Feeds

back to article Intel 'Tukwila' born after long and painful labor

Intel officially unveiled its long-delayed Tukwila "mission-critical" server processor today - now dubbed the Itanium 9300 series - providing a few more details about the 2-billion transistor part and giving some color on why it was over three years late. Made in conjunction with this week's International Solid-State Circuits …

COMMENTS

This topic is closed for new posts.
Thumb Down

tukwila and nehalem

We thru out HP-UX for Power6 and Nehalem early last year.

It became apparent when Tukwila was canceled that it would never equal Nehalem or IBMs power product line.

We tested itanium and nehalem on the low end and EP crushed Montvale. With Tukwila only doubling the cores and not increasing core performance we moved all core priced middleware and apps to Power and low end to Nehalem.

I applaud Intel and HP for keeping this alive for customers who dont want to move, but we chose Unix to be able to move.

This announce reaffirms our decision and migration last year.

We are standardized on two platforms X86 and power, and looking at dumping Oracle next as they are becoming more like CA who we dumped years ago.

Bottom line is fall two generations behind or charge exordinate maint fees or dramatically increase your prices and you get on the divest list.

0
0
Paris Hilton

can I ask a silly question?

Why would people upgrade to this?

Vs I dunno, buying Xeons or Powers instead?

I honestly would like to know.

1
0
Gold badge

Why would people upgrade to this?

Three words: deployed code base.

Companies using in-house coded apps have mostly moved away from Itanium by now (or are in the process of doing so.) (There are always the exceptions of course; there are still some very niche areas where a very well coded app will perform better on Itanium than other platforms.) There are however still many companies using code written for Itanium by third parties. These companies may not have ported their apps, the apps may have been one off, or the ports may require a completely new (and expensive) set of licences.

The long story short is that some companies that use Itanium-coded applications may need more speed, (and thus would stump up some cash for hardware upgrades,) but don't have the budget to have thier application completely re-written (and then re-certified) for a new platform. Others may be in the middle of this process, but need ot buy time until the port or certification is complete.

Either way, Itanium is dead and everyone knows it. Unlike the Alphacide though, Intel is doing well by its customers, and giving a few more generations of faster chips to those that need it, allowing for a smooth transition path off to CoreX or Power.

0
0
Stop

Always late and not worth the wait

The latest Itanium processor is again a disappointment. Not sure why it is the 9300, maybe the DDR2 version which never made it to market was 9200.

Some glaring deficiencies are the 65nm fab and 1.73GHz. While Intel is bragging about 32nm in the beginning of the pitch they do not mention the nm on Tukwila.

While they brag about 8 threads per processor they are actually only saying it is still only two threads per core. For a 2 Billion transistor processor and 24MB of cache you would think it could have more than 4 cores.

I see SGI and Unisys have dropped out of the Itanium vendor list and the only people left are HP and people that are only looking to create a 4 socket mostly Intel box.

Strange seeing a big announcement for a chip that will not be in systems for 90days. Not sure the purpose except to say to HP's current customers that there is some hope.

Curious where the 2nd generation virtualization came from or was the 1st gen just on Xeon. Does any OS/software take advantage of any Itanium virtualization. All I have seen is partitioning and HP's IVM which is HP-UX and has 20%-50% overhead. True hardware virtualization is certainly needed.

Funny quote:

"The 9300 is Intel's seventh generation of Itanium...There's one thing that people really demand here, and that's the ability to not have to do 'forklift upgrades' on a regular basis. They want investment protection." Kirk Skaugen, Intel vice president of the Intel architecture group and general manager of the data center group

==> Itanium 9300 is a forklift upgrade. Their is no upgrade path from any prior chip.

clear example of a negative and make it look like a positive.

Curious how many QPI ports are on the chip and if it will suffer from the same chip hop problem that Nehalem EX will have beyond four socket. Reminds me of the AMD hyper-transport problem past 3 sockets.

RedHat RHEL6 - another ISV to drop support.

Are their any virtualization offerings out there that can scale a virtual machine past 8 threads on Itanium? VMWare does not work on Itanium and HP's Integrity Virtual Machine last I checked had an 8 thread limit which makes it limited to one Tukwila chip.

The best they could do for a performance number is spec_int and fp but not actually put the numbers in the press release. I am betting Nehalem and Power are 2X to 4X Itanium.

Will there be a real benchmark any time soon like SAP or TPC.

Any vegas odds on Poulson 2012=>2014having the same schedule as Tukwila. 2006=>2010.

When does the Intel agreement with HP end?

2
0

Won't someone think of the children?

"What we've done is now we've actually built a completely seamless and single tools suite...which greatly enhances the predictability. We have hundreds and hundreds of engineers working on Poulson and Kittson right now...We're on an approximately every two-year beat rate for this processor, so that's what we're staffed to go do."

Kids, when you grow up, please don't talk like this. The future of humanity is at stake.

1
0
FAIL

Even IBM...

...had the good grace to give up on PS/2 machines when it turned out no-one was buying them.

0
0
Silver badge
Happy

Obviously

The tools and compatibility features are so that later Disillusioned Itanium users can Migrate to Xeon rather than something not Intel.

For Green Field startups Itanium was dead, is dead and will be dead. I guess it's good that Intel is supporting existing customers.

Bad the way DEC stuff was killed when they were bought. The Alpha kept NT cross platform and they had really been the originators of MS Cluster and the ARM designs Marvell bought from Intel.

Is AMD really still credible competition? Was the heyday of x86 at end of 486 and start of Pentium cycle when we had IBM, Cyrix, AMD and others making various PC chips. The Geode x86 derived from those days is losing fast to ARM, ST20 and MIPS in embedded systems. Is VIA credible competiton in x86 or a hobby of the owner (who owns the rather more successful HTC)?

0
0
Silver badge
Troll

LMAO!

Oooh, did someone rattle the IBM marketting cage? Looks like IBM have realised the same as hp did a long while ago - enterprise UNIX is now a two-horse race, Power and Itanium. And it looks like the IBMers are coming out swinging wildly.

RE: tukwila and nehalem

".....when Tukwila was canceled...." Tukzila wasn't cancelled, only delayed. Now, please try and pretend every Power chip has come in right on the originally announced deadline. Just for our amusement. Do you need a little reminder that Power6 was originally slated for 2006? Then explain how Nehalem, which won't scale above eight sockets even when we get the EX chips, was able to "crush" Montvale? What application were you using? If it only needed scale of less than eight sockets and ran on both Itanium and Xeon then it would have to be Linux or Windows - the former is a complete recompile on Power and the latter doesn't run there at all. And anything that low down the enterprise scale would run faster on Xeon than Power anyway. I detect more than a faint aroma of male bovine manure!

RE: can I ask a silly question?

There are no silly questions. The simple answer is that Itanium scales much higher than Xeon can currently in a single system image. That is the UNIX SMP market where Power needs to score but where Itanium and hp-ux are winning the big deals. A new but growing market segment, large SMP instances of Windows, is also an area where Power's inability to run a variety of OSs is starting to hurt. There is pressure from Windows shops to run large instances of MS SQL, and IBM simply can't do that on Power or Xseries. For large MS SQL instances the only answer is still Itanium, and will be even when we get 8-socket Nehalem EX servers as they still won't scale as high as Tukzilla in a single Windows instance. Oh, and those 8-socket Nehalem EX boxes will be from hp - has IBM even said they will make an EX server larger than four sockets? And if they do, the real laugh will be they will need to use "node controllers" to glue two four-socket Nehalem EX motherboards into one system - derived from Itanium's cell-based backplane switch technology!

RE: Why would people upgrade to this?

"....there are still some very niche areas where a very well coded app will perform better on Itanium than other platforms....." Yes, like these three; the largest SAP installation in the World, the largest telecoms billing systems in Europe, and the largest RAC instance in Europe. All of which I understand to have been won in competition with Power and AIX. Doesn't sound so niche to me!

".....the ports may require a completely new (and expensive) set of licences....." Hilarious! IBMers trying to push the "double-the-cores-double-the-licences-only-happens-on-Itanium" bilge still? So how many core is Power going to? From four to eight, so doubling the licence costs if you buy exactly the same number of sockets. So Power is equally non-bad in this complete non-argument from the IBMers. It's telling that this is the best the IBMers can come up with! But if you want to look at this further, if a current Integrity user needs the same number of cores then Tukzilla gives you faster cores in a smaller package, saving on rackspace and system power. As TPM carefully avoided mentioning in his "analysis", a Tukzilla 4-core will consume less power than two of the existing dual-cores when we just look at the bare chips alone, and as Tukzilla servers will require less sockets to make the same core count, that means they need less motherboards and therefore even less power than the previous generation of Integrity. I wonder why TPM didn't think to mention that?

RE: Always late and not worth the wait

Well, not much to say to this constant regurgitation of the same flawed IBM FUD. I suppose we can all guess where those awful Sun marketting people went to hide after the Sunset.

".....Itanium 9300 is a forklift upgrade. Their is no upgrade path from any prior chip....." Itanium in Integrity has used the same chassis and many server components since 2001, and the new generation looks like it will be good for not just Tulzilla but also Kitson and Poulsen. I'd say that's a lot better than Power. Every Power generation has required a forklift upgrade. As it looks now, the change to Power8 will require one too. Trolls that live in glass houses.....

"....Some glaring deficiencies are the 65nm fab and 1.73GHz....." <Yawn> Been here, debunked this before! Can't you get anything new to whine about? Please, it's getting boring, there's just no sport in it if this is the best you can do! Firstly, as I tried to explain before in terms a complete layperson might understand, clock isn't the only factor, which is why lower-clocked 9100-series Itanium can beat higher-clocked Power6, and no doubt why Tukzilla will cane Power7 in the same way. It's what is getting done with each clock tha matters, and part of that is keeping the cores spinning with data - how many of those fast clocks does Power spend sitting idle whilst the rest of the system tries to get data to the cores? Ever wonder why IBM had to design such massive cache into Power? And then rig TPC benchmarks so they could use a whole system's cache for one core, just to make Power look faster? DDR3 will maintain Itanium's advantage over Power in making better use of those clocks. Secondly, Intel have already proven a die process down to 25nm, which is way ahead of IBM's Power, and there is nothing now to stop Intel ramping down the die size on Tukzilla other than the fact that it will only be two years until we get the next gen. Sorry, but does the public Power roadmap say anything other than "Future" for Power8?

".....VMWare does not work on Itanium...." And since when has VMware run on Power? And with IVM, why would hp need VMware on Itanium?

"....RedHat RHEL6 - another ISV to drop support....." IBMers shouldn't mention Linux on Itanium, it only allows comparisons to Linux on Power. Like how Itanium has always trounced Power when it comes to Linux shipments, such as in 2005 when hp shipped five times as many Integrity servers with Linux as IBM did Power servers with Linux. And hp has maintained that advantage ever since. Whilst IBM brags about being multi-OS, their limited options and the limited will to support the customer going with Linux mean those customers prefer to buy their Linux servers from hp. I won't mention ProLiant either, as the story of hp's dominance of the x64 Linux market is even funnier!

"....I see SGI and Unisys have dropped out of the Itanium vendor list....." Whilst i can't comment on Unisys as I don't know anyone that works there, a friend at SGI would like to point out to you that SGI are still a member of the Itanium Solutions Alliance, and one of the small group of sponsor members. Now, time for a little IBM history that will probably really upset you.

Long, long ago, in the day when Ponytail was stil a semi-respected figure in the IT industry, and Sun was still mentioned without the automatic wince or laugh, a group of companies brought out their first Itaniums servers. Amongst that group was IBM with the x450, though many suspected IBM were still there more as a spoiler than as a real participant in the Itanium market. In fact, IBM salesgrunts only used the new IBM Itanium systems as a means to get into conversations they would otherwise have been excluded from, and then followed the standard IBM sales mantra - "try and convert the deal to a mainframe, if that doesn't work try selling a Power server and AIX, then if all else fails try selling Xseries, and don't mention things like Linux and Windows unless the customer does first" So, with all that determination to not sell, you'd think that IBM would never have sold a single Itanium server, so you may be susprised to know that IBM used to be the second largest Itanium server seller! Yes, those poor IBM salesdrones dug up 10,000+ instances where an Itanium server was the only solution, despite their desperate attempts to flog Power instead!

"....For a 2 Billion transistor processor and 24MB of cache you would think it could have more than 4 cores...." Well, I suppose that - seeing as you are obviously not a semiconductor engineer - that you could expect anything you like. The reality is Intel are still designing large and complex cores, whereas Power's cores are getting weaker with every generation. After spending years deriding EPIC and CMT, IBM are now desperately designing EPIC features into their RISC-based Power design and at the same time simplifying the cores as Sun tried with CMT, all so they can squash more cores onto a die. This desperate IBM design process is generating a mongrel chip with little hardware compatibility between generations and hence the constant need for forklift upgrades, and is driven by the fact Xeon is going 8-core, which is eating the Power base from below. In the long-run, as Sun found out, you can't keep simplifying the cores and still make it compete with a real enterprise CPU. The only way forward for IBM that will generate viable and progressive improvements in both core power and throughput is a complete and radical redesign of Power, and that will require a complete overhaul of the OS as well. Sounds expensive, and we all know where IBM likes recouping their expenses - from the customer.

/SP&L

1
3
Gold badge
Unhappy

@Matt Bryant

Hey Matt, for the record, I am not an IBMer, I don’t work for IBM, and with the exception of my treasured IBM model M keyboard, I am almost 100% certain I don’t own anything made by IBM.

I am a systems administrator for an SME in Western Canada. If you want the truth, I whitebox everything, (servers and desktops) from a Canadian distie called Supercom. I send them the specs, they slap a system together for me, brand it under their in-house brand and give me a three year warrantee.

Itanium is dead, everyone including you knows it. I choose to see HP and Intel’s actions as noble, and in support of their customers. They are supporting this platform in it’s twilight years, providing speed upgrades for their customers for the next decade or so, and helping businesses gently transition to new platforms. This is a GOOD THING. It means that the people in charge see the writing on the wall, and decided to keep the goodwill of their customers, and to not generally be douches.

Why would you jump up and down and say “Itanium is not dead, it will take over the world, HP and Intel are confident in it’s continued dominance?” That would imply that rather than all of the noble, customer oriented motives I mentioned above, the chieftans of HP and Intel were instead myopic and grasping at a dead architecture, trying to wring every last penny out of their investment, and screw the customer. I like my interpretation better.

In the end, put yer tinfoil hat away, Matt. I wasn’t attacking Itanium. I was, in fact, giving an answer as to why you might WANT Itanium, or at least to upgrade your existing Itanium fleet to the new stuff. (As opposed to abandoning it for other architectures as is becoming common for Itanium users.) This doesn’t make me an IBMer, or even an IBM customer. I am simply a completely neutral, uninterested third party who wanted to answer a question from a fellow commenttard.

That over with, I’ll reply to what you said:

“Yes, like these three; the largest SAP installation in the World, the largest telecoms billing systems in Europe, and the largest RAC instance in Europe. All of which I understand to have been won in competition with Power and AIX. Doesn't sound so niche to me!”

Well Matt, I don’t know about you, but “largest [X] in the world” sounds pretty dang niche to me. As I had said, there are certainly a few niches where Itanium is the king, and the single best possible choice. When you are building the “largest [x] in the world” you quite simply have access to resources that no one else does. You can write your applications in house, or at least afford some niche (and expensive) support from companies like HP. If I was designing a computer system to make trades on a stock exchange in the shortest possible time frame, HP with Itanium and IBM with Power would be the two first phone calls I would make.

If I am designing a web server, a database server or any other “standard” type of computer system, most especially if I am not a fortune 1000 company, the very last thing in the entire world I would put into that box would be Itanium. Remember please my origins, Matt. I am from the SME space. Every system I design must be the absolute lowest possible TCO for a given job that will still get the job done. Price, performance, wattage, skills availability and ubiquity, long term support costs for upgrades, long term viability of the application platform without requiring rewrites or ports and oh yeah, did I mention price? This means I largely live in an x86 world, but I will admit to a few Power DB servers and a smattering of Sun squillion-thread webservers.

The point being that while Itanium is largely irrelevant to the computing world, it VERY MUCH SO HAS IT’S PLACE in these niche environments. As you pointed out, the “largest [X] in the world” space. These are people who would consider upgrading to the newest Itanium because it might just be the best thing available to them. Everyone else who doesn’t need it, however, is moving off to commodity hardware…maybe not because the commodity hardware is the best, or better…but because it’s cheaper, and so are the skills and software to support it.

Next on your list:

"Hilarious! IBMers trying to push the "double-the-cores-double-the-licences-only-happens-on-Itanium" bilge still? [blah blah blah some rant about cores at more of an angle to reality than amanfromMars]”

Matt, buddy, I think you totally misinterpreted what I said. I said “the ports may require a completely new (and expensive) set of licences.” The *ports.* As in if you port AWAY from Itanium to IBM, Sun, or even to CoreX. Depending on your application, your licences may only be good for Itanium cores. Which means if you rip the Itanium boxes out, replace them with faster Itanium boxes, your licensing doesn’t change dramatically. Let us say that you go forth and buy, (for example,) 15 IBM Power 7 boxen to replace the 15 HP Itanium boxen that were responsible for ensuring your toilet water flowed clockwise instead of counterclockwise. Let us also suppose those 15 Itanium boxen already had “Toilet Flow Ultimate: 30 socket licence (Itanium only).” Suddenly, you have to go buy a new 30 socket licence (Power only) for Toilet Flow Ultimate. This then is a great argument for *not* moving to Power, and sticking with Itanium.

0
0
Silver badge
Happy

RE: @Matt Bryant

Trev, the whole point of mentioning those three largest instances are that they are all non-niche, popular, commercial applications, and all three use standard servers, not handbuilt HPC specials. I mentioned them as they clearly demonstrate that in tightly contested and competitive enterprise deals, those customers (and believe me, telecoms are some of the most demanding companies I have ever worked for) chose Itanium, not Power or SPARC64 or Niagara. Whilst that whole enterprise UNIX market is shrinking every year as x64 eats it from below, it is still worth about $5bn per year to the vendors, which is why Intel and hp are still pushing Itanium and IBM pushes Power. And the systems they provide look after business and government processes that probably are valued by us customers in the hundreds of billions. So, whilst there may be little call for either in your space, they are not "dying" or "dead". The enterprise UNIX market is going to be here for quite a while as it simply is not possible to replace everything with x64 just yet, and whilst it is still a profitable business to be in Intel will keep making Itaniums and IBM will keep making Power.

The licence angle is true, but many of the key commercial apps come from vendors like Oracle and SAP that will cut you a deal to ensure that when you migrate between platforms you don't also change apps. Nothing makes an Oracle salesgrunt drop his prices faster than leaving an IBM Software salesgrunt's business card out on your desk. ;)

I would agree that in the low-end, SMB/SME market there will be little call for Power or Itanium as x64 does the job at that level better and a lot cheaper. However, please bear in mind that your view from the SME/SMB environment does not match what others may see higher up the enterprise ladder.

0
1
Gold badge

@Matt Bryant

Matt, my point was that "higher up the enterprise ladder" *is* a small niche. There are a heckofalot more folks out running SMB/SME and even large enterprise shops where Itanium will never be relevant.

x86/64 has been successfully clustered for huge HPC rollouts, for all sorts of weird custom gear, and it scales all the way down to tiny little MIDs. There are a few areas it isn’t suited for…yet…but those areas are few and far between. (Thus why I call them “niche” areas.)

Sparc is a player in the large enterprise space, (albeit a dubious one with the whole Sunacle thing,) and frankly works like a hot damn for an SME/SMB webserver. This, to me, makes it more relevant than Itanium because it can serve in more places. Apps can be written once and scaled up or down. The fewer architectures a shop needs to worry about, the happier the admins taking care of it are!

That leaves Power and Itanium. Now, if you sell in this space you might well know better than I the realities of what’s on offer, but I will tell you what I have seen, and you can chase your marketdroids if I am wrong, because they have obviously failed in their jobs.

Power boxes come pretty darn small, and can even be had for fairly cheap. (Cheaper than I’ve seen Itanium, certainly.) I am not talking price/performance, but simply the cost of entry. (How cheap can I get a widget to play with/develop on, etc.) This is combined with what seems to be a larger diversity of applications available for the platform. From a technical standpoint, I know full well that Power and Itanium can go toe-to-toe, the differences in overall performance usually boiling down the kind of AMD/Intel hairsplitting that ruins forums all over the internet. What’s more, I am constantly reading about Itanium deployments moving over to Power.

Regardless of whether the hair-splitting technical superiority/inferiority, the actual diversity of applications or the raw number of deployed boxes…the overall industry perception appears to be that Itanium is dead. A legacy platform being abandoned for Power on the high end, and x86/64 on the low end. If it makes you feel any better, people are deserting SPARC in the same way.

Sure, Itanium has some staunch supporters, and enough Itanium boxes are shifted each year for Intel and HP to keep investing in the chip…but there are no indications that this is something that will go on forever. Itanium isn’t increasing it’s market share, and it isn’t busting out of it’s niche. Like SPARC, it’s stuck in the few areas where Big Unix has always been relevant. The interesting bit about Power is that it is slowly leaking into the low end, and in doing so increasing market share. Power looks like it is moving out of the “ultra high end” niche it has traditionally occupied, while not letting go of any of that market share in the process.

If I am wrong on this, I’d love to see figures that prove it…but don’t blame me, blame the marketdroids at HP for not making Itanium seem relevant, and the ones IBM for making their stuff seem within reach.

Matt, my point was that "higher up the enterprise ladder" *is* a small niche. There are a heckofalot more folks out running SMB/SME and even large enterprise shops where Itanium will never be relevant.

x86/64 has been successfully clustered for huge HPC rollouts, for all sorts of weird custom gear, and it scales all the way down to tiny little MIDs. There are a few areas it isn’t suited for…yet…but those areas are few and far between. (Thus why I call them “niche” areas.)

Sparc is a player in the large enterprise space, (albeit a dubious one with the whole Sunacle thing,) and frankly works like a hot damn for an SME/SMB webserver. This, to me, makes it more relevant than Itanium because it can serve in more places. Apps can be written once and scaled up or down. The fewer architectures a shop needs to worry about, the happier the admins taking care of it are!

That leaves Power and Itanium. Now, if you sell in this space you might well know better than I the realities of what’s on offer, but I will tell you what I have seen, and you can chase your marketdroids if I am wrong, because they have obviously failed in their jobs.

Power boxes come pretty darn small, and can even be had for fairly cheap. (Cheaper than I’ve seen Itanium, certainly.) I am not talking price/performance, but simply the cost of entry. (How cheap can I get a widget to play with/develop on, etc.) This is combined with what seems to be a larger diversity of applications available for the platform. From a technical standpoint, I know full well that Power and Itanium can go toe-to-toe, the differences in overall performance usually boiling down the kind of AMD/Intel hairsplitting that ruins forums all over the internet. What’s more, I am constantly reading about Itanium deployments moving over to Power.

Regardless of whether the hair-splitting technical superiority/inferiority, the actual diversity of applications or the raw number of deployed boxes…the overall industry perception appears to be that Itanium is dead. A legacy platform being abandoned for Power on the high end, and x86/64 on the low end. If it makes you feel any better, people are deserting SPARC in the same way.

Sure, Itanium has some staunch supporters, and enough Itanium boxes are shifted each year for Intel and HP to keep investing in the chip…but there are no indications that this is something that will go on forever. Itanium isn’t increasing it’s market share, and it isn’t busting out of it’s niche. Like SPARC, it’s stuck in the few areas where Big Unix has always been relevant. The interesting bit about Power is that it is slowly leaking into the low end, and in doing so increasing market share. Power looks like it is moving out of the “ultra high end” niche it has traditionally occupied, while not letting go of any of that market share in the process.

If I am wrong on this, I’d love to see figures that prove it…but don’t blame me, blame the marketdroids at HP for not making Itanium seem relevant, and the ones IBM for making their stuff seem within reach.

0
0
Silver badge
Happy

RE: @Matt Bryant

Sorry, Trev, I hadn't noticed your response.

"Matt, my point was that "higher up the enterprise ladder" *is* a small niche...." Gartner estimates the whole server biz - x64 and RISC/EPIC - is worth about the $15bn figure per year. So that means the enterprise, which is a much smaller number of systems, actually accounts for a third of the grand total. Add to this that the enterprise segment offers greater chances for attached services, more margin-rich support contracts, and much better opportunities for assocaited system sales (such as SAN storage and backup libraries), not to mention software, and it is likely the figure is almost a break-even for the importance of enterprise compared to anything below. So, niche would seem to be the wrong descriptor. You may call it niche, but I'm sure the vendors will refer to it as important.

"....but those areas are few and far between...." Not really. The $5bn figure alone implies a lot of server units, even if we're not talking a tenth of the number of x64 unit shipments. Having said that, both hp and IBM will wax lyrical about their penetration of the Fortune 1000 list, which implies the requirement to run something bigger than what x64 can provide is actually quite widespread. Then you might want to consider universities that do not appear in any Fortune list, but usually muster several UNIX boxes apiece for academic research. Universities aren't exactly overflowing with spare cash (trust me, I've worked for a uni IT department before), and if they could get away with doing everything on LIntel then they would!

"....x86/64 has been successfully clustered for huge HPC rollouts...." Yes, I was involved in an academic project with Beowolf that eventually scaled to 32 servers. The hassle and skills necessary to just set it up, let alone the management problems of maintaining it in operation, meant it did not replace the uni's obsolete VAX. Then you may want to consider that most commercial apps that run on today's large SMP UNIX boxes won't run on such esoteric software as Beowolf, so they don't provide a practical alternative to UNIX, especially as the vendors of such esoteric platforms can't provide real business critical support as IBM or hp can. Betting on scale-out clusters is what Larry Ellison is doing at Oracle, but only because he doesn't have an in-house SMP option. At the moment, the only large SMP Oracle can offer are the re-badged Fujitsu SPARC64 offerings, and nobody seems to know how committed Fujitsu is to keeping that party going.

"....Sparc is a player in the large enterprise space, (albeit a dubious one with the whole Sunacle thing,) and frankly works like a hot damn for an SME/SMB webserver....." Sun's biggest problem was its installed base was largely one-way Netra boxes running as webservers. First the dotcom bubble bursting really hurt them as half their customer base disappeared in the space of six months, then they proved to be the low-hanging fruit for Windows and Linux x86/64 which moved into that webserving space. Niagara was supposed to defend the Sun base against x86/64, but, unfortunately for the Sunshiners, whilst Niagara makes a very good webserver platform, it is now rediculously expensive compared to x64 and far too inflexible, as x64 has scaled past it.

"....if you sell in this space you might well know better than I the realities of what’s on offer...." I don't sell anything and I don't work for hp or IBM. Amusingly, at different times, I have been accused in these forums of being a salesman for both IBM and hp! But, as a customer that does use both companies products (and has suffered Sun's in the past), I do keep an eye on what they have to offer, and before we buy anything I insist it goes through a thorough POC or shootout, as I don't trust marketting droids or salesgrunts. Or policy is quite simple - put up or shut up! If a vendor isn't willing to go to a POC or shootout then we don't buy from them. Both IBM and hp will, so we always get to see what the kit can do before we buy, and for some solutions we buy Power, others we buy Integrity. So, from my point of view, neither is the "best" or "fastest" for every role, and I expect the same for Power7 and Tukzilla. Until I've test-driven both I'm keeping an open mind.

"....The interesting bit about Power is that it is slowly leaking into the low end...." Sorry, but the Gartner and IDC market figures simply don't support that conclusion. A quick look at list prices shows the Power kit is not getting cheaper, but the x64 option certainly is. Power is not "penetrating down" into the SMB, it is being eaten from the bottom just the same as the other UNIX platforms. Meanwhile, hp has dominated the high end of the enterprise market for many quarters and has gained share in that area which also has the highest margins and attach. Seeing as hp is also the main benefactor from the x64 movement, IBM is the one losing ground overall. For a long time the IBM mainframe business has subsidised the AIX one, but now mainframes are also under pressure as the UNIX SMP servers get more and more capable, and IBM is being forced to lower margins there too That does not bode well for IBM's longterm ability to compete against hp's Integrity given the ecnomoy of scale that the successful ProLiant buxiness gives hp.

Sure, it's not going to be a quick fight, both will see a spurt as the SPARC base is converted, but it may just be a long slugging match with no clear winner before the wave of x64 obliterates RISC/ERIC completely. Unless Power7 really does have a big and clear advantage, it will do no more than tread water. If you look back to the Power6 launch you will find plenty of dribbling from Power fanbois telling the World that Power was the only answer, just as years ago we had Sunshiners telling us SPARC-Slowaris was the only choice - hp-ux and Itanium survived both quite well. Tukzilla would have to have a serious defect to cede the field to Power7, and given that the core has been stable for at least two years (the only real change was too add memory controllers for DDR3, the rest of the core design has passed testing twice), that's a bit unlikely. Given that hp-ux also seems to come out quite comfortably in comparisons with AIX, many would say better in HA, unless a company has an historic preference ("we already have it on Power so we trust IBM"), it will probably come down to the better salesteam and the better overall offer. Whilst Power fans may claim that IBM can offer an integrated software stack as well, hp will point out that Itanium can not only support the same IBM stack but also other stacks on Windows and OpenVMS that Power can't. It's going to be an interesting few years, but for Power7 fanbois to just declare victory before either competitor has even taken to the field would seem to be the height of either complacency, naivete, or just bluster.

/looks forward to some interesting times ahead!

0
2
Gold badge
Pint

Does the high-end need some mid-range or low end to survive?

Mey Matt, you make some excellent points, and remind me why I love being a commenttard. Beyond the articles, there is much to learn here, and many new points of view to be exposed to.

First, I think the argument about the nomenclature of “niche” is really just that: wanking about nomenclature. I acknowledge wholeheartedly that as a non-commoditized area, high-end non-x86 boxes are where the real money is made. (It’s very hard to make a profit on x64. The margins are absolutely razor thin, all the more so the closer you get to the actual customer.) I view “niche” as low % of total customers, not as % of total revenue. Commoditisation changes everything.

“both hp and IBM will wax lyrical about their penetration of the Fortune 1000 list…if they could get away with doing everything on LIntel then they would!” Consider me suitably informed. I was under the (obviously mistaken) impression that there were perhaps a grand total of 3000 high-end/HPC customers worldwide. (Maybe 10,000 if you consider SPARC to be a part of this.)

"Betting on scale-out clusters is what Larry Ellison is doing at Oracle,” strangely enough…this is perhaps the biggest reason I still see large clusters of x64 or even SPARC as viable. No individual is perfect, and no individual consistently makes the right choices…but Larry Ellison is one of those people who seem to be quite capable of making [socked descriptor] amounts of money. Personally, I’ve little use for Oracle, and heard absolutely nothing but horror stories about dealing with them…but I wonder what the likelihood is Ellison would be totally out to lunch on this.

“…it may just be a long slugging match with no clear winner before the wave of x64 obliterates RISC/ERIC completely.” Until I’ve seen some real consolidation on the high-end, however, I’d still not buy stock in either of ‘em. I personally believe in the inevitability of commodity chips steamrollering the high-end stuff. The only thing lacking is high-speed interconnects to become truly commoditised. (Well, and the software to catch up.) The better Linux distributions become at lashing clusters of servers together, the more strongly I believe there aren’t enough customers to support two vertically integrated high-end heavy hitters.

“Unless Power7 really does have a big and clear advantage, it will do no more than tread water.” No offence, but isn’t that essentially all Itanium has been doing for a while now too? There are just a limited number of people and organisations who will buy non-commodity hardware, and more importantly, non-commodity software.

“…but for Power7 fanbois to just declare victory before either competitor has even taken to the field would seem to be the height of either complacency, naivete, or just bluster.” Well, I’ll be honest when I say I can’t debate you at all when it comes to the Unix players on these platforms, Matt. It honestly and truly is deep into territory I just don’t get to play in. What I can say is that while I have never seen an Itanium box with my very own two eyes, I do know folks who run Itanium shops. (HP-UX, as a matter of fact.) Equally I know a lot of shops using Solaris in one form another, (guilty as charged,) as well as only a sprinkling of AIX customers. Each and every one has decided that their compute needs can be met by Linux, and that it’s worth putting the money in to get the heck away from Unix. They are all talking about Linux on Power when they do so, not Linux on Itanium. I acknowledge my personal experience can not be taken as representative of the world as a whole, but I think that as the Unix software market is eroded from the bottom, this doesn’t meant he hardware has to be as well. Again, I’ll be honest when I say here is where IBM’s marketing folks have kicked the ass of HP.

If 15 years from now HP-UX on Itanium is the only Unix around, I honestly think it will be because Linux on Power evolved as the “mid-range” offering of choice. Maybe a “high-end Unix processor” can’t exist unless there is at least a certain level of economy of scale. I can’t say for sure, and I don’t trust any statistics company or marketing department to tell me the future. I only speak about what I personally see; and that is the low-end of the Unix folks fleeing Solaris, HP-UX and AIX like rats off a sinking ship heading for Linux on x64 and Linux on Power. As you say, interesting times seem to be ahead…

Also, I'd like to add a thank you for keeping up with the long replys. Despite what seems to be some different areas of focus and experience, I consider it a genuine pleasure to be exposed to new viewpoints. Your comments have given me much to think about!

Also: All apologies to Sarah (or [insert other sub-ed here]) for these long-winded comments. I really hope you in no way bothered to read them all the way through, because it would probably have been a fantastic waste of your time.

0
0
Pint

This makes me sad

Not for itanic, it hit the ground running like a norwegian blue. But for alpha and parisc. They were good chips, which itanic <teen speak>so totally</> is not. Even the execs know it and tell you to go x86_64. As to amd, well, where did you think intel had x86_64 from? The irony is blistering.

The problem here is that beyond ibm's POWER line, there's nothing innovating except what fits in the x86 straitjacket. And that is on a deeper level just as bad as forcing your sofware world of innovation in the micros~1 straitjacket. The only other contender in chip innovation is not in HPC: Amazingly ARM is trying to chew through from the bottom, despite intel and micros~1 ganging up on it to try and kill it dead.

Hoisting one in remebrance of lost friends, and to PA Semi and Transmeta, who at least tried to stir things up a bit.

1
0

Not only sad, but also very bad.

Itanium's death (and HP not launching a product is nothing less) is actually very bad news for all of us. 'Us' as in 'UNIX folks'.

No Itanium means no HP-UX. This leaves the comercial UNIX market with AIX and Solaris.

With SPARCs future being cloudy at best, it is forseeable that most of us will either do Solaris on x86 or AIX on POWER. The latter one with a mind-boggeling surcharge for being on the last surviving 'true' unix.

Eithr way, we are fsck´ed.

0
0
WTF?

A question.

Why does intel leave feature engineering to marketing these days, even at the cost of year-long delays? Given that itanic originally was another intel engineering design, not slavishly pandering to the wintendo market, this is somewhat ironic. Not as ironic as it showing intel's engineering has trouble doing anything outside the wintendo mindset, or itanic's subsequent failure, but still. I'm somewhat surprised that no fellow commentard picked up on that little tidbit yet, either. Why didn't the engineers think of what apparently even the repetetive dimwits at marketing came to see as obvious?

0
0
FAIL

"Mission critical innovation"???

Shall we have a look at some of the RAS claims in the "Technical White Paper"? It's at

http://h20341.www2.hp.com/integrity/downloads/21568_Intel_Tukwila_Tech_WP_r06.pdf

Some of the "mission critical RAS" claims include:

Memory scrubbing (caches and main memory): sorry, been around for years on x86 and elsewhere, particularly on cache, e.g. from 2004: http://pages.cs.wisc.edu/~shubu/papers/cachescrub-prdc2004.pdf

Dynamic link rerouting: Hypertransport had that long before Quickpath came out.

Processor onlining/offlining: VMS had that when SMP was introduced (mid 1980s) though (as below) it's not *that* relevant as a RAS feature.

Processor hot plug/hotswap: Proliant had that years ago, though like onlining/offlining it's actually mostly a slideware feature as few (not none) OS/application combinations can sensibly use it. Same goes for memory hotswap.

"Chipkill" equivalents and variants (posher than ECC memory protection): initially from IBM, then in Sun and Proliant and elsewhere. Since the 1990s.

Memory migration: been in the ex-Tandem boxes for years, no specific chip dependency involved, just sensible system design.

PCI (Express) Hot Plug: Proliant had PCI hot swap ages ago, can't say for PCI Express right now.

Quickpath RAS: Now the same across IA64 and Xeon Quickpath RAS? So where's the IA64 RAS advantage?

Etc.

Actually, that's basically it. Nothing new to see here.

If you want to stay with HP-UX or VMS, best to just contact your HP rep and ask when they're introducing a Proliant/AMD64 version of your favourite OS.

1
0
Anonymous Coward

HP Itanium 9300 systems

Someone please help me out here.

I am looking for the systems that HP announced yesterday and I can only find information about the chip. Not to be sarcastic but I learned about the chip two years ago and want to know what HP is doing with the chip. Will HP have a follow-on to the SX2000? When will there be a Superdome replacement? Will a superdome be a scalable SMP or a hodge-podge of blades?

I cant find any benchmarks either, just a reference to SPECint and SPECfp that HP submitted but not on any site or press release.

IBM announce four systems with Power7 and it looks like an impressive story.

HELP!

Juniata

0
0
FAIL

But at least Supermicro is a new OEM !!

umm... well maybe not

SAN RAMON, Calif., Aug 20, 2008 -- The Itanium Solutions Alliance welcomes Super Micro Computer, Inc. as a sponsor of the non-profit association. Supermicro's sponsorship will be announced tonight as part of the first annual Innovation Celebration event honoring the Alliance's 2008 Innovation Award winners in San Francisco. Supermicro joins the Alliance as a longtime manufacturer and vendor of Itanium-based platforms.

September 12, 2003 – Super Micro Computer, Inc. announced the debut of its Intel® Itanium® 2 solutions at the Fall Intel® Developer Forum (Sept. 16-18) and then in Asia at the Taipei Computex tradeshow (Sept. 22-26). At these shows, Supermicro will unveil the world’s first 1U Itanium® 2 platform to meet the ever-growing demands of the high-computing server segment. The products unveiled will include the most sophisticated 1U Itanium® 2 serverboard in a standard form factor: the SUPER i2DMR-8G2, and the equally impressive SC813S-500 chassis, specially optimized for Itanium® 2-based solutions.

0
0
Anonymous Coward

Matt Bryant

You are wrong on every aspect. POWER7 is much faster and better. Itanium is dead. Face it. This will likely be the very last generation of the slow Itanium chip. Come on, is 1.73GHz clock speed anything to call home about? Admit it, 1.73GHz is nothing compared to 4-5GHz as the mighty POWER wields. 1.73GHz is nothing, it is chicken sh*t. Itanium is no more, after this one. I promise you that.

We are now in the process of migrating from Itanium to POWER7, as everyone else.

0
0
Silver badge
Happy

RE: Matt Bryant

"You are wrong on every aspect. POWER7 is much faster and better....." Hmmmmm, seeing as it would have been impossible for you to get hold of both and do truly comparable benchamarks, I would suggest you are talking from an orifice not too far separated from your butt cheeks!

".....Itanium is dead....." How is it "dead"? Do you mean "no-one will buy that"? So you have personally polled every CIO in every possible customer in the World and they all came back and gave you a rock-solid guarantee that they would not buy it? Or are you merely speculating based on your own and obviously limited assumptions? Gee, I wonder which!

"....Admit it, 1.73GHz is nothing compared to 4-5GHz...." <Yawn> Please go read what I posted in response to Mrs Kebabfarts regurgitating of that FUD. If clock was the only thing that mattered then the current 9100 range of Itaniums would never be able to compete with Power5, let alone Power6/6+. Yet I can assure you that current Itaniums can outperform Power6+ as I have seen it done. And with real apps and real data, not carefully constructed "comparative benchmarking" sessions as pushed by vendors.

"....Itanium is no more, after this one. I promise you that....." Now I am beginning to worry - how exactly are you going to "promise me" that? That is the kind of scary, obsessive statement that suggests you really need to talk to someone, preferably someone with pyschiatric training! Should Intel take out a Restraining Order to keep you away from their premises?

"....We are now in the process of migrating from Itanium to POWER7, as everyone else." Ah, once again you have talked with every Itanium customer in the World. I'm guessing a lot of that talking was happening in your head, n'est pas?

/SP&L - can we have a "rabid troll alert" icon, please?

0
1
Anonymous Coward

Matt Bryant

I have not talked to every customer, but we all know that Itanium is dead. We advice against Itanium and recommend POWER7. It is obvious that 4-5GHz is faster than a measly 1.73GHz. You lie when you say that Itanium is outperforming 4-5GHz cpus. It can not happen. I dont see any reason to buy Itanium when POWER7 exists. No one buys Itanium anymore, everone is throwing it out in favour of POWER7. POWER7 is faster and better. You are just too stubborn to admit it. But you know that Itanium is defeated by POWER7. You know it.

We are soon done with our migration to POWER7 and we see performance improvements of 4-5x times and increased stability! We had lots of stability and performance issues earlier with our Itanium servers. Sorry, but Itanium does not cut it. Too bad. It is not stable enough.

0
0
Silver badge

RE: Matt Bryant

"I have not talked to every customer...." I am trying to look surprised, honest!

".....but we all know that Itanium is dead...." So, you didn't talk to them, but you speak for them all.... ESP maybe? Seeing as you admit you haven't spoken to all customers, and I'm really not going to accept any suggestion of ESP (it really wouldn't help your argument!), what that leaves is that, in your opinion derived from your personal experience and appraisal of the market, you feel you are perfectly positioned to state unequivically, that no-one is going to buy any Itanium servers. Well, seeing as that's just your opinion, and it looks like your position should probably be a padded cell, I feel quite comfortable in just pointing and laughing at you.

"....We advice against Itanium and recommend POWER7...." Ah, you "advice" (sic) on servers, so we can safely assume you are a consulting house or IBM reseller, maybe? Yes, it's quite clear now what your "position" is.

"....It is obvious that 4-5GHz is faster than a measly 1.73GHz....." In what way? As a direct comparison of just meghaertz, 4-5GHz (sorry, can you be more exact seeing as you claimed you'd tried Power7 already?) would be faster than 1.73GHz. But, as I could try explaining to you in very small words, clock is not the whole story. Just go back and look at articles about the early Opterons versus the Xeons of the day -the Opterons completely thrashed faster-clocked Xeons, beacuse the Opteron cores did more per clock. In the same way, going by your reasoning, you would expect the jump from Power5 to Power6 to provide almost twice the performance per core, but what customers actually saw was usually only around 10% per core. Well, that's what we saw, but I can't speak for all other customers, unlike yourself.

"....It can not happen..." Thus spoke God! Well, actually He didn't, and it did happen. I have seen hp Integrity beat Power servers in competitive shootouts, and I am much more inclinced to believe my own eyes than someone that says they know what all customers think without speaking to them. That's not a very good line to take if you are in consulting.

"....No one buys Itanium anymore, everone is throwing it out in favour of POWER7...." When you say "everyone", I assume you mean everyone talking in your head? Mayber you can't hear me over the rest in there, should I mentally shout a little louder? I can assure you that we have bought Itanium (some just last week, actually), have plans to buy the new kit when available, and only have contingency plans to replace Itanium with Power should something dire happen, like the senior management at hp start taking advice from people talking in their heads. We also have contingency plans to replace our Power kit with Integrity should there be a disaster at IBM, like if they employ someone like yourself in a real decision-making role.

"....POWER7 is faster and better..." Your opinion, or the opinion of the voices in your head? I think we can all disregard your assertion that you have put one up against the other in a real test.

"....You are just too stubborn to admit it. But you know that Itanium is defeated by POWER7. You know it...." Well, yes, I do admit to being a tad stubborn, as the Sunshiners that have crossed my path will agree. But, seeing as I also haven't put a Power7 solution up against a Tukzilla solution in one of our shootouts, I "know" nothing of the sort. Were you wagging your finger at the screen whilst you typed that, assuming they let you type out of a straightjacket.

"...We are soon done with our migration to POWER7 and we see performance improvements of 4-5x times and increased stability!...." How? The servers have just been launched, there is no Power7 hardware on the street other than some seed units, and it is highly unlikely that a real company would put mission critical business apps onto seed kit. Would you care to withdraw or temper that statement, because otherwise we might conclude you are just a big, fat liar. Well, some may just be arriving at that conclusion, others may have already got there.

"....We had lots of stability and performance issues earlier with our Itanium servers. Sorry, but Itanium does not cut it. Too bad. It is not stable enough." Like what "issues"? Did the servers not respond to your mental commands, maybe? Seeing as hp-ux on Integrity is five-nines rated and has industry awards for stability (try a Yahoogle of The Uptime Instsitute for starters) , and currently runs a large number of 24x7x365 biz crit business solutions all over the World, I think that if you really have had stability issues they must be due to your own incompetence, which doesn't make very good grounds for accepting your advice. All in all, I hope for your sake your customers don't read The Reg forums!

/SP&L

0
1
Headmaster

@Matt Bryant - It isn't...

non, c'est ' '...n'est-ce pas...'

Merci.

0
0
Silver badge
Happy

RE: @Matt Bryant - It isn't...

Apologies to true Francophiles, it has been a long time since I regularly abused French ears with my schoolboy French! I promise to stick to winding up other trolls in broken English alone in future.

0
1
Megaphone

@ Matt re "Uptime insitute" ???

Matt, I did as you suggested, searching for "uptime institute" and "hp-ux", and all I initially found was a pile of articles referencing the Uptime Institute and the Uptime Institute's Fault Tolerant Power Compliance Certification, which didn't really seem to relate to HP-UX. Was there somewhere else I should have looked?

Do Proliants have Uptime Institute chittys? If not, why doesn't it matter for Proliant?

0
0
Silver badge
Happy

RE: @ Matt re "Uptime insitute" ???

I'm told the ProLiants also have Uptime Institute certificates, but the main point here is that it is an example of how hp look at all areas of availability before classing their machines as five-nines, in a total design process. The Integrity and ProLiant range share more and more components, including power supplies, which means it is cheaper for hp to make Inetgrity servers than the competitors, but both benefit from the drive for high-availability by design that happens in the Integrity range.

Having spent the design time doing that ground-up work, and knowing how important power resilience was to customers (especially in places like California), hp decided to let an independent institute test their kit. Whilst I know the IBM kit is also quite good at this, IBM haven't seen fit to put their kit out to a similarly independent tester.

0
1
rch
Go

Re: Uptime Institute - What irony

I found it.

http://www.upsite.com/TUIpages/tuihome.html

Where they say:

#The Site Uptime Network’s mission is to identify, quantify and improve infrastructure availability.#

And when trying to look at the White Papers I got the following.

#Can't open.../w3w19033/web/upsite/webroot/cgi-bin\admin\templates\template_wp.tpl No such file or directory at PARSE.pm line 15.

For help, please send mail to this site's webmaster, giving this error message and the time and date of the error.#

I didn't find the webmaster but I guess a mail to HP will suffice.

Oh wait, the only servers certified there are HP Integrity servers so I assume other types are complete crap.

Big yay to Uptime Institute!

0
0
FAIL

@ Matt

You either blind or you work for HP...

I really doubt HP has the largest customers in these segments you mentioned.

Please, tell us their names... It's pure bullsh*t...

The market speaks for itself Matt... Itanium market share is diving since Tukwilla failed to come to market 2 years ago.

I have not heard about anyone buying Itanium machines for 2+ years in my country.

So, stop spreading FUD cause nobody at El Reg is buying it.

Cheers!

0
0
Silver badge
FAIL

RE: @ Matt

"You either blind or you work for HP...." Why, just because I disagree with you? Very open and objective - not! I don't work for hp or Intel, nor do I work for some rentamarketting consultancy paid by hp or Intel. Just to show you that obsessiveness is not unique to IBM fanbois, I have had Sunshiners on this forum accuse me of working for IBM! Get over it - just because someone doesn't agree with your myopic viewpoint it doesn't mean they work for a competitior.

"....Please, tell us their names... It's pure bullsh*t..." I suggest you speak to hp about their customer base, I'm sure they can provide you with plenty of case studies to prove even to you that Itanium has been selling quite well to many well-known companies. If you don't feel inclined to talk to an hp salesgrunt (and who could blame you!), you could always look here:

http://h20341.www2.hp.com/integrity/w1/en/messaging/case-study-industry.html

Is that enough "bullsh*t" for you? You may have to take your head out of your rectum and your size twelves out of your gob in order to view that page.

"....I have not heard about anyone buying Itanium machines for 2+ years in my country....." I see IBM are even outsourcing their trolls now! Which country is that? You could always go to the www.hp.com website, put in your country in the dropdown and then contact your local sales office for a little local information. Well, if you were at all interested in the truth, that is.

"....So, stop spreading FUD cause nobody at El Reg is buying it...." Buying what, the "FUD" or Itnaium? My "FUD" is free, you don't have to pay me for it, but you can if you wish as I understand IBM customers do feel they have to pay through the nose for everything. As for Itanium, I suggest you check the IDC or Gartner figures, they show that plenty of people have bought Itanium even during the delay period for Tukzilla. That's if you can access Gartner's or IDC's website from The Kingdom of Ihaventgotafudgingclueistan, or wherever it is you post from.

0
1

This post has been deleted by its author

Joke

Uptime Institute is down and out since 2007?

Thank you rch.

"the only servers certified there are HP Integrity servers"

It looked like that might be the case when I looked but I didn't have time to be definitive.

I just had a look at the URL you provided and the "Upcoming Events" section is dated 2007. Oh dear Matt, what have you done here. [http://www.upsite.com/TUIpages/tuihome.html]

So Integrity servers are certified by a dead (or at best careless) Institute, even though the market (outside of HP-UX, NonStop, and OpenVMS) isn't interested in Integrity, and Proliant servers (the server which wrote the book on "industry standard", which runs most of the world's business systems) doesn't rate a mention at the Institute.

Is that a rat I can smell, or is there something dead and fishy in the room?

Still, as we see here, Matt's always game for a laugh. One day his sponsors might realise he's doing more harm than good though, and then we lose our entertainment, so maybe we shouldn't be too harsh.

Or maybe he's an IBM employee going undercover ;)

0
0
Silver badge
FAIL

RE: Uptime Institute is down and out since 2007?

And IBM's independent analysis or certification of their RAS features, let alone power structure is where....?

"....even though the market (outside of HP-UX, NonStop, and OpenVMS) isn't interested in Integrity...." Ignoring the large hp-ux, OpenVMS and NonStop markets? That's a bit like saying there is no market for Power7 outside of pSerie and zSeries. Oh, wait a sec.... There is no market for Power7 outside of those two! And don't make us all laugh by trying to talk about Linux on Power, iSeries is just too niche to mention.

So, you want to ignore the hp-ux, OpenVMS and NonStop on Itanium markets - so what about Windows on Itanium? All those large SQL databases, where SMBs that have traditionally used only WIndows but now find they have outgrown the capabilities of MS SQL on x64, what do you think they use? Do you serioulsy think they go to the expense of choosing new apps, retraining their MS SQL DBAs on DB2 and their sysadmins on AIX? Yeah, right! They buy WIndows on Itanium so they can use their existing skills and tools, and Power is never in that discussion as Power can't run WIndows.

"....Still, as we see here, Matt's always game for a laugh....." With you IBM fanbois it's either laugh or cry! I really though we'd seen the last of such moronic FUD with the Sunset, but it looks you lot have decided to make up for the drop in Sunshiners, either that or they have simply migrated to what they think is the new fashionable vendor.

"....One day his sponsors might realise he's doing more harm than good though....." Jokes on you! I don't have any "sponsors", but I can see why you would want to paint me as an interested party and have me stop debunking your FUD. Not keen on a little counter-information, is it making your sales a little more tricky when the customers might actually get more than one side of the story?

"....Is that a rat I can smell, or is there something dead and fishy in the room?...." Check your Brads, it may be time for some more deodorant. Definately time for some bew IBM trolls, these ones have exhausted both what little material and wit they had available.

/SP&L

0
1

OpenVMS

Actually, I have heard much about OpenVMS legendary stability. I have heard of single OpenVMs machines with uptime of 17 years. And I have heard much about OpenVMS legendary clusters that never goes down. 100% uptime for 20 years or more. In the attacks against Twin towers, nine eleven, the finance companies saw the buildings collapse but the OpenVMS systems in there did not loose a single transaction, because they were clustered in another town.

Although I am a Sun fanboy, I doubt Solaris is as stable as OpenVMS, because I have not read those things about Solaris. In my company, a large world famous finance company, we use OpenVMS on one of our largest systems. All sysadmins speak very well about OpenVMS.

I rate OpenVMS above Solaris, in terms of stability and clustering. But OpenVMS is not that sexy to use, though. And it is not open, nor exist on x86. If it were released on x86, there is high chanche that I ditch Solaris for OpenVMS. I am a geek, and only support the best technology. However, Solaris is not bad, with ZFS, DTrace, etc. OpenVMS is not sexy for desktop, but OpenSolaris is sexy on the desktop.

But I think everyone here rates OpenVMS highest in terms of stability and uptime and clustering. Next to Mainframes. Then comes all Unices: Solaris / AIX / HP-UX / etc and lastly comes Linux/Windows/etc.

I dont know about the other OSes that HP has done, Ive only read about (and tried) OpenVMS. However, there is a Solaris zealot in the OpenSolaris forum, that praises HP-UX as very stable. But very bare bone.

0
0
Silver badge
Happy

A note to all the IBM fanbois.

Seeing as there seems to be little evidence of reason in your posts, let's try and set a baseline for you. I appreciate that many of you are far from the coalface in your sales offices but please try and put aside your petty bias and consider the following, if only for your customers' sake.

First off, Itanium is not dead. For you to state this is simply wishful thinking in the extreme. There is a large installed base and for Itanium to be dead you would need the majority of that base to be looking to migrate to another platform. Please consider that any migration is both expensive in terms of time and resource, and introduces risk, which businesses hate. A change in platform also usually means adding new staff to gain the platform skills for the new platform/apps or re-training existing staff, which disrupts daily operations. Large enterprise customers may be able to absorb this, but many medium-sized businesses will baulk at the added cost. Please feel free to use the same argument to explain to customers why IBM mainframes still exist. If you feel the need to understand this further, please talk to customers running SPARC-Slowaris and see how many are just switching platforms at the drop of a hat, even after the Sunset.

So, current Itanium customers, whether they are buying hp-ux on Integrity or running RHEL on an OEM Intel whitebox, will consider a shift in platform the least attractive option. This could only be over-ridden if Power7 in the new pSeries offered some radical business advantage and/or significant savings after the pain, risk and cost of migration. Currently, from the information available, I don't see Power7 offering that. For you to blithely state otherwise, when there is zero chance that you have been able to benchmark and compare both systems in an objective manner, is simply laughable. In short, it makes you look like trolls, and stupid ones at that.

So now we get to the customer bit. Us customers, despite what you think, are not stupid. Whilst the Power shops may be happy to swallow the FUD, the companies like mine that have both IBM pSeries and hp Integrity are not so guillable. If our IBM sales team walked into our boardroom tomorrow and said "Itanium is dead" they'd immediatly lose credibility. Luckily, they are not that stupid. Equally, we would not expect our hp account team to walk in and say "Tukzilla is so good, Power7 is a donkey."

Just over three years ago I decided to advise our board that the Sun SPARC option was not a good one. Seeing as we had a close-on 50% Slowaris-SPARC estate this was not what they wanted to hear. It took six painful months building a careful business case, with the support of both IBM and hp sales teams, and against considerable internal resistance, to convince them what I was suggesting wasn't just wishful thinking. If I had just walked in and said "Sun is a dead because I say so" it wouldn't have happened. Even then, we designed contingency plans so that if Ponytail had pulled off a miracle we could still have used the Rock-based "Supernovas" as an equal choice to pSeries and Integrity.

So, before you post any more trollisms, please think of your customers and try and come up with something worth posting that is also empirically proveable, and then post it in an intelligent and adult manner. Any more childish/rabid claims of "Itanium is dead and you know it" will only cause us customers to think even less of you.

/SP&L

0
2
Paris Hilton

Will the real Matt Bryant please stand up :-)

I think there are two of you.

1) The one that blindly defends Itanium and regardless of Intel's seamingly total disregard make unsubstantiated claims and accusations. i.e. saying "IBM would not be able to have octo-core chips before October"

2) The Matt Bryant that wrote this response.

- Itanium is not dead: Yes Intel is very proud that Itanium system revenue is now higher than sparc and is flat lining around 60% of Power systems. Having watched both announce webcasts it was apparent IBM is very proud of the thousands of migrations and Intel was trying to show a sustainable business model. The claim of consolidating 8 superdomes to one eight socket box is very interesting.

- The business value comes from the vast superiority of the PowerVM technology vs. IVM. Power7's huge leap in performance extends the consolidation capability

- ROCK is dead, SPARC64 is dead. Itanium will be late and will have a very hard time getting attention from Intel as Nehalem eats aways the bottom and Power7 takes away the mid to high space. Does HP need it for Non-stop, VMS, HP-UX? yes....but can it compete for new business and keep customers from moving key applications and middleware? only because of the cost of migration.

- Sometimes people get too I told you so when it comes to EPIC vs. RISC and the history of the good fight.

- You have been warming to Power over the last year or so and the bloggers appreciate that.

I don't think I said Itanium was dead...i have heard a alot of people say everything is dead except for x86, Power and Mainframes....but that is not a timeframe statement. Rock is dead, Disco is back.

Obviously Itanium is a disappointment though. 65nm / 1.73 GHz/half the cache per core / no systems yet / and a seemingly unwillingness for HP to invest in migrating HP-UX to Nehalem at odds with Intels unwillingness to invest much in Itanium. IVM only supporting 8 thread VM's.

And while HP "submitted" SPEC benchmarks no one was willing to admit to any result anywhere.

0
0
Silver badge
Troll

RE: Will the real Matt Bryant please stand up :-)

"....The one that blindly defends Itanium....." I don't blindly defend anything, I test and compare in my own environment, with our data, and our app stack. How a server or solution performs anywhere else really doesn't matter to me.

".....saying "IBM would not be able to have octo-core chips before October"..." I said it was likely or probable that IBM would start with low-end servers and use chips with failed cores as it sorted out the manufacturing process, not that it was a definitive fact.

"....The claim of consolidating 8 superdomes to one eight socket box is very interesting...." And, frankly, not much to crow about. I could squeeze the load from eight first generation SDs running single-core CPUs into a much smaller system today. And if I really didn't care too much about performance or resilience then I'm sure I could squash it down into an eight-socket using the current Itaniums, let alone Tukzillas. Of course, I will probably be able to do the same with some ancient pile of Power2, Power3 or even Power4 servers into a single 8-socket Tukzilla server with ease, but I'm much more likely to take advantage of hp's superior blades offering and replace those old Power servers with Tukzilla blades and use such goodies as Virtual Connect to increase resilience. After all, one server isn't a very highly-available solution, which does make me question how truthful it could be that a customer using even first generation Superdomes - a high-availability design - would agree to put all their eggs in one Power basket. Unlikely, unless it's another one of those IBM "case studies" that never happened outside some carefully contrived IBM Labs test.

"....The business value comes from the vast superiority of the PowerVM technology vs. IVM...." <Yawn> What "superiority"? Please explain. I'm not couting on it being very accurate given that you have again repeated a falsehood about IVM lower in your post.

"....Power7's huge leap in performance...." And again - have you benched it, in a real environment, with real data and real apps? No, I didn't think so.

"....SPARC64 is dead...." Debateable. Fujitsu still have a very good chip, much better than anything Sun came up with, but they need to produce a credible roadmap. It may be they are waiting to see if Larry throws them some development money - if he doesn't, then SPARC64 will probaby wander off into the HPC wilderness and die a lonely death. Until then, your statement is a bit premature. Please read my previous post as to why you may want to be a bit more careful.

"....Itanium will be late..." Well, seeing as the chip has been launched, how can it be late? Do you mean future versions such as Poulsen and Kitson will be late, or are you referring to systems from vendors such as hp? At least the future versions are on a detailed, public roadmap, you can't say the same for Power8! And, seeing as hp had the server designs ready last year, it is highly unlikely they will be delayed.

"....and will have a very hard time getting attention from Intel..." Why do they need attention from Intel? Are you implying there is some sort of lack of resource at Intel, that they only have enough engineer teams can only work on one CPU at a time? Hey, this is Intel, not Sun! Last time I checked they had more chip design teams than IBM, which makes you kind of wonder maybe you should be more worried about future pSeries given the pressures on IBM to keep mainframes competitive, replace Cell, and design the next PowerPC generation. I think we're back to the trolls that live in glass houses thing....

"...You have been warming to Power over the last year or so..." I have always rated pSeries as a good solution, but not the best or only solution. But I do much prefer hp-ux to AIX, probably since I have more experience with hp-ux. I'm sure those that have never tried hp-ux will be quite happy with their new Power7 AIX servers, and those that have never tried AIX will be happy with their Tukzilla hp-ux, OpenVMS, NonStop or Windows servers. And the Slowaris users should probably get some experience on one or other of the options soon!

"...Obviously Itanium is a disappointment though..." Why? After all, for customers happilly using Montvale systems now, Tukzilla will provide an attractive upgrade if only to reduce rackspace and power requirements. Since it will also provide a performance boost I can't see how it can be a "disappointment", not unless you're an IBMer and are just disappointed that Power7 doesn't have the field all to itself.

"....a seemingly unwillingness for HP to invest in migrating HP-UX to Nehalem...." This point often raised in IBM FUD really makes me laugh. Please show me the AIX or z/OS roadmap that has a port to x64 on it, or can we start saying AIX and z/OS are dead because IBM is "unwilling" to port them to Nehalem? Please, try and come up with an adult argument for a change! There is a very simple reason hp don't switch hp-ux to Nehalem - they don't think Nehalem is ready for the scale required. For now, Itanium is the only option that gives that scale. In the meantime, as the leading Windows and Linux on x64 vendors, hp isn't exactly desperate for another x64 OS, they're beating IBM in x64 without needing hp-ux or OpenVMS on x64.

"....IVM only supporting 8 thread VM's...." And now we get to the repetition of the falsehood mentioned before. Please explain your statement as it appears patently untrue. You may be referring to a limitation form an early trial version, maybe the fact that IVM 4.0 could only have eight virtual CPUs, though this did not limit the number of VMs. But current IVM 4.1 can scale higher than eight VMs with ease. Indeed, I personally gave up after I had tested IVM with 32 VMs as - whilst IVM was happy to allow me to create more VMs - I was running out of memory in the test box I was using, and giving each VM less than 1GB wasn't producing a really pratical server for what we wanted to do. In theory, you can run an hp-ux VM with only 750MB of RAM and Linux with as little as 128MB, but we expect an enterprise server VM to run with at least 8GB for the work we do. Anyway, I suggest you ask your FUD writers to update their work - you wouldn't want a customer to point out you're telling porkies!

/SP&L

0
2
Happy

Matt Bryant

Fail again. All customers that are migrating from Itanium to POWER7 in this very moment, had some hefty discounts. They dont "baulk at the cost". As I said, we had stability issues with our Itanium servers. They are not stable enough. I hear rumours from HP management that further Itanium development will be canceled. You will be kne deep in sh*t if you dont migrate to POWER7 now. The sooner the better. The Itanium is too late and it is too difficult to write the compilers. Itanium is soon-to-be dead, because of slow development, it can not keep up with POWER7. 1.73GHz is too slow, compared to 4-5GHz. It is nothing.

0
0
FAIL

Last AC you really are quite dull.........

Since when has core clock speed on different processor architecture ever been a relevant comparisson? Someones already posted it but AMD had a big advantage performance wise over Intel at one point with Lower clocked CPU's so that kind of disproves that statement.

Its not how fast it clocks its what you do with it that counts is whats important i.e. something measureable like SAPs per CPU

No doubt we'll have to continue to put up with this until some real benchmarks are published so we can all make an educated comparisson between the two processors not one based on rumour and pure fud from the (Mainly) pro IBM brigade. In the mean time this is all speculation.

PS I think Tukwila's biggest issue won't be performance but its lateness to market.

You hear rumours, I heard a rumour that IBM sold some product to Nazi Germany - oh no thats true!

0
0
Pint

A chip without a home

Where are the HP systems announcements there Matt? Since you work there - you should know! Oh - right - there are none because once Intel found out that IBM pulled a fast one with the POWER7 announce date, the marketing dept at Intel jumped the gun.

But all that was announced was a dud that might benchmark well on spec with DDR3 memory but I doubt she'll catch POWER6 on systems benchmarks (the ones that involve disk too).

I believe once HP announces systems, they will only be of interest to those poor souls clining to HP-UX.

Booooooooooooooorrrrrrrriiiiiiiiiiinnnnng! Back to my BEvERage.

0
0
Silver badge
Troll

RE: A chip without a home

"....Since you work there - you should know! ..." Oh dear, I detect another troll with a reality problem. I don't work for hp. I do work in the IT industry. I seriously suspect you don't work, fullstop. It was IBM that rushed forward their Power7 announcement, not Intel. Maybe the Nehalem EX really has them worried and they didn't want hp announcing Tukzilla and Nehalem EX before they got Power7 out the door.

"....But all that was announced was a dud that might benchmark well on spec...." Yeah, well, call back when you have a job where you might get to test either. In the meantime, I'm waiting until I get my grubby hands on both before making any judgements.

"....I believe once HP announces systems, they will only be of interest to those poor souls clining to HP-UX....." I'm guessing you also believe that Elvis is still alive and living as a petrolpump attendant in Utah. So, what happened to all the OpenVMS and NonStop users? Oh, sorry, I shouldn't expect such a poor troll to actually have any knowledge about another vendor.

".....Booooooooooooooorrrrrrrriiiiiiiiiiinnnnng!...." Yes, you are.

".....Back to my BEvERage." Sounds like you've had plenty already, but please don't let that stop you from taking your time and having plenty more. After all, it's not like we'll miss you or your non-contribution.

/Beyond laughting at, such a poor effort.

0
1
Gold badge
Troll

Trolls, trolls, trolls

Okay, serious question: is everyone in this thread honestly of the opinion that every other person in the thread is astroturfing?

(I admit, I have thought Matt Bryant worked for HP for ages, but never thought he was astroturfing in a professional capacity. His claim that he’s actually an admin for a company that buys Big Iron and just likes to troll the living piss out of people online seems perfectly credible to me. We’re geeks, trolling is in our blood.)

If everyone here honestly believes everyone else is a corporate shill, then the lot of you need some tinfoil-hat-ectomys. I ask you, honestly and truly, who the merry fnord would astroturf Big Iron on El Reg? You'd think anyone who was considering spending [$supid_amounts_of_money] on this gear would not base their decisions on what was read in the comments section here. It strikes me that there is absolutely zero return on income from spending money astroturfing Big Iron in the El Reg comments section.

If in the other hand we are all aware that nobody in the thread is a corporate shill…then this entire thread has simply become one gigantic trolling wank-a-thon. This is (by internet standards) perfectly normal and acceptable, and by all means everyone continue.

Still, I am beginning to fear some of you folks might be serious about the astroturfing accusations, and this worries me...

0
0
Silver badge
Happy

Dear Ali - Integrity Virtual Machines scale better than PowerVM

For those not familiar with the hp documentation tools available to customers, there's a website called Product Bulletin that has detailed specs (Quick Specs) on a lot of hp products;

http://h18004.www1.hp.com/products/quickspecs/productbulletin.html#intro

Just to keep the Compaq trolls happy, I'll admit this was one of the few good ideas from the old Compaq days. ;)

There is a Quck Spec sheet for Integrity Virtual Machines at;

http://h18004.www1.hp.com/products/quickspecs/12715_div/12715_div.HTML

Please note the line "up to 20 virtual servers per processor core" - so that means a theoretical limit of 80 VMs per Tukzilla.

Now, run on over to the IBM website and take a peek at this page;

http://www-03.ibm.com/systems/power/software/virtualization/features.html

Please note the limitation "Enables up to 10 partitions per processor core" - so Power7 will need twice as many cores to cater for the same number of VMs as Tukzilla.

Ooooh, busted!

0
1
FAIL

Integrity Virtual Machine does not scale like PowerVM

Matt,

IVM has a limit of 8 threads per virtual machine. Thanks for posting the link to confirm this:

===> "Support of larger VMs (max: 8 virtual processor core)"

each Integrity virtual machine can only scale to 8 threads or the equivalent of one Tukwila chip.

The number of VM's supported per core is not interesting when a Power7 4 socket box supports 320 virtual machines. What is important is any PowerVM virtual machine can scale to 128 threads based on weightings on that four socket box

PowerVM can scale any virtual machine to 64 cores today with 256 threads, soon to be 256 cores and 1024 threads.

The reason HP did not yet announce the systems is because they would have to admit they will only have one 2 socket rack based system. All systems above two sockets will be retired and customers will have to move to blades cobbled together to create scale up systems. The rx3600, rx6600, rx7640 and rx8640 do not have any rack replacements.

vpar and npar will not be supported on the Tukwila systems either.

HP's "blade everything strategy" is more like "blade only strategy".

This strategy does not make sense given the HP DL980 system (Nehalem EX) is an 8U 8-socket rack system with gluefull chips for scalability.

Juniata

0
0
Silver badge
Stop

Juanita is just cut'n'pastng the same posts into any Itanium thread.

Please, your FUD is so poor, at least try and improve on each itteration instead of just dumping the same junk every time. I'll save the readers some time and just point them to the other Itanium thread you posted exactly the same bit of waffle, and then they can read the reply I put their completely debunking your clueless statements.

http://forums.theregister.co.uk/forum/1/2010/02/09/intel_tukwila_feeds_speeds/

Please, someone get some new IBM trolls, their latest ones are failing miserably and us trolls have standards, you know!

/SP&L, in a pitying kinda way.

0
2
This topic is closed for new posts.