back to article Unisys threatens Itanium with death

The Itanium server chip from Intel needs all the allies it can get these days, particularly after the latest in a string of delays for the quad-core "Tukwila" chip. But it doesn't look like Intel can count on Unisys to be a particularly enthusiastic supporter of Tukwila when it does get out the door. Unisys is touting a new …

COMMENTS

This topic is closed for new posts.
  1. Algis Petraitis
    Dead Vulture

    Itanium becomes too risky to invest

    Intel inability to keep the old promise for Itanium to become socket compatible wih Xeon was the largest failure with this chip. Nobody will be able to keep developing chipsets for so low volumes.

    Second failure is that Itanium became so much technologically delayed after Xeons in terms of nanometers.

    I am HP customer and before I was investing into Integrity rx8600 series for SQL loads (Itanium based). Now I am investing into DL785 Proliants (8processor / 32core Opteron server with 512GB of RAM). Performance is the same or better, cost is just a fraction, and electricity / cooling costs is just a fraction, compared to Itaniums. Also, Opteron server is smaller in size. HP probably is not so happy, but I am happy as a customer, and I stayed as HP customer because they have proper product.

    With increasing chip performances there will be less and less applications that do not fit into 8P Opteron or Xeon machine. Five years ago billing system for 1 million telecom users was not able to run without special expensive machines. Now ordinary 8P x64 server is more than enough.

    Itanium chip architecture was good, but implementation - failed.

  2. Anonymous Coward
    Pirate

    Waiting for Matt Bryant...

    No comment

  3. John PM Chappell
    Unhappy

    Victim of its own success?

    In many ways Itanium peaked too early; it's arguably a superior 64 bit architecture to AMD64 but the way that AMD was able to match or even better the performance of comparable 80x86 processors running 32 bit code and also run 64 code pretty much killed a lot of Itanium's potential (since its emulation was significantly slower than native code and usually far slower than an 80x86 processor at a similar price point or even native performance point).

  4. Matt Bryant
    Pirate

    Tukzilla

    Niagara and Rock are worldbeaters and hp supports Solaris on ProLiant with no Transitive needed. With Itanium dead and H-SUX, well... sucking... HP has hedged their bet and should be safe with Solaris on Proliant. No need for Unisys - they don't even show up on Gartner's Magic Quadrant.

  5. Anonymous Coward
    Anonymous Coward

    Webster disapears

    And we get Matt.

    Ps Webster was nutty on all the forums he was on .

  6. Anonymous Coward
    Dead Vulture

    re: Tukzilla

    Well, it seems that only HP is making money from Itanium... But have they really? Has HP made their money back from the $5 Billion infusion from 2006? I doubt it. Intel does not break out Itanic from their other profitable chips, so it's impossible to know whether Intel is making money on Itanic. From the less than enthusiastic comments made by Intel execs such as the most common comment that Itanic has "... not met expectations" I would say that Intel would drop Itanic in a heart beat if HP was not pumping money into it...

    Times running out for Itanic.

  7. kain preacher

    I forgot to mention

    Suns corporate HQ , is a renovated , closed state mental asylum .

    Agnews Developmental Center is a psychiatric and medical care facility, now located in San Jose, California.

    When the west campus closed, the use of the land was the subject of local controversy. In April 1997, it was announced the state would sell an 82.5-acre (33.4 ha) parcel of the campus to Sun Microsystems for use as its corporate headquarters and R&D campus.

  8. Anonymous Coward
    Anonymous Coward

    re: I forgot to mention

    huh?

  9. Matt Bryant Silver badge
    Happy

    RE: Tukzilla, re: Tukzilla, and RE: re: I forgot to mention

    RE: Tukzilla

    Ah, the Sunshiners are out early, I see, along with the fake Matt Bryant login.

    "Niagara and Rock are worldbeaters..." in the over-priced and incapable chip segment. So niche they make Itanium look mainstream - oh, hold on a sec, Itanium is. Especially in the lucrative UNIX high-end where Niagara can't play and Rock will never get to play, Sun have given up that for the SPARC64 chips they used to FUD so much, and their customers are so unimpressed they're shifting to hp Integrity as shown by the marketshare figures. Dream on, Sunshiners.

    "....and hp supports Solaris on ProLiant with no Transitive needed...." Yes, and Sun makes zero revenue from any ProLiant running Slowaris, and no revenue from using IBM's Transitive. Not that there are many, indeed Sun's own Galaxy sales show Sun's own customers prefer their x86 with Linux by a ratio of five-to-one, even when Sun give Slowaris away for free. You see the customers prefer a real Linux like Red Hat to a poor Linux clone like Slowaris x86 and its tiny application base.

    ".... With Itanium dead and H-SUX, well... sucking... " Would that be the hp-ux and Itanium team that are trouncing Sun in the high-end? Face it, Sun would just love that kind of "sucking", they might even be able to make a profit from it despite Ponytail.

    "....HP has hedged their bet and should be safe with Solaris on Proliant....." Well, more likely they'll be safe with their number one server vendor, number one disk storage vendor, number one software vendor and number one print vendor positions. Sun is number one in only one area - losing money. What you need to realise is that Sun is making no money from Slowaris x86 or Open Slowaris and still losing money overall, so they won't be able to innovate anything new in Slowaris to try and keep it anywhere even close to Linux or hp-ux or Windows or AIX. Do tell where you think the money is going to come from, because the analysts all think Sun is currently junk stock.

    ".... No need for Unisys - they don't even show up on Gartner's Magic Quadrant...." Because they are just about bust. In fact, Ponytail should dredge up some capital and try and buy them seeing as they actually make far better x86 kit than the uninnovative, me-too Galaxy kit. But then Ponytail has his heart set on the cloud, so making intelligent hardware purchases takes second place to buying up software companies he hasn't got the time or money to integrate into a viable stack. Even if Southern Asset Management let him stay around long enough, he'll run out of cash and have to break Sun up long before his cloud "strategy" can get to the point it makes a profit. Better learn some Linux, Sunshiners, it's your only chance.

    RE: re: Tukzilla

    "Well, it seems that only HP is making money from Itanium... " Looking at their annual report the answer is a very obvious "yes", and they look like they'll keep on making money on Itanium long after Sun have killed Rock.

    "....But have they really? Has HP made their money back from the $5 Billion infusion from 2006?...." Try using some maths, check the annual reports, add up the totals including the money hp make from support and servcies from all those Integrity sales eating up the Sun base, and you'll see a profit. Oh, hold on a sec, don't use that Sunshiner maths you use for "benchmarking", otherwise you'll be out be a factor of aout ten.

    "....I doubt it....." I doubt you have the first clue about what you're trying to FUD, but carry on 'cos it is very amusing.

    "....Intel does not break out Itanic from their other profitable chips, so it's impossible to know whether Intel is making money on Itanic...." No, but hp do, and they alone make Intel and hp a profit on Itanium, so any bonus from other Itanium players is still more money than Sun have made from vapourware Rock, micro-niche Niagara and Open Slowaris.

    ".....From the less than enthusiastic comments made by Intel execs such as the most common comment that Itanic has "... not met expectations" I would say that Intel would drop Itanic in a heart beat if HP was not pumping money into it...." Yeah, a Sunshiner would advise dropping the chip currently mopping up business in the high-end UNIX space, where all the money is, but then Sunshienrs aren't very good at commercial decisions. It's so hard to relate to the real world from those ivory towers, which is why Sun customers are dropping Sun from their business critical high-end UNIX space - if you don't trust a company you don't run their kit for the jobs your company depends on to survive.

    "....Times running out for Itanic." Looking at Sun's market cap compared to hp's, and the trend in the market figures for the UNIX high-end, it looks more like time is running out for you Sunshiners. What snakeoil will you hype when Sun is gone?

    RE: re: I forgot to mention

    What I think Kain Preacher is implying is that it is somehow fitting that Sun's HQ should be the site of an old loonybin. This is of course both low and insulting to the previous incumbents, comparing their unfortunate mental illnesses to Ponytail's self-inflicted incompetence and lack of vision.

  10. Matt Bryant Silver badge
    Happy

    On the topic of the thread....

    Anyone else think it's ironic that the Sunshiners are gloating at the idea of one Intel chip possibly taking share from another? See how far Sun have fallen - they somehow see it as good news that Itanium's only real competitors are Power and the next gen Xeons. They have forgotten the key fact - hp are number one in both Xeon and Itanium shipments, so either way is win-win for hp and more lose-lose for Sun. One of the reasons Unisys are in such a bad state is that vanilla x86 systems have scaled up and eaten a lot of their marketshare. A lot of that eating up has been by hp's ProLiant line, none of it has been by Sun's Galaxy. It's all so funny when you remember back to such sterling Sun statements as "Solaris on SPARC and nothing else!" Schwartz was wrong then and he's still wrong now, and it all means more humblepie in store for the Sunshiners.

  11. Ken Hagan Gold badge

    IA64 is the SCO of the hardware world.

    ia64 was still born. Until a few years ago, 99% of all sales were a single box that some fool had stuffed with several thousand chips. The name lives on simply because there's a nutter who can't take a hint and enjoys throwing vast sums of money at it.

  12. Anonymous Coward
    Dead Vulture

    RE: Matt's inability to break his mind free from Sun

    The article's about Unisys and Itanium, why is Matt talking about Sun? What a loon...

    How has Sun affected your life so much that you see everything in the guise of Sun? Can you not see Itanium and HP as proper and right in their own accord without having to prop them up on the pitfalls of the lowly Sun? Perhaps not... There is something sad about the lack of self worth and confidence that one must insult others to make oneself feel vindicated and right. Matt comes off as the type that would put others work down to make his own look better.

    A quick aside... I've seen the Agnews campus, and it is quite beautiful. The county of Santa Clara required that Sun keep as much of the campus open to the public as possible and Sun did that and more. I agree, however, that it is quite humorous that Sun's HQ is located in a mental asylum.

    As far as myself, I really don't see Intel keeping Itanium going forever. Intel relies on volume sales and Itanium is nowhere near volume, per Intel's own admission. Other vendors such as IBM and to a lesser extent Sun can get away with a lower volume chip. Intel must make money from the actual fabrication of the chip, while IBM, HP and Sun can make money from the servers, services, software, etc... HP made a big mistake getting rid of PA-RISC and going with Itanium. I'm not sure when Intel will drop Itanium, but I am certain it will happen... This is my opinion, and only based on observation (not fanciful wishing as a certain HP Fudster is wont to do).

  13. Bill
    Happy

    re: Waiting for Matt Bryant...

    aaaannnnddddd.... there he is!

    Someone dare question Matt's view of the world! How dare they!

  14. Rob Dobs
    Dead Vulture

    oh god more Sun haters

    "prefer a real Linux like Red Hat to a poor Linux clone like Slowaris x86"

    Linux Clone?? omg you really don't have any idea what you are talking about, and I've realized this is like arguing religion with a fanatic. I would love to go on about SUN smarts for selling x86 boxes, or why having HP sell your OS is a win, but whats the point?

    as stated above what does this have to do with SUN? Take your HP promoting SUN hating shill argument elsewhere please. I no longer see a point in arguing such blather, with a single minded fan boy / SUN hater. SUN to me seems a great company and that is why I invested in them and put my money where my mouth is, time will prove which of us has our head on straight. If you disagree STF up and go buy some overpriced HP stock, which honestly at today's market prices still wouldn't be a terrible investment. Though its bad investment advice to buy heavily in the company you work for, if they go bust you loose job and retirement... not good.

  15. Daniel B.
    Boffin

    The problem is one: "Intel Inside".

    I'd rather use PA-RISC (even when its nearing its end-of-life) than use anything made by Intel.

    Too bad that most desktop systems have been saddled with the x86 arch. RISC is much much better.

  16. Anonymous Coward
    Pirate

    What Matt Bryant is and is not good at

    "Sunshienrs aren't very good at..."

    Matt Bryant isn't very good at spelling.

    Matt Bryant is very good at being a hypocrite. It was just days ago when Matt was criticizing someone for personal attacks and yet he is at it again.

  17. Anonymous Coward
    Pirate

    Matt Bryant can't spell, again.

    Matt spelled "Solaris" incorrectly again.

    Not very sharp.

  18. Andy Masland

    A different point of view

    NEC remains 100% committed to Itanium, as do the sponsors and members of the Itanium Solutions Alliance.

    The Alliance's blog provides a different perspective on the Itanium platform, including its success in Windows and Linux environments...http://blog.itaniumsolutions.org/

  19. Anonymous Coward
    Pirate

    Funny

    Let's see how long people stay at HP. Maybe the brightest will get asylum at Sun

    http://www.proletar.com/

    Sun 4.4 Stars - HP 1.7 Stars (of five)

  20. David Halko
    Go

    Helping Matt Understand

    "Anyone else think it's ironic that the Sunshiners are gloating at the idea of one Intel chip possibly taking share from another?" --- Intel is a Sun Partner. The x64 based chip is a Sun Solaris supported product. The Itanium is not a Sun Solaris supported architecture. Fast Intel x64 CPU's make fast Sun platforms. There is no irony.

    "the lucrative UNIX high-end where Niagara can't play" --- That's OK, where Niagra is playing, it seems to be beating up competing vendors in a pretty lucrative market.

    http://www.sun.com/servers/coolthreads/t5440/benchmarks.jsp?display=3#3

    "Sun makes zero revenue from any ProLiant running Slowaris" --- That is completely incorrect. There is support, services, and other software license fees. Solaris is generally faster than other operating systems on identical server platform. Solaris gives better visibility through DTrace to allow people to make applications run faster under Solaris than other operating systems on identical server platforms. S

    "Would that be the hp-ux and Itanium team that are trouncing Sun in the high-end?" --- It was just last year when Sun published high-end SMP platform results outrunning IBM & HP by 2x and 2.75x, respectively. Sun was "trouncing" Power and Itanium. ;-)

    http://www.sun.com/servers/highend/m9000/benchmarks.jsp?display=6#6

    "Sun is number one in only one area" --- For most of 2007 & 2008, SUN had the fastest:

    1 CPU socket performance

    http://www.spec.org/cgi-bin/osgresults?conf=rint2006&op=fetch&proj-COMPANY=256&proj-SYSTEM=256&proj-CORES=256&proj-CHIPS=256&critop-CHIPS=0&crit-CHIPS=1&proj-CORESCHP=256&proj-THREADS=0&proj-CPU=0&proj-CPU_MHZ=0&proj-CPUCHAR=0&proj-NCPUORD=0&proj-PARALLEL=0&proj-BASEPTR=0&proj-PEAKPTR=0&proj-CACHE1=0&proj-CACHE2=0&proj-CACHE3=0&proj-OCACHE=0&proj-MEMORY=0&proj-OS=0&proj-FS=0&proj-COMPILER=0&proj-HWAVAIL=0&crit2-HWAVAIL=Jan&proj-SWAVAIL=0&crit2-SWAVAIL=Jan&proj-COPIES=256&proj-PEAK=256&proj-BASE=256&proj-400PEAK=0&proj-400BASE=0&proj-401PEAK=0&proj-401BASE=0&proj-403PEAK=0&proj-403BASE=0&proj-429PEAK=0&proj-429BASE=0&proj-445PEAK=0&proj-445BASE=0&proj-456PEAK=0&proj-456BASE=0&proj-458PEAK=0&proj-458BASE=0&proj-462PEAK=0&proj-462BASE=0&proj-464PEAK=0&proj-464BASE=0&proj-471PEAK=0&proj-471BASE=0&proj-473PEAK=0&proj-473BASE=0&proj-483PEAK=0&proj-483BASE=0&proj-LICENSE=0&proj-TESTER=0&proj-SPONSOR=0&proj-TESTDAT=0&crit2-TESTDAT=Jan&proj-PUBLISH=256&critop-PUBLISH=-1&crit2-PUBLISH=Nov&crit-PUBLISH=2008&proj-UPDATE=0&crit2-UPDATE=Jan&dups=0&duplist=COMPANY&duplist=SYSTEM&duplist=CORES&duplist=CHIPS&duplist=CORESCHP&duplist=THREADS&duplist=CPU&duplist=PARALLEL&duplist=BASEPTR&duplist=PEAKPTR&duplist=CACHE1&duplist=CACHE2&duplist=CACHE3&duplist=OCACHE&duplist=COPIES&dupkey=PUBLISH&latest=Dec-9999&sort1=PEAK&sdir1=-1&sort2=COMPANY&sdir2=1&sort3=CORESCHP&sdir3=1&format=tab

    2 CPU socket performance

    http://www.spec.org/cgi-bin/osgresults?conf=rint2006&op=fetch&proj-COMPANY=256&proj-SYSTEM=256&proj-CORES=256&proj-CHIPS=256&critop-CHIPS=0&crit-CHIPS=2&proj-CORESCHP=256&proj-THREADS=0&proj-CPU=0&proj-CPU_MHZ=0&proj-CPUCHAR=0&proj-NCPUORD=0&proj-PARALLEL=0&proj-BASEPTR=0&proj-PEAKPTR=0&proj-CACHE1=0&proj-CACHE2=0&proj-CACHE3=0&proj-OCACHE=0&proj-MEMORY=0&proj-OS=0&proj-FS=0&proj-COMPILER=0&proj-HWAVAIL=0&crit2-HWAVAIL=Jan&proj-SWAVAIL=0&crit2-SWAVAIL=Jan&proj-COPIES=256&proj-PEAK=256&proj-BASE=256&proj-400PEAK=0&proj-400BASE=0&proj-401PEAK=0&proj-401BASE=0&proj-403PEAK=0&proj-403BASE=0&proj-429PEAK=0&proj-429BASE=0&proj-445PEAK=0&proj-445BASE=0&proj-456PEAK=0&proj-456BASE=0&proj-458PEAK=0&proj-458BASE=0&proj-462PEAK=0&proj-462BASE=0&proj-464PEAK=0&proj-464BASE=0&proj-471PEAK=0&proj-471BASE=0&proj-473PEAK=0&proj-473BASE=0&proj-483PEAK=0&proj-483BASE=0&proj-LICENSE=0&proj-TESTER=0&proj-SPONSOR=0&proj-TESTDAT=0&crit2-TESTDAT=Jan&proj-PUBLISH=256&critop-PUBLISH=-1&crit2-PUBLISH=Oct&crit-PUBLISH=2008&proj-UPDATE=0&crit2-UPDATE=Jan&dups=0&duplist=COMPANY&duplist=SYSTEM&duplist=CORES&duplist=CHIPS&duplist=CORESCHP&duplist=THREADS&duplist=CPU&duplist=PARALLEL&duplist=BASEPTR&duplist=PEAKPTR&duplist=CACHE1&duplist=CACHE2&duplist=CACHE3&duplist=OCACHE&duplist=COPIES&dupkey=PUBLISH&latest=Dec-9999&sort1=PEAK&sdir1=-1&sort2=COMPANY&sdir2=1&sort3=CORESCHP&sdir3=1&format=tab

    4 CPU socket performance

    http://www.spec.org/cgi-bin/osgresults?conf=rint2006&op=fetch&proj-COMPANY=256&proj-SYSTEM=256&proj-CORES=256&proj-CHIPS=256&critop-CHIPS=0&crit-CHIPS=4&proj-CORESCHP=256&proj-THREADS=0&proj-CPU=0&proj-CPU_MHZ=0&proj-CPUCHAR=0&proj-NCPUORD=0&proj-PARALLEL=0&proj-BASEPTR=0&proj-PEAKPTR=0&proj-CACHE1=0&proj-CACHE2=0&proj-CACHE3=0&proj-OCACHE=0&proj-MEMORY=0&proj-OS=0&proj-FS=0&proj-COMPILER=0&proj-HWAVAIL=0&crit2-HWAVAIL=Jan&proj-SWAVAIL=0&crit2-SWAVAIL=Jan&proj-COPIES=256&proj-PEAK=256&proj-BASE=256&proj-400PEAK=0&proj-400BASE=0&proj-401PEAK=0&proj-401BASE=0&proj-403PEAK=0&proj-403BASE=0&proj-429PEAK=0&proj-429BASE=0&proj-445PEAK=0&proj-445BASE=0&proj-456PEAK=0&proj-456BASE=0&proj-458PEAK=0&proj-458BASE=0&proj-462PEAK=0&proj-462BASE=0&proj-464PEAK=0&proj-464BASE=0&proj-471PEAK=0&proj-471BASE=0&proj-473PEAK=0&proj-473BASE=0&proj-483PEAK=0&proj-483BASE=0&proj-LICENSE=0&proj-TESTER=0&proj-SPONSOR=0&proj-TESTDAT=0&crit2-TESTDAT=Jan&proj-PUBLISH=256&critop-PUBLISH=-1&crit2-PUBLISH=Jan&crit-PUBLISH=2009&proj-UPDATE=0&crit2-UPDATE=Jan&dups=0&duplist=COMPANY&duplist=SYSTEM&duplist=CORES&duplist=CHIPS&duplist=CORESCHP&duplist=THREADS&duplist=CPU&duplist=PARALLEL&duplist=BASEPTR&duplist=PEAKPTR&duplist=CACHE1&duplist=CACHE2&duplist=CACHE3&duplist=OCACHE&duplist=COPIES&dupkey=PUBLISH&latest=Dec-9999&sort1=PEAK&sdir1=-1&sort2=COMPANY&sdir2=1&sort3=CORESCHP&sdir3=1&format=tab

    SUN still has the fastest 64 socket performance

    64 CPU socket performance

    http://www.spec.org/cgi-bin/osgresults?conf=rint2006&op=fetch&proj-COMPANY=256&proj-SYSTEM=256&proj-CORES=256&proj-CHIPS=256&critop-CHIPS=0&crit-CHIPS=64&proj-CORESCHP=256&proj-THREADS=0&proj-CPU=0&proj-CPU_MHZ=0&proj-CPUCHAR=0&proj-NCPUORD=0&proj-PARALLEL=0&proj-BASEPTR=0&proj-PEAKPTR=0&proj-CACHE1=0&proj-CACHE2=0&proj-CACHE3=0&proj-OCACHE=0&proj-MEMORY=0&proj-OS=0&proj-FS=0&proj-COMPILER=0&proj-HWAVAIL=0&crit2-HWAVAIL=Jan&proj-SWAVAIL=0&crit2-SWAVAIL=Jan&proj-COPIES=256&proj-PEAK=256&proj-BASE=256&proj-400PEAK=0&proj-400BASE=0&proj-401PEAK=0&proj-401BASE=0&proj-403PEAK=0&proj-403BASE=0&proj-429PEAK=0&proj-429BASE=0&proj-445PEAK=0&proj-445BASE=0&proj-456PEAK=0&proj-456BASE=0&proj-458PEAK=0&proj-458BASE=0&proj-462PEAK=0&proj-462BASE=0&proj-464PEAK=0&proj-464BASE=0&proj-471PEAK=0&proj-471BASE=0&proj-473PEAK=0&proj-473BASE=0&proj-483PEAK=0&proj-483BASE=0&proj-LICENSE=0&proj-TESTER=0&proj-SPONSOR=0&proj-TESTDAT=0&crit2-TESTDAT=Jan&proj-PUBLISH=256&critop-PUBLISH=-1&crit2-PUBLISH=Jan&crit-PUBLISH=2009&proj-UPDATE=0&crit2-UPDATE=Jan&dups=0&duplist=COMPANY&duplist=SYSTEM&duplist=CORES&duplist=CHIPS&duplist=CORESCHP&duplist=THREADS&duplist=CPU&duplist=PARALLEL&duplist=BASEPTR&duplist=PEAKPTR&duplist=CACHE1&duplist=CACHE2&duplist=CACHE3&duplist=OCACHE&duplist=COPIES&dupkey=PUBLISH&latest=Dec-9999&sort1=PEAK&sdir1=-1&sort2=COMPANY&sdir2=1&sort3=CORESCHP&sdir3=1&format=tab

    Of course, CPU benchmarks bounce around with new releases, but it is clear that Sun is very often #1 in performance benchmarks for many months at a clip. Sun is very competitive. Let's see what Sun comes out with, this year.

    "Yeah, a Sunshiner would advise dropping the chip currently mopping up business in the high-end UNIX space, where all the money is" --- Isn't 64 socket the "high-end"? From the published benchmarks, it seems Sun is "mopping up" HP. Perhaps the other poster was advising HP drop the Itanium chip, since "high-end" Sun servers provide nearly 140% higher throughput, socket per socket, in the high-end.

    http://www.spec.org/cgi-bin/osgresults?conf=rint2006&op=fetch&proj-COMPANY=256&proj-SYSTEM=256&proj-CORES=256&proj-CHIPS=256&critop-CHIPS=0&crit-CHIPS=64&proj-CORESCHP=256&proj-THREADS=0&proj-CPU=0&proj-CPU_MHZ=0&proj-CPUCHAR=0&proj-NCPUORD=0&proj-PARALLEL=0&proj-BASEPTR=0&proj-PEAKPTR=0&proj-CACHE1=0&proj-CACHE2=0&proj-CACHE3=0&proj-OCACHE=0&proj-MEMORY=0&proj-OS=0&proj-FS=0&proj-COMPILER=0&proj-HWAVAIL=0&crit2-HWAVAIL=Jan&proj-SWAVAIL=0&crit2-SWAVAIL=Jan&proj-COPIES=256&proj-PEAK=256&proj-BASE=256&proj-400PEAK=0&proj-400BASE=0&proj-401PEAK=0&proj-401BASE=0&proj-403PEAK=0&proj-403BASE=0&proj-429PEAK=0&proj-429BASE=0&proj-445PEAK=0&proj-445BASE=0&proj-456PEAK=0&proj-456BASE=0&proj-458PEAK=0&proj-458BASE=0&proj-462PEAK=0&proj-462BASE=0&proj-464PEAK=0&proj-464BASE=0&proj-471PEAK=0&proj-471BASE=0&proj-473PEAK=0&proj-473BASE=0&proj-483PEAK=0&proj-483BASE=0&proj-LICENSE=0&proj-TESTER=0&proj-SPONSOR=0&proj-TESTDAT=0&crit2-TESTDAT=Jan&proj-PUBLISH=256&critop-PUBLISH=-1&crit2-PUBLISH=Jan&crit-PUBLISH=2009&proj-UPDATE=0&crit2-UPDATE=Jan&dups=0&duplist=COMPANY&duplist=SYSTEM&duplist=CORES&duplist=CHIPS&duplist=CORESCHP&duplist=THREADS&duplist=CPU&duplist=PARALLEL&duplist=BASEPTR&duplist=PEAKPTR&duplist=CACHE1&duplist=CACHE2&duplist=CACHE3&duplist=OCACHE&duplist=COPIES&dupkey=PUBLISH&latest=Dec-9999&sort1=PEAK&sdir1=-1&sort2=COMPANY&sdir2=1&sort3=CORESCHP&sdir3=1&format=tab

    Less bigotry is needed in the community and more objectivity. Multiple vendors and architectures demonstrate a healthy industry - the community does not need fewer choices, but more.

  21. kain preacher

    Agnews

    What I think Kain Preacher is implying is that it is somehow fitting that Sun's HQ should be the site of an old loonybin. This is of course both low and insulting to the previous incumbents, comparing their unfortunate mental illnesses to Ponytail's self-inflicted incompetence and lack of vision.

    Or that Matt a sun troll is a bit looney him self

  22. Macka
    IT Angle

    Tukwilla .va. Beckton

    Itanium's fate really rests on how fast Intel and HP can deliver Tuwilla into the market and convert that into real server sales. If it slips again, it's going to run into the Beckton Core i7 release schedule, and that will really hurt its chances.

    The Beckton Core i7, 8-core (16 thread) chip with 2.3 Billion transistors, 24MB L3 cache, up to 2 TB of DDR3-1600 memory, 4 memory channels and 4 Quick Path Interconnect links is a real monster. It's due early in 2010 and Intel can't drag their feet and wait for AMDs to bring their 6 core or 12 core MCM chips to market, so Beckton will have to release on time.

    I struggle to see how Itanium is going to compete with all the high octane x86 contenders coming down the pipe. Beckton should blow it away on performance and cost. So where is the value in Itanium now? I just don't get it anymore.

  23. Anonymous Coward
    Happy

    Re: Agnews

    I think they're both loony. Perhaps they should have a stroll together and discuss jelly beans.

  24. Anonymous Coward
    Unhappy

    Re: Matt

    Ahh, come on guys. Let's leave poor Matt alone. He's obviously suffering from relevance envy.

    Besides, Matt has such a bad name on this site, no one really takes him seriously... He's like the old guy that yells at the kids to get off his lawn.

    'Hey you kids! Get off my lawn!" I can see it now... Really quite endearing. I wonder if he has pink flamingos and white plastic lawn chairs...

  25. Matt Bryant Silver badge
    Happy

    RE: Sunshiners

    "The article's about Unisys and Itanium, why is Matt talking about Sun?...." Because, as with any article that mentions Itanium, the same Sunshiner trolls immediately start posting FUD. Whay are you guys so fixated with bashing Itanium and hp? Oh, I forgot - your jobs are on the line what with Integrity eating up Sun's high-end marketshare, because everyone knows that's where the money is for contractors and admins too. In the webfarm niche Sun is stuck in the money is poor and the jobs go to PFYs straight out of college, not old Slowairs hackers.

    RE: Helping Matt Understand

    And there goes the Sun salesguide again. Honestly, David, if customers don't believe that schpiel when Sun salesgrunts are throwing it at them, do you really think anyone is going to believe it when you post it here? It's a forum - try posting an opinioned argument and not just regurgitating Sun material.

    RE: Funny

    "Let's see how long people stay at HP. Maybe the brightest will get asylum at Sun...." Hmmmm, I wonder if those figures were from before the recent Sun job-cuts. I hear hp employees in the past have made pay sacrifices to avoid job reductions, looks like they're more loyal than Sun employees. And why on earth would anyone want to go work for Sun? The company's financial position is so bad you'd get a pink slip five minutes after you walk in the door!

  26. Anonymous Coward
    Anonymous Coward

    re: Tukwilla .va. Beckton

    It's all about reliability. The x86 architecture does nothing to address this. There's no reason they couldn't though...

  27. David Halko
    Linux

    @Macka: Itanium Struggle is Valid

    "I struggle to see how Itanium is going to compete with all the high octane x86 contenders coming down the pipe. Beckton should blow it away on performance and cost. So where is the value in Itanium now? I just don't get it anymore." --- There is not a significant software base for Itanium to draw from, which differentiates Itanium it from Intel Windows x86, Intel Windows x64, SPARC Solaris, Intel Linux, or even new Open Source variants like Intel Solaris?

    I think a lot of people are struggling with how Itanium will compete with x64. Will HP & Intel both decide to kill Itanium? With the global recession, I don't see how both companies can pump so much money into Itanium, when that is not the major source of revenue from either company.

    SPARC is THE major revenue resource for SUN, so this 64 bit architecture seems to have a rather secure immediate future - especially since SPARC has been 64 bit for almost a decade, built a significant application base, and is still showing dynamic architecture growth (the first octal core 64 bit processor with octal crypto engines thrown in for free.)

    I can see the vision behind x64, SPARC, and Power in the future with servers - I am not sure if Itanium has any long term viability. I just don't see very long word instruction sets as having enough commercial appeal to differentiate it.

  28. Anonymous Coward
    Dead Vulture

    Re: Tukwilla .va. Beckton

    Agreed. Itanium has no stickiness... Users would and could move to X64 without ever looking back. HP-UX has so few applications that even the lowly Solaris X86 has double the number of applications. Itanium is slower than SPARC64 on the high end and Niagara on the low to midrange. I'm not even mentioning POWER! What does Itanium have to keep it alive going forward? It has two major vendors (HP and Intel) and it appears that only one of them is making any money from it. Unless some other vendor gets the religion, Itanium has nowhere to go but down. Tukwila is a nonstarter.

    Matt, your dislike for Sun is legend, and it makes any opinion that you posit without merit. You constantly "defend" Itanium by bashing Sun. This is a losing argument, and anyone that has any level of logic should be able to see that. But instead, you say that Itanium is good because Sun is bad... Childish logic...

  29. Matt Bryant Silver badge
    Pirate

    RE: @Macka: Itanium Struggle is Valid & Tukwilla .va. Beckton

    RE: David Halko

    "....There is not a significant software base for Itanium to draw from, which differentiates Itanium it from Intel Windows x86, Intel Windows x64, SPARC Solaris, Intel Linux, or even new Open Source variants like Intel Solaris?...." <Yawn> Dave, your Sun FUD is out of date yet again. Sunshiners were desperately pushing that line in 1998, it's been over a decade and hp-ux not only has a vast number of applications, it has all the business critical ones, which is why it is cleaning up in the UNIX high-end. If hp-ux didn't have the applications then the dmoniation of the high-end by Integrity simply wouldn't be happening. What you also forgot to mention was that hp-ux on Itanium has a wie range of drivers which allows it to meet many differeing custoemr requirments - Slowaris x86 has a cripplingly poor set of drivers that limits the amount of hardware it can run on.

    "....SPARC is THE major revenue resource for SUN...." Which is why Sun is still making a loss, and has a market cap that threatens it with delisting. Intel and hp have much healthier financial positions and make a profit from Itanium. I know understanding finances and business is just not a strong point with Sunshiners so I'll spell it out for you - hp and Intel have the money and reason to continue with Itanium; Sun does not have the money and can't decide whether it is a software company or a hardware company.

    "....can see the vision behind x64, SPARC, and Power in the future with servers - I am not sure if Itanium has any long term viability...." Which is why you have a lowly Mcjob as a Sun marketing droid. Hp Integrity is domintaing the UNIX high-end, this means major corporations are going with Itanium and hp-ux for business critical solutions. These customers are both very demanding and very careful - they don't buy without a lot of consideration and reassurance that what they are buying is both vaild and has a future. They employ the top people in IT to make those decisions for them, people with obviously a far higher degree of knowledge, experience and vision than yourself. Bleat all you like but the marketing figures expose your lies - when the customer has to bet his buisness on the job, he chooses hp-ux on Integrity, and not SPARC.

    RE: Anonymous Coward

    "Agreed. Itanium has no stickiness... Users would and could move to X64 without ever looking back....." Which just goes to show you know nothing about compilers, chip designs or the market. From the start, hp and Intel designed Itanium to be an easy porting platform. An example is that is equal endian, meaning it can feed registers from the highest or lowest value first. Older chip designs were one or the other, making their code incompatible even if they came from the same families. The AMD and Intel x64 desings are different endian to SPARC, which is why Slowairs x86 and SPARC binaries are not compatible, you don't just need to recompile you need to rewrite them, making porting off SPARC Slowaris onto Slowairs x86 much more difficult. Nehalem is still not equal endian and does not have the large amount of registers Itanium has, so whilst it will be a much better chip than anything coming out of Sun or Fujitsu, it will still not be the equal of Itanium, and still not be as good a programming or porting platform as Itanium. As for the rest of your drivel about "Itanium has nowhere to go", please read my reply to David above.

    "....Matt, your dislike for Sun is legend...." What you Sunshiners fail to undertand is us customers don't want Sun kit. We are not buying it not because we have an axe to grind but because Sun have let us down in the past and don't provide a future we consider ether convincing or one we want to bet our businesses on.

    ".....and it makes any opinion that you posit without merit....." That's so Sun - don't listen to the customers, don't listen to complaints, blank out any argument that contradicts your core beliefs, just keep pushing Slowairs on SPARC regardless. Sun needs to wake up and start listening to the customers a LOT more.

    "....You constantly "defend" Itanium by bashing Sun....." Actually if you look at the forums you'll see I usually post in RESPONSE to FUD from Sunshiners like yourself. What's the matter, you don't want people to hear both sides of the story?

    ".....This is a losing argument, and anyone that has any level of logic should be able to see that...." Seems like the market disagrees, but then I suppose you don't want that mentioned either.

    "....But instead, you say that Itanium is good because Sun is bad... Childish logic..." Actually, you guys are always first to post the Sun whitewash and your level of childishness is both extreme and amusing. I will try to explain how us users in the real world make purchasing decisions for high-end UNIX kit, as I think will help you understand why us users get annoyed by your Sunshiner posts.

    Your complete argument is that any choice should ALWAYS go on Solaris on SPARC, without any further thought, simply becasue you say it is better than sliced bread. This is simply incompetent - the prime reason we want a platform choice is because it offers chances to reduce costs both in the immediate purchase and ongoing operational costs. A rule of thumb is that a system will costs four to five times as much in opex over its lifecycle as the immediate purchase price, so a wrong choice can be very expensive. Especially if that system is running a service your busimess depends on, in which case a wrong decision cold mean your business fails too.

    We invite the vendors to help us with formulating propoals for new projects so as to both garner discounts and get the best possible projections. Yes, this means we talk to IBM, Sun and hp. We ALWAYS warn vendors we are not interested in hearing FUD about the other vendors, just info about their own kit. We have been burned in the past by vendor FUD dragging out purchases, which is why we insist on try-before-you-buy shootouts - nothing proves a system like running real data through it in your own environment. It is my experience that Sun are the worst culprits for continually trying to FUD competitors. I don't pretend my expereince is a complete marekt view, but it's what I see. Which is why I post counters to your online FUD - I suspect you are doing it not out of the customer's interests but your own, and there is nothing you want less than for them to get a real view.

  30. Bill
    Happy

    Re: RE: @Macka: Itanium Struggle is Valid & Tukwilla .va. Beckton

    "hp-ux on Itanium has a wie range of drivers which allows it to meet many differeing custoemr requirments - Slowaris x86 has a cripplingly poor set of drivers that limits the amount of hardware it can run on."

    Wow Matt. That's a stupid comment even coming from you. I was trying to stay away from all things Matt today, but that is one stupid comment. HP-UX only runs on one vendors hardware, that being HP's. To say they have a "wie" range of drivers is just stupid. The last I saw, Solaris was certified on over 500 different servers. How many different servers does HP-UX run on?

    shaking head... pause... giggle.... shaking head again...

  31. Matt Bryant Silver badge
    Happy

    RE: Re: RE: @Macka: Itanium Struggle is Valid & Tukwilla .va. Beckton

    ".....That's a stupid comment even coming from you.... HP-UX only runs on one vendors hardware, that being HP's. To say they have a "wie" range of drivers is just stupid. The last I saw, Solaris was certified on over 500 different servers. How many different servers does HP-UX run on?...."

    And into the trap walks gormless Bill! I couldn't wait for you to tell me how many incomplete and incompatible bits of kit Slowaris runs on. Hp has carefully produced solutions and drivers for all their customer needs, and you can look them up on the hp websites and check EXACTLY what is supported with what, which server with which card with which storage device with which firmware, etc, etc. Sun has problems getting even it's own stuff to work! Try looking up the Sun support matrix for Slowaris x86 cluster on Dell Poweredge with Falconstor and Adaptec iSCSI cards - oh wait, you can't! You have to take pot luck and cross your fingers and just hope it all works. Compare to even Linux like Red Hat which has far better support and far better documentation, and customer trust from years of use in real business environments. The real giggle is Sun's desperate attempt to copy Linux and then an even better laugh is when they try and play it off against a real enterprise solution like hp-ux on Integrity.

  32. Jesper Frimann

    Re:David Halko

    What a nice cherry picking, why don't you do the same lookup and compare equal number of threads or equal number of cores, or even equal price ?

    1 CPU socket performance

    Today Niagara is being trashed (like everyone else) by Intel Core i7. All the niagara results use 63 Threads, which means that unless your app can utilize many threads you won't really get to the throughput in the benchmark. Furthermore running 63 threads compared to for example 8 to get the same work done will use a fair chunk more memory. Anywhere from 8 times to 1.5 times.

    2 CPU socket performance

    This is pretty much the same as for the 1 CPU socket performance, there aren't really Core i7 results, but the hex core Xeons are pretty much equal to Niagara.

    4 CPU socket performance

    This is pretty much the same as for the 1 CPU socket performance, there aren't really Core i7 results, but the hex core Xeons are pretty much equal to Niagara.

    8 CPU socket performance

    The M5000 is trashed by a factor of 2 by a power 570 with half the cores.

    16 CPU socket performance

    The M8000 gets beaten by both power 570 (with half the cores) and a hex Xeon based Unisys by nearly 50%.

    32 CPU socket performance

    The M9000 gets beaten by the power 595 with half the number of cores with 70%.

    64 CPU socket performance

    Well the M9000 with 256 cores and 64 sockets manages to beat a 64 core 32 socket p595 with a astonishing 8%. That is 4 times the cores to get 8% extra. And it gets a solid trashing by a Altrix with equal number of cores.

    And You should really check out the reality behind the SUN benchmark quotes that you present, I mean it's hard to take SUN marketing posts seriously when they quote old stream benchmarks where they benchmark themselves up against old systems. Try having a look at the real facts:

    http://www.cs.virginia.edu/stream/top20/Bandwidth.html

    // Jesper

  33. Anonymous Coward
    Dead Vulture

    IDC numbers are just in HP's Unix business is down 10% in 2008

    With Unisys dumping itanium and SGI refusing to confirm/deny Tukwila support you will see a huge drop in HP Unix and subsequently Itanium sales in 2009

  34. Matt Bryant Silver badge
    Happy

    RE: Re:David Halko

    Careful, Jesper, they'll start accusing you of having a plastic flamingo and white lawn chairs! They somehow think that works in lieu of a technical riposte.

  35. Matt Bryant Silver badge
    Happy

    RE: IDC numbers are just in HP's Unix business is down 10% in 2008

    "With Unisys dumping itanium and SGI refusing to confirm/deny Tukwila support you will see a huge drop in HP Unix and subsequently Itanium sales in 2009" Whilst a refreshingly short bit of Sunshine, how exactly do you come by that conclusion? SGI and Unisys are nothing to do with hp, so if anything both events offer MORE opportunities for hp Integrity sales. Do you Sunshiners even stop a half-second to think before you post your twaddle?

    Presuming you're just reading from The Reg article "Server sales cratered in Q4, says IDC", you no doubt didn't want to post the bit about how the only growing sector of server sales is blades, up 16.1%, with hp having 54.8% of that. Sun's share of the blades cake is too small to merit more than the passing chuckle. Or that whilst the economic downturn has knocked hp's server sales down 10.1% to $3.91bn, Sun's server sales are worse hit down 14.1% at only $1.25bn. So hp aren't dropping sales as fast as Sun, and still making more than three times as much revenue. Looks like more red ink for Ponytail and more value off the Sun market cap!

  36. Anonymous Coward
    Happy

    Re: RE: Re:David Halko

    Matt, regarding Jespers comments. He makes arguments, unlike you who make grand statements of opinion and accuse others of the same acts as you. The funny part is that Jesper makes no mention of Itanium as even being in the discussion. Itanium is the goat of the IT industry.

    As far as your argument on Solaris vs. HP-UX and supported hardware. Yes, the x86 industry is rife with incompatibilities. That is what you get in an the x86 market. If you want to compare compatibility of hardware and software, then compare all of the X64 hardware that Sun sells along with the SPARC hardware that Sun sells. There are double the number of systems that Sun sells to run Solaris as HP sells to run HP-UX. Of course, after today HP will have announced full support for Solaris, so even you will be able to run Solaris. You are ignorant in the purest definition of the same.

  37. Dan JOhnson

    What is with the guy "Matt Bryant?"

    Is he for real? Does he actually have a job? It would appear not. I certainly wouldn't have time to post so many long rants. What's worse, he doesn't know what he's talking about.

    No doubt about it, Itanium is in serious trouble. Intel has said that all the money made on Itanium so far has been reinvested into the development of next versions. It has become the albatross around their neck. Only HP pushes Itanium these days. It's enough for now to keep Intel building the things, but probably not for long. Sales of Itanium are too low to justify anything after Tukwila, so Intel is going to have to dump the chip. That puts HP in a very uncomfortable position. They are pushing Integrity Superdomes hard. But if customers are faced with a dead-end processor after having invested big bucks into new Superdomes, HP can kiss the high-end UNIX market goodbye.

  38. Anonymous Coward
    Dead Vulture

    Re: HP-UX

    Using Matt the HP Fangirl Bryant's logic, Solaris is much better than HP-UX because Gartner says so:

    http://www.hp.com/hpinfo/newsroom/awards/hp2354.pdf

    http://mediaproducts.gartner.com/reprints/sunmicrosystems/article4/article4.html

    Solaris is *Very Positive*

    HP-UX was just upgraded to just *Positive*

    Obviously Solaris is the greatest and HP-UX is worse than Amiga. Again, this is per Matt the HP Fangirl Bryant's logic.

    P.S. This logic comes from the fact that Matt the HP Fangirl Bryant constantly quotes Gartner as being all knowing and dependable.

  39. David Halko
    Go

    @Jesper Frimann--- cherry picking

    "What a nice cherry picking, why don't you do the same lookup and compare equal number of threads or equal number of cores, or even equal price" --- because that is not what was being contested, Jesper.

    Clearly, when Matt Bryant said, "Sun is number one in only one area" --- he was wrong and I demonstrated with clarity by citing benchmarks to prove it over years. I demonstrated it using sockets per platform, regardless of what vendor, how many cores a vendor had, and how many threads a vendor had.

    Sure, I can also say that SUN has more threads per CPU per SMP chassis than any other vendor on the market, that would be a true statement.

    I could also say that there have been years on end which SUN had been #1 in price-performance in various midrange, and demonstrated it with dozens of links.

    There was a FUD statement, that was clearly wrong, and I demonstrated it. That was the point.

    "Today Niagara is being trashed (like everyone else) by Intel Core i7" --- If you compare crypto performance, used in web servers, Intel Corei7 does not hold a candle to 1 year old Niagra. With Integer and floating point, 1 year older Niagra is a little behind. Niagra is still clearly competitive, and is not "being trashed (like everyone else)".

    If you are looking for single socket performance for crypto with 10 Gigabit Ethernet - it may not be the best choice for an embedded system (with lots of components to fail, high temperature, and not necessarily the best encrypted throughput while running particular applications.)

    The Intel Core i7 is a nice chip, it has it's problems with some architecture aspects with throughput, but it is still a nice chip. Niagra has problems with single threaded applications, but it is still a nice chip.

    You will also note that I had taken that into consideration in my posting, indicating that CPU performance bounces around with new releases. Really, that is OK, since SUN sells Intel based platforms. This does not demonstrate a problem for open vendors like SUN, where they are now CPU agnostic. All their software runs on multiple architectures.

    People just have to just get away from all the bigotry and just enjoy the benefits of innovation that various vendors bring to the marketplace!

  40. Matt Bryant Silver badge
    Pirate

    RE: Dan JOhnson & David Halko

    "s he for real?...." Actually, no, I'm just a voice in your head. You'd best go book a session with the shrink, otherwise I might stay in here and really make you suffer....

    "....Does he actually have a job?...." Well, as voices in your head go, I have a very good job, thanks, though it is very echoey in here what with the large amounts of empty space

    ".....I certainly wouldn't have time to post so many long rants....." Well, it doesn't take five minutes to formulate replies to the pointless FUD posted by you and your fellow Sunshiners. After all, I'm in your head, I see your fear, and I already know when you're lying.....

    /psych off.

    "....No doubt about it, Itanium is in serious trouble....." Yeah, being the number one chip in the lucrative UNIX high-end is such a pain.

    "....Intel has said that all the money made on Itanium so far has been reinvested into the development of next versions....." Really? I've never heard anything of the sort. But then at least Intel has the money to put into development, whereas Sun's dwindling reserves and continual losses means Sun doesn't have the money to invest in any development for long, hence their desperation to get Slowaris supported by other server vendors before they have to pink slip the whole Sun hardware organistaion.

    ".....Sales of Itanium are too low to justify anything after Tukwila, so Intel is going to have to dump the chip...." Now what have we told you Sunshiners about how just because you wish it was so it doesn't make it happen? Intel has the winning hand, it has Xeon cleaning up in the low-to-mid server bizz and Itanium for everything above that, so it makes strategic sense for Intel to maintain Itanium too keep the pressure on IBM (they don't have to worry about Sun anymore). Itanium is the only option to Power and IBM's mainframes, areas Intel would dearly like to feast on, and only Itanium gets them up there. Nehalem will be very good, but still not good enough to take on Power. The high-end has become a two-horse race and neither horse is Sun's.

    "....But if customers are faced with a dead-end processor after having invested big bucks into new Superdomes, HP can kiss the high-end UNIX market goodbye....." But seeing as Superdome sales are growing, and hp is dominating the high-end UNIX space, it would seem hp don't have to worry at all. Face it - as Rock, if it ever arrives, can only scale to eight sockets, leaving Sun reliant on the uncertain future of SPARC64 in the M-series servers, Sun has kissed goodbye to the high-end for good. And the announcement that Slowairs will be sold on hp ProLiant means Sun has just kissed goodbye to the low-to-mid-end server bizz too. Like I warned you a while back, you better start learning Linux fast!

    RE: David Halko

    "....because that is not what was being contested, Jesper....." No, you said Slowaris was fastest all the time in all situations, and then posted a load of cherrypicked, out-of-date and invalid benchmark results, which Jesper then showed up nicely.

    "....There was a FUD statement, that was clearly wrong, and I demonstrated it. That was the point....." No, you merely showed for a fact that Sun can't compete on performance anymore, and hasn't been able for a while, as shown by Jesper's figures and my own posts.

    "....With Integer and floating point, 1 year older Niagra is a little behind. Niagra is still clearly competitive, and is not "being trashed (like everyone else)"....." Naiagara lags today's Xeon on performance and is massively more expensive, so it is most certainly uncompetitive.

    "....The Intel Core i7 is a nice chip, it has it's problems with some architecture aspects with throughput, but it is still a nice chip. Niagra has problems with single threaded applications, but it is still a nice chip....." Yes, but businesses don't base decisions on "niceness", they want the best performance with the best reliability for the lowest price, and that means Niagara is dead in anything other than the webserving niche.

    "....Really, that is OK, since SUN sells Intel based platforms....." If Ponytail or the customers had any faith in the Galaxy range then Sun wouldn't be so desperate to sign up hp's ProLiant. The truth is Galaxy is like the rest of Sun's products - a limited range with poor features, limited management tools and from a limited range of unintegrated products. Sun just outsourced their Xeon server business to hp, IBM and Dell because those vendors have real integrated x64 ranges with a future.

    "....This does not demonstrate a problem for open vendors like SUN, where they are now CPU agnostic. All their software runs on multiple architectures....." All their software in binarily incompatible from one platform to the next. This triples the expense of developing for Slowaris, which is why developers will simply take the easy path of developing for the larger Linux base first and then making the simple port to Slowaris x64. Most won't bother making three different SPARC binaries for Niagara, Rock and SPARC64 (well, Rock is dead so they actually won't bother to make the other two). So again, Slowaris will lag the real open product, Linux.

    "....People just have to just get away from all the bigotry and just enjoy the benefits of innovation that various vendors bring to the marketplace!" Cough*kettle*cough. Well, you enjoy your Sunshiner fantasy, just don't expect the rest of us in the real world to fall in step.

  41. Anonymous Coward
    Go

    Matt B. Logic

    In the logic of Matt:

    - I've never heard that before so it must not be true.

    - You like Solaris so you must be an idiot.

    - You said something bad about HP, so you must be the enemy.

    - Because I said so.

    - I'm trusted by all the "Important" vendors so they give me all of their inner secrets,

    while you are untrusted and can never know what I know.

    There are other Mattisms, but I've gotten bored.

    Oh yeah, one other thing... Gartner likes Solaris better than HP-UX and even though Matt refuses to comment on this fact after he stated otherwise, it's still true.

  42. David Halko
    Unhappy

    @Matt Bryant - can't read

    "you said Slowaris was fastest all the time in all situations"

    It seems you can't read, either.

    It seems the only word which "fastest" appeared in the entire listing of comments from anyone was targeted specifically at number of sockets.

    We now know that Matt Bryant:

    - can't spell

    - can't read

    - can't quote

  43. Anonymous Coward
    Flame

    @Matt Bryant, just stupid

    Matt Bryant, in another article comment, claimed endian-ness was the reason why SPARC machine code was incompatible with Intel machine code.

    - does not understand computer architecture

  44. Matt Bryant Silver badge
    Happy

    RE: @Matt Bryant, just stupid

    "Matt Bryant, in another article comment, claimed endian-ness was the reason why SPARC machine code was incompatible with Intel machine code....." Actually, I claimed it was one reason why porting from SPARC to x64 was much harder than SPARC to Itanium, but don't let a little truth get in the way of your frothing rant.

    "....- does not understand computer architecture" I'm now waiting with amused interest to hear how our Anonymous Sunshiner would like to claim that having different endian CPU designs has no effect on code migration. Not holding my breath, but mainly because I'm still laughing at him.

    /and pointing.

  45. Matt Bryant Silver badge
    Happy

    RE all the personal attacks

    Aw, is someone a bit upset? It's always fun when the Sunshiners fall back on basic name-calling. Scarily, this is also the tactic I've seen used by Sun salesgrunts. All the vendors do customer analysis, they're quite open about the fact they are looking for the influencers and decision-makers, whom has the say on money and whom is the person that really needs to be onboard before a deal can be struck. What I first noticed a few years ago was that the Sun salesteam were not just using it as a means of identifying positives, but also as a way to try and negatively influence decisions. As soon as they identified someone in our business that backed another vendor or voiced doubts about Sun, the Sun salesgrunts would start making derisive comments, belittling their knowledge or experience, and trying to exclude them from the conversation. Sometimes it got to the point where Sun was not only trying to tell us who we could have on a project but how we should test systems, all with the aim of making their products look better in shootouts. At that point we told Sun to change the salesteam. Seems things haven't changed.

  46. David Halko
    Thumb Up

    @Matt Bryant -- endian-ness, compilers, code migration, and architecture

    Matt Bryant said "The AMD and Intel x64 desings are different endian to SPARC, which is why Slowairs x86 and SPARC binaries are not compatible"

    Anonymous said, "Matt Bryant, in another article comment, claimed endian-ness was the reason why SPARC machine code was incompatible with Intel machine code.

    Matt Bryant said, "I'm now waiting with amused interest to hear how our Anonymous Sunshiner would like to claim that having different endian CPU designs has no effect on code migration."

    The comment seemed to make a lot of sense to me, let me take a stab at it.

    The reason "binaries are not compatible" have to deal with operating system designer decisions to how to read binaries, compiler designer decisions with how to form binaries, and the instruction sets of the CPU's.

    For example:

    - Apple Macintosh running MaxOSX on the PowerPC architecture, Motorola 68K binaries can run just fine, because the Operating System vendor had placed the intelligence into the PowerPC based Operating System to execute 68K binaries.

    - Apple Macintosh running MacOSX on the PowerPC architecture, binaries compatible with both Intel & PPC run just fine, because the Operating System vendor had placed the intelligence into the PowerPC based Operating System to ignore the x86 portions of the binary.

    - Apple Macontosh running MacOSX on the Intel architecture, binaries compatible with both Intel & PPC run just fine, because the Operating System vendor had placed the intelligence into the Intel based Operating System to ignore the PPC portions of the binary.

    - Apple Macontosh running MacOSX on the Intel architecture, binaries not compatible with both Intel & PPC run just fine, because the Operating System vendor had placed the intelligence into the Intel based Operating System to execute the PPC portions of the binary.

    In these cases above, the endian-ness of the 68K, PPC, and Intel architectures did not affect compatibility of the binary.

    The Operating System Vendor makes a decision as to how much investment they wish to make with compatibility with binaries.

    The endian is a factor in the overall architecture, but you are clearly incorrect when you said, "...are different endian to SPARC, which is why Slowairs x86 and SPARC binaries are not compatible"

    Anonymous, "Matt Bryant... - does not understand computer architecture'

    The anonymous poster clearly caught you.

  47. Matt Bryant Silver badge
    Happy

    RE: @Matt Bryant -- endian-ness, compilers, code migration, and architecture

    Same old Dave - lot's of frothing but no answer to the question.

    "....Apple Macontosh running MacOSX on the Intel architecture, binaries not compatible with both Intel & PPC run just fine, because the Operating System vendor had placed the intelligence into the Intel based Operating System to execute the PPC portions of the binary....." Nope, Transitive licensed their emulator tech to Apple in the form of the Rosetta software, which allowed Apple to escape PowerPC and get onto the Intel bandwagon. The same tech is used in the Quick Transit software to allow Slowaris 9 binaries to be run unchanged on top of Red Hat on Itanium, which is a lot easier than the painful migration to Slowairs x86 and a darn sight faster than Slowaris on x86 or SPARC. And what does MacOS have to do with Slowaris, or do you really have so much of a muddle in your mind you think MacOS and Slowaris are somehow related? And you still didn't explain how different endian x86 and SPARC can run the same binaries, but that may be because they can't.

    Your confusion clearly caught you.

  48. David Halko
    Go

    @Matt Bryant -- endian-ness, operating systems

    Hi Matt,

    David said, "....Apple Macontosh running MacOSX on the Intel architecture, binaries not compatible with both Intel & PPC run just fine, because the Operating System vendor had placed the intelligence into the Intel based Operating System to execute the PPC portions of the binary....."

    Matt said, "Nope, Transitive licensed their emulator tech to Apple..."

    Ummm... what is the word "Nope" for? That seems like a very poorly placed transition, since you just added context which confirms my statement. I indicated the "Operating System vendor had placed the intelligence into..." and you confirmed where the OS vendor received received the technology from.

    I want to thank you for confirming my statement and further demonstrating your incorrectness regarding the first statement regarding "different endian" causing "binaries" to "be not compatible". It is great to see that you agree that your original statement was incorrect.

    Matt, "And what does MacOS have to do with Slowaris..."

    You spelled Solaris wrong again.

    It was a technological citation that the anonymous poster suggesting you do "not understand computer architecture" seemed plausible.

    Matt suggests, "you still didn't explain how different endian x86 and SPARC can run the same binaries, but that may be because they can't."

    You never asked anyone to explain that, so the phrase "still didn't" does not apply. It seems your memory is failing. I will answer your newly posed question, further demonstrating that you don't understand a decade of multi-vendor computing architecture.

    SUN had clearly not followed the same path as Apple in running "different endian x86 and SPARC" on the opposing platforms. It is clearly not because "endian" is the problem, since Apple could do it. It is because there was no business decision to do it.

    SUN had taken a different approach, to create a virtual machine (Java), and move software into the JAVA stack. Funny how SUN had clearly tackled & succeeded the "endian" issue through Java, since SUN now support Solaris SPARC, Solaris Intel, RedHat Linux,Ubuntu Linux, and MS Windows. (SUN's Java also supports HP HP-UX and IBM AIX, by the way, regardless of the "endian" issue.)

    SUN's investment in Java had made the entire non-x86 UNIX market viable. Companies like HP with OS's like HP-UX received the benefit of SUN's investment. Using Java to make software compatible between Solaris x86 and Solaris SPARC was an innovative move.

    SUN and Solaris is in a great position, to move forward with, as they Open Source everything, build very popular software (MySQL, Open Office, etc.) for everything, move internal software to platform independent Java, allow the hardware to compete internally with each other, continue to participate with Open SPARC with internal & external competition, multiple vendors sourcing SPARC chips, and not tie it's end to a single vendor. Certain SPARC platforms are earning 40% return on investment with growing revenues, other SPARC platforms are not growing revenue or profit, software is earning a return on investment that is growing revenue.

    HP, as a company, is in a very good position, being diversified with the Compaq purchase and consulting services purchase.

    HP-UX is does not have such advantages, right now. On the high-end platforms, SUN SPARC is still beating them in 64 socket; HP-UX is bound to a single vendor for Itanium; HP-UX is proprietary; HP-UX does not offer a competitive low-end Itanium platform; no one really knows if Itanium is making HP or Intel any profit. No one really knows if HP-UX is earning HP any profit.

    This being said, I am glad that HP is out there, I am glad that SUN is out there.

    Java being available on Itanium is a tremendous benefit for HP, so support of Itanium under HP-UX becomes easier for software vendors (regardless of "endian".)

    Solaris being available on Intel & AMD is a tremendous benefit for HP, so HP gets a modern high-performing OS on heavily threaded systems, where Solaris has proved to be superior to every other mainstream OS on the market, due to SPARC R&D.

    The ongoing SUN and HP partnerships have tremendously benefited the entire computing industry. HP is increasingly dependent on SUN. HP contributing to OpenSolaris indicates there may be Solaris on HP Itanium in the future. HP royalty payments to Microsoft and other companies for contributed technology would go away with a migration from HP-UX to Solaris. Porting Java apps from HP-UX to Solaris platforms on Itanium would be little more than adjusting install scripts (which have already been done, since most significant HP-UX software run under HP-UX and Solaris) and SUN had already invested in the framework for adding Itanium support.

    Continuing partnership between HP and SUN is an excellent hedge for the death of Itanium... would be a good thing for the market!

  49. Anonymous Coward
    Thumb Down

    Matt Losing His Mind: Can't Remember His Own Writing

    Anon~ "Matt Bryant, in another article comment, claimed endian-ness was the reason why SPARC machine code was incompatible with Intel machine code....."

    Byrant ~ Actually, I claimed it was one reason why porting from SPARC to x64 was much harder than SPARC to Itanium, but don't let a little truth get in the way of your frothing rant.

    Matt Bryant lied again. This was his quote:

    Byrant ~ "The AMD and Intel x64 desings are different endian to SPARC, which is why Slowairs x86 and SPARC binaries are not compatible"

    Go back to school before you post again, go back to church or temple to learn how not to lie, and get some medication to help you with your memory.

This topic is closed for new posts.

Other stories you might like