 business is actually the fastest growing X64 business in the industry. I'm too lazy to look up the numbers, but I think that Sun owns ~5% of the X64 server business. When you consider that this is achieved in only a few years, I think that is rather remarkable.

    "...and still hasn't commited to a roadmap for ANY Sun product?"

    Well, by law Sun and Oracle have to work as separate entities until the purchase is complete. Oracle cannot, by law, make specific commitments to something they don't own yet. However, Oracle has said that they will expand investment in SPARC and expand the market for SPARC. These comments were made both verbally and in the SEC submission. The Fed's don't take kindly to being lied to, so there must be at least a bit of truth in these statements.

    "...just several incompatible flavours of Slowaris on different SPARC or x64. So much for vendr lock-in."

    Heck Matt. There are more applications on Solaris x64 than there are on all OS's that run on Itanium put together. How, can you say that there is more choice on Itanium? Also, what exactly do you mean by "incompatible flavours of Solaris"? This statement is fairly nebulous. I think you're actually starting to believe your own FUD.

    "...before Sun copied Xen to get zones/containers."

    I almost just ignored this nonsense, but couldn't help myself. This just shows your continued ignorance. I'm sure you'll ignore my response here as it points out how wrong you are... XEN was first released in 2003. Zones/Containers came from the Solaris Resource Manager that used Fair Share time scheduling in Solaris 2.6+. This was back in 2001. Sun then slowly started bundling pieces of SRM into Solaris starting with Solaris 9 (if I remember right). They then continued with this move by turning SRM into the idea of Zones, later to be called containers.

    "...and also look at a Power-Lite or Cell variant to take on Xeon."

    Well, seeing how IBM has already said that they are going the Cell route for all future Power (after Power7), this doesn't seem like much of a prognostication...

    Matt, you used to actually have a few facts to share. You seem to be resorting to lies and half-truths.

  24. Anonymous Coward
    Anonymous Coward

    "... enterprise ... lucrative services and support pull-through"

    I don't know about HP UX or NonStop but it's been reliably reported that VMS Engineering (and the folks in Engineering that end up being the backstop for Support) has been shut down in the US, with decades of experience being replaced by new recruits in India.

    How well do you think that will go down in Enterprise markets, Matt?

    Would HP do that if they were serious about VMS in "the high end where all the lucrative services and support pull-through" is ?

    Why will it be any different for HP-UX?

    Non-Stop was, is, and likely will be a special case with special economics.

    http://www.theregister.co.uk/2009/05/12/openvms_to_india/

  25. Anonymous Coward
    Flame

    Re: Matt

    "We're definitely not going to exit the hardware business," Ellison was quoted as saying.

    'Nuf said.

  26. Anonymous Coward
    Flame

    Itanium is dead.

    Itanium is dead. Intel has decided to try to push X64 into the high-end. The fact is that Intel does not understand the high-end. HP _did_ understand it, but Intel took over complete control of development of Itanium, so there is little to no focus there.

    SPARC is dead.

    Itanium is dead.

    POWER will continue, as IBM has the capital to push it forward.

    Try to do anything with X64 past 4 proc in a single chassis and learn what it's like to pull your hair out. Zero diagnosability. Flat to negative scalability (hence the need for VM)...

  27. Anonymous Coward
    Dead Vulture

    You will see the windjammer chipset support x86 before Itanium

    Not only with the delay of Tukwila kill Itanium, but HP will release windjammer that will support up to 64 8core Nehalem chips in a scale up architecture.

    only idiots buy montvale which is only .06GHz higher than montecito and HP never updated the SX2000 chipset to support he faster front side bus.

    Basically HP customers have been making do with a late chip which finally made it to market in 2006.

  28. Anonymous Coward
    Stop

    *Itanium Business is *NOT* Growing*

    Folks,

    Please get your facts straight on Itanium and what you claim to be marketshare gains. Based on IDC's latest quarterly server tracker, HP's Itanium revenue and unit increases are not at the expense of IBM or Sun *but at the expense of HP's own EOLed PA-RISC and Alpha*.

    If you look at the IDC #s from 2004, HP did ~$5.4B in RISC (PA-RISC & Alpha) sales and then in 2008, that declined to ~$1.3B (a loss of ~$4.1B). During that same time, in 2004, EPIC(Itanium) was at ~$.97B and increased to ~$4.6B (an increase of ~$3.6B). So in all, HP did a reasonable job of converting a majority of those PA-RISC/Alpha sales to Itanium, but clearly HP has NOT completely replaced those lost sales and has not replaced any IBM/Sun sales.

    HP Revenue in 2004 as per IDC

    RISC $5,423

    EPIC $969

    total $6,391

    HP Revenue in 2008

    RISC $1,303

    EPIC $4,604

    total $5,907

    Same is true for # of systems shipped:

    HP Server Units Shipped in 2004

    RISC 67,132

    EPIC 22,059

    total 89,191

    HP Server Units Shipped in 2008

    RISC 9,685

    EPIC 56,138

    total 65,823

  29. Anonymous Coward
    Stop

    Regarding Power slippage "Tim, power chips have never slipped their timeline"

    RE: I've worked for IBM since 1997 and we have NEVER slipped our timeline on Power chip releases.

    You need to verify your facts! Power 6 was expected to be released in 2006 as per IBM's roadmaps (slide 20 in here: http://www.cs.utexas.edu/users/cart/arch/fall03/KallaSlides.pdf), and it only started showing up in late 2007 but not across the product line till mid 2008-2 years later!

  30. Anonymous Coward
    Anonymous Coward

    "Zero diagnosability. Flat to negative scalability "

    Are you talking about New Improved Xeon or the real AMD64 stuff? Are they really the same anyway? They certainly look pretty much the same once you put a lowest common denominator OS on them (definitely Windows, maybe Linux) but that doesn't mean it'll stay that way forever. Either way, if you have facts (links, even) to back up your assertions folk might be interested to read them.

  31. Matt Bryant Silver badge
    Happy

    Dear Sunshiners

    Thank you for providing such good humour, I nearly bust a gut laughing so hard at your rantings! Your venom for Intel is hilarious - for years you denied Intel could make a chip that could replace SPARC, then finally you caved and tried to bolster the crumbling Slowaris bizz with a bizarre series of flip-flops between Slowaris x86, Linux, and back to Slowaris x86 again, but all with the hated x86/x64. All the time, you kept up this amusing snootiness, acting like having to sell Xeon (and Opteron) was some dirty task forced on you by the shareholders that "just didn't understand the beauty of SPARC", when the reality was Xeon was your only hope of survival, and consequently you stuffed the survival bit. You never forgave Intel for eating your bizz from below, and you definately never forgave hp for partnering with Intel on Itanium. You acted like Intel was some backward bunch of spivs and amateurs and not the largest CPU vendor in the business, and tried to pretend that failing SPARC was soooo superior, and would "win everywhere if only customers weren't so cheap and stupid". Today, Sun is a zombie company waiting for Larry to hand over the cash so he can butcher it up, SPARC is as good as dead, and the only Sun servers likely to survive long are the very Xeon kit you despise so much. You guys are just so tragic!

    /SP&L

  32. Anonymous Coward
    Stop

    Zero core performance improvement!!!!!!!!!!!!!!!!!!

    Ouch!!!!

    "Tukwila is tracking to 2X performance vs its predecessor chip."

    It has twice the number of cores which means it will have zero performance improvement per core.

    Oracle is already getting ready to increase the core factor to .75 from .5 and customers will find they are paying 50% more for the same performance.

    Mr. MOJO

  33. David Halko
    Thumb Down

    Gordon Haff, Illuminata, Intel, Tukwila, and SPARC

    'According to Gordon Haff, an analyst at Illuminata who broke the Tukwila delay story, Intel has issued a statement to his fellow analysts... "...allow Tukwila systems a greater opportunity to gain share versus proprietary RISC solutions including SPARC"'

    I almost find Intel's statement regarding being late on the 4 core Itanium providing "greater opportunity to gain share" to be as ridiculous as calling (open source, open specification, multiple vendor) SPARC proprietary.

    A single vendor, closed source, non-consortium instruction specification choice of Itanium is certainly proprietary.

    Such parroted statements from Gordon Haff and Illuminata reduce their credibility - they should have called Intel on their choice of inaccurate terminology and not bothered to repeat it.

    Perhaps, because these analysts don't understand what they are parroting, a 4 core Itanium will not be delayed or will not show up at all to the gun fight in 2010.

  34. Anonymous Coward
    Flame

    Re: Dear Sunshiners

    Hilarious. Matt can't defend any of his lies, so he attacks Sun. Classic.

  35. Matt Bryant Silver badge
    Happy

    RE: Re: Dear Sunshiners

    ".....Matt can't defend any of his lies, so he attacks Sun. Classic." What lies? And by past efforts, you Sunshienrs still go on posting the same bilge even when you have been soundly shown up for the idiots you are. And the truth is it is Intel-envy that seems to drive a lot of you rabid Sunshiners. "We're so much smarter than them, it's just not fair that they are so much more successful!" LOL!

    I just loved the Sunshiner posting selective sale figures from 2004! Please look at the figures for 2007, 2008 and so far in 2009 - they show the decline in Sun sales and the matching rise in hp Integrity quite clearly. The whole post is almost as funny as those SAP figures Novatose posted, comparing four-year-old Itanium and brand new Opteron. You guys are so desperate your FUD is really getting slack and feeble. Here's a "big lie" for you to disprove - Sun has declined from a $200bn company to being bought for cash equivalent of $5.6bn. Meanwhile, hp's value has risen from $40bn to $100bn, during the same period in the same economic conditions. Now, if Sun's sales haven't declined and hp hasn't benefited, please explain Sun's crash and hp's rise (big hint - don't tell us it's all ink sales, us customers don't believe you).

    Here's hoping Larry drags out the purchase so we get a few more chuckles out of the frantic Sunshiners before he axes the lot!

    /SP&L

  36. Anonymous Coward
    Happy

    Re: Matt B being a HP Sales Bod / Troll again.

    Matt, the article & comment thread should be about the fact Itanium is delayed (again) and has yet to creep past the 2-core figure but don't let that worry you, The other vendors will carry on prepping their 8, 12 & 16 core CPU's so I'm sure you & HPUX will be just fine when the 4 core version pops out. Maybe the fact it could be you who needs to change their technical skills to keep on the gravy train is a bit too much to bear. Anyway, onwards.....

    Your lies were pointed out above & opened one by one so let's reverse the question & keep it on one topic shall we? Can you answer which part of this part of one sentence was not a lie? :

    "before Sun copied Xen to get zones/containers."

    Matt oh Matt, anything for a cheap shot at the "Sunshiners", you fully fit the description of a troll.

  37. Anonymous Coward
    Unhappy

    matt bryant can't think or read. showing paranoia. not bright.

    anonymous ~~~ HP Revenue in 2004 as per IDC... HP Revenue in 2008... HP Server Units Shipped in 2004... HP Server Units Shipped in 2008... clearly HP has NOT completely replaced those lost sales and has not replaced any IBM/Sun sales

    matt bryant ~~~ I just loved the Sunshiner posting selective sale figures from 2004! Please look at the figures for... 2008

    anon included multiple architectures, revenue, units, and two years of figures - anon was not very selective but very thorough

    anon included 2008 sales figures that matt couldn't read in half the comment - matt's not very observant

    anon compared hp against against ibm as well as sun - anon was not very sunshining but was brighter than matt

    matt can't think or read - why is matt so paranoid?

  38. Anonymous Coward
    Dead Vulture

    Matt B and the Ink manafacturer loving...

    Here Matty Matty Matty.... There there, be a good boy will you while I spoon feed you some figures to soothe your confused mind : http://www.theregister.co.uk/2009/05/28/idc_q12009_server_nums/

    "According to IDC, IBM accounted for 31 per cent of the Unix (real Unix, not Unix plus Linux) pie, compared to Sun Microsystems and Hewlett-Packard, which had 27.7 percent of the Unix pie each."

    So El Reg even seems to be able to unpick your lies, your beloved HP isn't doing as well as you'd like to believe (unless you count the Ink or the x64 wintel sales which dropped 28%). HP's Unix Sales are on Par with Sun, below IBM and it looks like with this 4-core Itanium knockback (2 years late did I here you say) I'd guess the numbers will fall next time around.

    A lot of system managers & accountants must be eyeing the last PA-RISC stuff and thinking about whether to risk (no pun intended) HPUX/Itanium or go Linux, AIX or Solaris instead. Why chance a temporary port to Itanium before a possible CPU death forces a port to x64 (but they said NO WAY! Matt, right?) or HP drop it completely. They just sacked another load of developers across Europe, mainly Bristol based so how much life is left in HPUX and VMS?http://www.theregister.co.uk/2009/05/29/hp_labs_closure/

    If I read the El-reg article correct it looks like OpenVMS is losing marketshare to Mainframes (or just generally not being replaced) so thats another Itanium knock for you to consider.

    Wakey Wakey Matt, x64 looks strong, PowerPC is really kicking ass @5ghz and while Sparc doesn't look certain either Oracle have plenty of cash if they choose to grow it's developement. My money says Intel wants Itanium dead, hence the gravestone....

  39. Matt Bryant Silver badge
    Happy

    Dear Bemused Sunshiners

    RE: Re: Matt B being a HP Sales Bod / Troll again.

    ".....The other vendors will carry on prepping their 8, 12 & 16 core CPU's...." Which vendors? Not Sun, that's for sure. Fujitsu? Only for HPC - hardly a threat to anyone! AMD? Last I saw they were still working on getting six cores out the door. IBM? Power is now Itanium's only real competition.

    RE: matt bryant can't think or read. showing paranoia. not bright.

    Obviously a better reader and far brighter than you, Sunshiner, seeing as I didn't hitch my career to a dying OS from a dea company. Enjoy your pink slip when Larry gets his hands on the Sun bizz!

    RE: Matt B and the Ink manafacturer loving...

    <Yawn> I see you haven't even tried to answer my challenge to explain Sun's demise and hp's rise if Sun did everything right and hp just sold ink. A bit of reality avoidance going on? Here's a hint - SPARC is dead, Slowaris on SPARC has been dead for years, the Linux clone Slowaris x86 is on lifesupport, whilst Windows, Linux, hp-ux, Itanium and Xeon march on making profits. I'm betting you failed to notice the last five winning products can be bought from hp, a market leader, whilst the zombified products all come from Sun, a dying has-been. Oh, and hp do also make a lot of money out of ink (it's that profitable innovation and diversification thing that Sunshiners like you just don't understand).

  40. Anonymous Coward
    Thumb Down

    matt the paranoid still can't spell and read

    anon ~~~ PowerPC is really kicking ass @5ghz and while Sparc doesn't look certain

    matt ~~~ Dear Bemused Sunshiners... I see you haven't even tried to answer my challenge to explain Sun's demise

    reading the anon comment made it clear - he was an ibm advocate, not a sunshiner as matt incorrectly indicated. also, being purchased does not necessarily mean demise, somtimes companies survive like to live another day like teradata while their brands survive like linksys.

    matt ~~~ Enjoy your pink slip when Larry gets his hands on the Sun bizz!

    matt ~~~ SPARC is dead

    matt said elsewhere that he did not write fud. more lies from matt bryant.

    article ~~~ "Sun Microsystems and Hewlett-Packard, which had 27.7 percent of the Unix pie each."

    matt ~~~ on SPARC has been dead for years

    by matt's standards, itanium on hp has been dead for years

    why is matt so obsessed with negative comments about sun?

    let me help out poor matty who can't spell - s o l a r i s - ha ha ha ha!

  41. Matt Bryant Silver badge
    Happy

    RE: matt the paranoid still can't spell and read

    "....matt said elsewhere that he did not write fud. more lies from matt bryant...." What, you seriously believe Larry is going to go the whole hog on Sun and not make any cuts? Are you breathing just pure helium? Even before the purchase there were a slew of Sun cuts in the offing (6000 if I remember rightly), and you think Larry - who didn't originally want the hardware business - is going to just let it carry on untouched? Or that customers are going to look at what the analysts are saying and not come to the conclusion that their company's future wouldn't be better off not using SPARC and Slowaris? Big hint - the massive decline in Sun sales is because, just like in our company, the last has already happened and will accellerate the decline in Sun server sales. The U in FUD is uncertainty, and Sun just made a massive heap of uncertainty with their botched sale. That means any Slowaris consultants and admins are going to be chasing a smaller and smaller market. My suggestion would be to get out now and get some cross-training. RedHat is always popular....

    "....by matt's standards, itanium on hp has been dead for years...." Look at where the marketshare is - hp-ux and Integirty in the high-end, where there is money to pay for admins and expensive consultants, so lots of opportunity for those with Linux and Windows on Itanium and hp-ux accreditations. And hp's Integrity sales are growing. Meanwhile, Suns recent sales have predominantly been in the webserving arena, where the drive is minimal cost, one-admin-per-hundred-servers, poor pay and limited opportunities, and Linux and Windows on cheap Xeon are eating you up. And Sun's server sales are declining. Only a complete dolt - or a Sunshiner - would be chosing to continue with Slowaris skills alone.

    "....why is matt so obsessed with negative comments about sun?...." Look in the mirror - why do you guys immediately attack anyone that says anything realistic about Sun, why do you bash Intel with such venom, and why do you fear Linux so much? Because you know your comfy little gravy train, milking your employer for your Slowaris skills, is shortly going to grind to a halt, no matter how much you whine and scream otherwise.

    "....let me help out poor matty who can't spell - s o l a r i s - ha ha ha ha!" The laughs on you, Sunshiner. As has been discussed here on other posts, the term Slowaris was coined by Sun's own customers. But don't worry, soon the only time you'll hear the terms Sun, Solaris, Slowaris or SPARC will be on some Discovery Channel program into failed IT businesses. It will probably be a short mention before they get onto more important historical technologies like Netware or OS/2.

    /Enjoy! SP&L.

This topic is closed for new posts.

Other stories you might like