Is HP Paying Intel To Keep Itanium Alive? 216
itwbennett writes "In a court filing, Oracle accused HP of secretly contracting with Intel to keep making Itanium processors so that it can continue to make money from its locked-in Itanium customers and take business away from Oracle's Sun servers. Oracle says that Intel would have long ago killed off Itanium if not for these payments from HP. For its part, HP called the filing a 'desperate delay tactic' in the lawsuit HP filed against Oracle over its decision to stop developing for Itanium."
Support (Score:5, Interesting)
Maybe Oracle should come up with better and faster servers so that they can win customers on their own merits?
Maybe Oracle should do something useful, period. (Score:5, Insightful)
Maybe Oracle should do something useful instead of being a massive patent troll and distributor of obnoxiously terrible software.
Re: (Score:3, Insightful)
You sound pretty biased. You sound like someone who inherited an all Oracle shop. Sorry :P
Comment removed (Score:5, Funny)
Re:Maybe Oracle should do something useful, period (Score:5, Insightful)
Oracle agreed under contract to support this platform on their products. They got good valuable consideration for that. Now they don't want to hold up their end. Well that's too bad. A deal is a deal.
I have no pity for either HP or Intel on this one - they're taking a bath with Itanic, as some of us said they would 7 years ago and more. But at least they're not having to be sued to keep their promises.
Re: (Score:2)
Re:Support (Score:5, Insightful)
Re:Support (Score:5, Interesting)
It's probably more that Oracle doesn't want to support Itanium anymore, but I'm guessing that so long as Itanium is viable they're stuck supporting contracts that they have with HP. HP is in the middle of suing Oracle for their declared end of support for Itanium products. If Intel continues to make Itanium at HP's behest, that might leave Oracle on the spot.
Sucks to be Oracle's contracts department, but that's what happens when one doesn't write in a good escape clause. It probably legally doesn't really matter why Intel is still supporting the Itanium line, because I'd bet that Oracle never saw this one coming, but since Oracle is a third party to Intel and HP's business dealings as far as the contracts between the two, there's probably not a lot more than complaining that they can do.
Re: (Score:3, Insightful)
The Itanium could be a very nice processor, if they continued developing it. Although I'd suggest using a new brand name, after the total disaster of the first version. A pure 64-bit chip with no limitations due to legacy architectures has a lot of potential, potential Intel never really took advantage of. It didn't help that they never wrote a decent compiler for it.
Re: (Score:3, Funny)
I hear Alpha AXP's not being used...
Comment removed (Score:5, Informative)
Re:Support (Score:5, Insightful)
I hate to start any conspiracy theories, but look at it from Intel's perspective. Intel likes a good skunkworks project. Try something new, if it pans out you make a mint, if not, well, cost of doing business. Take the tax deduction. (Incidentally, that is where the Core architecture came from. Israeli design team making improvements to the old P6, never expecting to need it outside of maybe laptops.)
So they come up with this crazy VLIW idea and realize it will cost a ton of money. At the same time, they can convince HP to transition away from their existing RISC architectures (PA-RISC and Alpha) and in so doing get them to pay a big chunk of the R&D costs. Then, if it works out, great! Intel is now the sole supplier of Itanium chips for HP's high end servers. And if it fails, great! Two more non-Intel RISC architectures dead and out of competition with x86, and Intel gets HP to pay half the cost of their own execution.
And at that point, once Intel is in the position where success or failure doesn't matter to them because they sell the same number of chips whether they're Itanium or x86, success becomes the more expensive option. Why keep developing new models of Itanic when you've already got a Xeon that is better in every significant way?
Re: (Score:3)
The problems with IA64 were many, but the biggest one was closed source software... IA64 makes a very good Linux box, running an entirely open source stack but few if any closed source applications were ever ported to the platform.
Another problem was Intel making their own compiler instead of improving gcc, since the vast majority of software capable of running on ia64 is open source and most of that is generally (and sometimes can only be) compiled with gcc.
Re:Support (Score:5, Interesting)
Another problem was Intel making their own compiler instead of improving gcc
Intel did improve GCC, although GCC at the time of the Itanium release was completely useless at optimisation. Modern GCC is still a joke at optimisation compared even to something like Open64, and Itanium needs more effort than any other target architecture, yet gets far less manpower because no one cares about it. LLVM dropped the Itanium back end a few months ago because no one wanted to maintain it (and the few people who might have been vaguely interested didn't have access to the hardware).
Re:Support (Score:5, Informative)
So they come up with this crazy VLIW idea
Who's "they"? Intel, or HP [hp.com]?
and realize it will cost a ton of money.
Which, as I understand it, is why HP partnered with Intel (not the other way around).
At the same time, they can convince HP to transition away from their existing RISC architectures (PA-RISC
Which, as I understand it, was HP's intent even before they got Intel involved.
and Alpha)
Which was, at the time the HP-Intel partnership was announced, DEC's RISC architecture - DEC hadn't even been bought by Compaq yet, much less Compaq bought by HP.
Re: (Score:3)
Intel didn't 'come up with this crazy VLIW idea'. Itanium was the second VLIW x86-killer from Intel [wikipedia.org]. Someone senior at Intel in the late '80s to '90s seemed to have been absolutely convinced that VLIW was the future. Probably a hardware person, since hardware people always seem convinced that any problem can be fixed by more complex software (software people, in contrast, know that problems can always be solved by more hardware).
If they weren't convinced that IA64 would succeed, they'd have made their
Re: (Score:3, Insightful)
Poor performance, high cost? That sounds exactly like systems that big business love the best (judging from looking around me).
Re: (Score:2)
Requiem Itanium: the clock won't ramp.
Itanium predates AMD64 and was one way out of the 4GB limit. PAE and AMD64 were two others. That the clock would never ramp was provable in 2004, but by then it was too late. The promises were already made. Now we've got this immense chip that sucks watts, runs slow, but is very reliable. It will see two more revs before it goes into the dark - and a new enterprise CPU architecture will take its place.
Re: (Score:2)
Re:Support (Score:4, Interesting)
Intel was quite capable of writing a compiler for it, they chose not to write one that was any good. Make no mistake, it WAS a choice. Their software divisions (they have many) are a mess, their contractor rates are terrible and the politics are cruddy. However, these are all fixable. Now that Intel owns the CILK++ code, they have a better chance than ever of doing it right -- if they can be bothered. Compilers aren't rocket science.
As for the original Itanic - they could have made the Itanium 2 from the get go. Again, they chose not to, for political reasons likely. Again, it was a choice. The new Itanium, the Itanium 9300, actually looks like a credible contender for HPC - again, if they get the compiler right.
Re:Support (Score:5, Insightful)
Compilers aren't rocket science.
Indeed not. They are far more complicated than that.
Re: (Score:2)
A pure 64-bit chip with no limitations due to legacy architectures has a lot of potential
And it's called AMD64. The x86 support is almost literally glued onto Hammer, let alone anything later. Every modern x86 CPU since the Am586 decomposes x86 instructions into micro-ops which are executed by an internally-RISCy core.
Re: (Score:2)
The AMD64 has 16 64-bit registers, which is merely a logical next step from the 8 32-bit registers of the prior generation. According to Wikipedia: AMD64 still has fewer registers than many common RISC ISAs (which typically have 32–64 registers) or VLIW-like machines such as the IA-64 (which has 128 registers). The "no execute bit" is new with the AMD64, but none of the other underlying components seem to be anything more than basic upgrades. Long mode is just a flat address table and you've been able
Re: (Score:3)
The AMD64 has 16 64-bit registers, which is merely a logical next step from the 8 32-bit registers of the prior generation. According to Wikipedia: AMD64 still has fewer registers than many common RISC ISAs (which typically have 32–64 registers)
Except for one rather common RISC ISA, but that one's mainly used in stuff like mobile phones, not Big Honking Servers (although with ARM64 they're at least thinking about servers - it might be more "tons of blade processors" than "a smaller number of Big Honking Processor Engines", though).
TheAMD64 is not a "pure" 64-bit chip. It is a chip that operates in 64-bits but has an internal architecture that has not significantly changed since the days of the 4040.
Internal architecture, not instruction-set architecture? If you truly mean "internal architecture" in the sense of "implementation of the instruction set", do you really mean to say that a superscalar out-of-order pipe
Re: (Score:3)
TheAMD64 is not a "pure" 64-bit chip. It is a chip that operates in 64-bits but has an internal architecture that has not significantly changed since the days of the 4040
Wow, so in one article you've shown that you know nothing about compilers or architecture. A few 'insignificant' changes since the 4004 that are present in any modern x86 CPU.
Of course none of those are engineering...
Re: (Score:3)
All designs have limitations. Nor would I claim that the Itanium is the greatest processor out there. It has some great ideas, the second iteration was actually a decent invention, and it shows some nice creative elements. The flaws in the concept could only be revealed by actually implementing it and - as Linux kernel developers keep pointing out - actually putting it out there to test. Developers can't put anything, hardware or software, through all the real-world experiences, only users can do that and d
Re: (Score:3)
Re:Support (Score:5, Interesting)
I completely agree. Itanium was a boondoggle years before it shipped. But if you were stupid enough to buy into all the marketing, at least HP hasn't just abandoned you. Better to have the choice to leave than to be pushed off. Besides, nowadays the Itaniums suck much less than the first couple of generations did.
Re: (Score:2, Interesting)
There is one area where the Itanium has shined: VMS
Once Alpha had reached endgame, The Itanium filled the gap.
X86 systems just do not have the I/O bandwidth needed for VMS applications.
And HP is contractually bound to support VMS and Oracle is bound to support RDBMS installations.
By attacking Itanium, they attack VMS and in turn RDBMS.
Re:Support (Score:5, Funny)
here you go
http://en.wikipedia.org/wiki/Chevrolet_Citation [wikipedia.org]
Re: (Score:2)
How long have you been waiting to do that? And are you actually going to give him the car?
Re: (Score:2)
Not really, I do get the joke. Which is why I asked "how long have you been waiting to do that?"
Re: (Score:2)
It was a question, not a statement. The man needs a citation! Its just a question of how far he is willing to take the joke. I don't go in for half measures.
Re: (Score:2)
Also turns out Itanium is good at some things (Score:5, Interesting)
No, you'd never want it in a desktop, much though Intel hoped that would be where it went, but there is something to be said for what it can do in ultra high end servers with a ton of CPUs.
What you want for a CPU for a bigass compute server isn't always what you want for a desktop. Hell you can see that even with Sun's new Ultrasparcs. Different from both the x86 and Itanium, they are all about tons o' threads. They offer up to 8 threads in hardware per core on the newer ones. Such a thing would be totally useless on a desktop, a waste of silicon. However on, say, a web server such a thing could be very useful.
Itanium isn't the One True Way(tm) for processors, but they are useful for somethings, which is part of the reason HP likes them.
Re: (Score:2)
No, you'd never want it in a desktop, much though Intel hoped that would be where it went, but there is something to be said for what it can do in ultra high end servers with a ton of CPUs.
For the cost savings of going to basically anything you can get at least twice as many cores and you can blow itanic out of the water with an ice cube, it doesn't take a berg.
Itanium isn't the One True Way(tm) for processors, but they are useful for somethings, which is part of the reason HP likes them.
HP likes them because they are expensive and they can mark them up a lot. But nobody actually wants them. People took them because they were forced into upgrades involving them. Yuba college has an 8-way itanic because it was the ONLY available upgrade path for their student records system which was formerly on an Alpha. It's massive o
Re: (Score:2)
Itanium in its earlier incarnations were actually pretty poor for large multi processor servers too, in that the multi processor support used a shared bus architecture... Shared bus doesn't scale very far, so makers of very large servers have to develop their own glue logic to connect many processors together.
Re: (Score:3)
Re:Support (Score:5, Informative)
Re: (Score:2, Informative)
Unfortunately, Oracle claims that no such contract exist and it has publicly challenged HP to produce such a contract and HP has so far refused to take up the challenge. So far HP has not given any credible evidence that analyst believes is valid and its stock price reflects this view.
Re: (Score:2)
Yes, but producing the contract would make Oracle back down, which is profitless. Getting into a lawsuit and winning would be worth a lot of money.
Re: (Score:3)
There's always promissory estoppel.
Did Oracle let HP rely on any implications?
Re:Support (Score:5, Insightful)
Aside from the fact that the flow-on consequence is that oracle then needs to develop the ia64 oracle side - I still cant see why oracle think this is something worth even mentioning.
HP paid intel to keep making a chip HP uses - OH FOR SHAME! Or is the big thing about it the "secret" bit cause well, contracts like that do tend to be rather "sensitive".
But "oracle whinges cause HP tries to keep its IA64 customer base from moving to oracle servers" just sounds kinda ridiculous. Even reading the article is really not helping me get the problem oracle are trying to get at here. It reads like:
Oracle to HP: We would like to steal your customers please
HP to Oracle: Um, no thanks?
Oracle to HP: HAH, NO ITANIUM FOR YOU!
HP to Oracle: im sorry, but see this piece of paper says you cant do that
Meanwhile at the HP cave:
HP to Intel: heres some cash to continue IA64 development work
Intel to HP: Sure, no problem, we'll make silicon for you, we do that.
Meanwhile back at the Oracle Cave:
Oracle to Universe: WAAAAH HP WONT LET US STEAL THEIR CUSTOMERS.
*much thumb sucking ensues*
Now if HP had pain intel to stop making the IA64 to gimp dell (or someone else) for instance, then sure thats worth mentioning.
Re: (Score:3)
Comment removed (Score:5, Interesting)
Re: (Score:2)
Interestingly, Microsoft used to have a contract with DEC/Compaq/HP to continue producing Windows for Alpha, and for the "server" platform to be at parity with versions for other architectures... Compaq let them off the leash when they decided to drop Alpha.
Re:Support (Score:4, Insightful)
What is wrong with a software publisher saying they will stop supporting a hardware platform in a future release? Redhat and Microsoft also dropped support for Itanium.
They are not just a software publisher. They have near monopoly levels of control on the big-iron database market and they are using it to leverage their otherwise anemic hardware platform. Whether that rises to the level of "tying" that is considered anti-competitive is for the courts to determine.
Re:Support (Score:5, Interesting)
no monopoly, plenty of other big-iron databases besides Oracle around. DB2 is the real big-iron database, costs less, scales bigger (on Unix to over 100 servers, on mainframes to 32 data sharing groups, each one can be made of multiple MVS systems), performs better
Re:Support (Score:5, Funny)
DB2 is the best big-iron database, and MS-SQL is the best big-iron database for people who don't really like databases.
I guess that leaves Oracle as the database for people who like databases, but not too much. The Mitsubishi Lancer of databases, as it were.
Re:Support (Score:4, Insightful)
Re: (Score:2)
Precisely how is this anticompetitive? It might screw over Oracle's ability to sell these companies new hardware, but ultimately this is market forces in action. IBM has found it to be more profitable to pay Intel to continue producing the chips that it needs in order to maintain those computers so that the contracts can continue to their completion. And if there aren't any contracts in play, then I'm not sure why it is that Oracle can't do something similar. I'm not sure I see how this is any different fro
Re: (Score:3)
What about Apple? Their OSs drop support for hardware platforms on a regular basis. (I know it's an Itanium discussion, but...)
Apple sell computers mainly to home computer users and some professionals; they don't sell enterprise-class servers, and I have the impression that "sorry, that machine you bought a few years ago has been kicked to the curb in our new release" might not go over as well in that market (although I also have the impression that "here's a shiny new release of our {OS, database, etc.}, you should upgrade to it right now" doesn't exactly go over well in that market, either).
HP is run by Vogons... (Score:4, Funny)
...and Itanium is the Ravenous Bugblatter Beast of Traal.
I'm not sure what sort of Faustian Bargain HP made with Intel over Itanium, but it certainly had nothing to do with quality products or customer service.
Anyone sane bolted for Linux long ago.
Re: (Score:2)
Linux supports the Itanium. Linux supports the VAX. You can probably find a Linux port for the intelligent toaster from Red Dwarf.
Re: (Score:2)
You can probably find a Linux port for a commercially-available toaster, too.
Re: (Score:2)
I think the point is that once you're on Linux, you aren't tied to any particular architecture. Which causes you to choose not-Itanium as your architecture.
Re: (Score:2)
If you want high-end processing, you can do more with the Itanium than you can with the X64 architecture.
Re: (Score:2)
Maybe HP shouldn't act fraudulently in signing contracts?
Re:Support (Score:4)
Neither should Oracle. Hell, in a perfect world, both companies would donate all their assets to responsible companies and then quit.
Re:Support (Score:4, Informative)
I guess Oracle in their infinite wisdom have decided that a processor in a very marginal market is the reason Sun hardware sales are dying/dead, and not the fact that the Sun line is a choice between Intel (available from everyone) or Sparc (either slow and questionable performance, or power / rack space hungry ).
Considering it is actually Xeons, x86, and IBM hardware in the virtualization space that is the main market now, I don't see how this legal sidebar with HP does Oracle any benefit.
It's not about better and faster servers (Score:2)
For some reason, HP seems to think there are contracts obliging Oracle to keep selling and supporting their database software products for Itanium and Oracle seems to think they have no obligation, since Itanium would be dead and buried if HP wasn't secretly paying Intel money t
Re: (Score:3)
well...
https://en.wikipedia.org/wiki/Apple_Lisa#The_end_of_the_Lisa [wikipedia.org]
In 1987, Sun Remarketing purchased about 5,000 Macintosh XLs and upgraded them. Some leftover Lisa computers and spare parts are still available today.
In 1989, Apple disposed of approximately 2,700 unsold Lisas in a guarded landfill in Logan, Utah, in order to receive a tax write-off on the unsold inventory.
Re: (Score:2)
I think that Oracle wanted to sabotage HP by shitting on Itanium.
Oracle's naysaying is nothing more than FUD designed to undermine market confidence.
Thing is if enough vendors believe it, it becomes a self fulfilling prophecy.
Haha, "Locked In" customers (Score:2)
Re: (Score:2)
Locked up?
Que? (Score:4, Insightful)
Since when are companies not allowed to pay each other for services?
HP is contracting chip production and development out to Intel.
So what? Who is harmed?
Re: (Score:3)
Re: (Score:2)
It's entirely probable that the terms of the contract contain an escape clause if the platform is discontinued. However it's unlikely that the terms of the agreement stipulated that IBM couldn't pay Intel to continue manufacturing the chips.
Ultimately, if Oracle wins the suit that would be very bad indeed as companies should be allowed to pay for products to be continued indefinitely. The big concern with antitrust law would typically be for companies to pay other companies not to produce chips or to lock t
Re:Que? (Score:4, Insightful)
Yeah, I don't get it. I don't see how Oracle could make an anti-trust case out of this, as obviously there's no monopoly (or anything approaching one) in the space they're operating in (high-end non-x86 servers, basically the space between mainframes and regular x86-64 servers). If HP wants to pay Intel to keep making those crappy chips, why shouldn't Intel take the money and do so, as long as this makes up for whatever they lose by not using those resources elsewhere (like making their regular chips)?
As for harm, obviously, Oracle is harmed by this since this keeps them from monopolizing this market, but too bad so sad.
And the villain here is...Oracle! (Score:5, Insightful)
"...take business away from Oracle's Sun servers."
Trust me Oracle, the only company that's having the slightest negative impact on your server sales is...Oracle.
Solaris 11 shipped last week. They added code to prevent it from running on the UltraSparc processors. Thanks assholes.
Re:And the villain here is...Oracle! (Score:5, Informative)
You mean code to prevent it from running on Ultrasparcs IV+ and anything earlier: http://lildude.co.uk/solaris-11-end-of-support-for-legacy-hardware [lildude.co.uk]
kind of surprising as many customer plan for more than 7 years with large Unix servers, IV+ was introduced in 2005.
Contract fab? (Score:5, Insightful)
So, HP has a processor that they use a contract fab to build. It's just that in this case the fab belongs to Intel. Big whoop.
In other news... (Score:5, Insightful)
In other news most companies will kill products that don't have paying customers. HP is paying to make sure their supply chain stays open to support their customers, Intel has a customer for Itanium so they're maintaining production of the product. Oracle's a whiny brat who's pissed that customers that still have support on their older stuff have less of an incentive to change providers... If Oracle can't give them a compelling reason to leave that isn't "your old stuff isn't supported anymore 'cause we sued intel to stop support for your hardware" I don't have much sympathy for Oracle
Just sell Intel's Itanium division to HP (Score:2)
I am thinking Intel should sell the Itanium division back to HP once they are sure no other vendors are using it.
Re:Just sell Intel's Itanium division to HP (Score:5, Informative)
Re: (Score:2)
Sorry, didn't realize other vendors are using it. If so ignore this.
A few things... (Score:5, Informative)
http://smarterquestions.org/wp-content/uploads/2011/10/UNIX_Revenue_08-2011.png [smarterquestions.org]
Second, comparing Oracle suddenly killing support to Microsoft and Red Hat killing support is ridiculous. Red Hat is continuing to develop the 5.x tree for IA64, despite the fact that maybe 5% of Itanium customers ran RHEL. Oracle, on the other hand, is just suddenly saying "No more. Nada." despite the fact that they build key apps for all three HP Itanium operating systems (Rdb for VMS, Oracle for HP-UX, Tuxedo for Nonstop.) There's also the fact that Oracle has its own competing UNIX OS and processor, one that hasn't performed particularly well in comparison to Power or Itanium for several years. The whole thing just looks like Oracle is being a bully.
Also not a surprise for MS (Score:2)
They've never been big in the ultra high end market. The area where they do tend to have stuff, well x64 has been making a big entrance there. These days you can get an x64 system with 2TB of memory and that really tends to do the trick for large databases, which is one of the main ultra high end markets you see MS in (MSSQL is a serious contender as a high end DB server).
These days, most things aren't going with one ultra high end system, they are going with a cluster of regular systems. Those clusters are
Re: (Score:2)
>windows
>itanium
Excuse me?
Windows on Itanium is being EOLed.
http://arstechnica.com/microsoft/news/2010/04/microsoft-its-the-end-of-the-line-for-itanium-support.ars [arstechnica.com]
Re: (Score:2)
Re: (Score:2)
YOU DO NOT TELL YOUR CUSTOMERS THAT YOU ARE EOLING YOUR SOFTWARE UNLESS YOU ARE SERIOUS ABOUT GETTING THEM TO MIGRATE OFF THE PLATFORM.
For reals. Migrating off of one system to another takes time. You don't play fucking games. Microsoft knows this. That's why they gave everybody enough warning.
I read the message. The GP is trying to convince people to buy Windows for Itanium, when Windows for Itanium is a DEAD END.
It is irresponsible, and frankly, an insult to the intelligence of potential customers.
W2
Re:A few things... (Score:4, Insightful)
Oracle's naysaying about Itanium is nothing more than FUD intended to undermine confidence in a platform relied upon by one of their competitors.
Uh, what? (Score:5, Insightful)
Re: (Score:2)
Itanium is far from crappy, it's a much better architecture than x86 for transaction processing. People use old software because it works. If you were a medium sized company that spent $10 million on a custom ERP, why would you spend another $10 million every few years to do it all over again? Then you get to train everyone and work through the kinks and bugs again... Most companies just want to use what works.
Re: (Score:3)
Itanium is far from crappy,
True, but...
it's a much better architecture than x86 for transaction processing.
...false. You cite ERP as an example but that is exactly what Itanium is not special at. For business apps and database apps, Itanium is not really any more exciting than Sparc or POWER (and is not as good as POWER). For that matter, it's not any more exciting than x86.
Itanium has lots of 64-bit registers, so if you're doing engineering, science, chess computers, etc. and write Itanium assembler (or have a good compiler), Itanium rocks. But for business apps like ERP, CRM, general databases,
And history repeats itself... (Score:5, Interesting)
In 2002 Sun alleges that people don't buy their product because too many people choose to use Microsoft.
In 2011 Oracle alleges that people don't buy their product because too many people choose to use Itanium.
Lame, lame, lame.
Is McNealey now working at Oracle?
Deja vu:HP first ported Linux to Itanium &SCO (Score:3)
The SCO Group (then Caldera) which had purchased the rights to sell copies of the old Unix from Novell, sued IBM because the freely available Linux competed the SCO Groups old Unix offering.
So Oracle has become the next SCO Group, quick somebody tell PJ!
So Oracle admits they were lying? (Score:5, Interesting)
That's funny. Not to long ago Oracle stated [oracle.com] that they have proof that Intel was killing Itanium and that HP was harming their own customers by not admitting it. Now they say that the exact opposite is true; that HP is paying to ensure that Itanium stays alive. Either this change occurred after Oracle dropped their support for Itanium (unlikely), or Oracle just admitted that they have been printing libelous statements about HP, in addition to breaking their contract with them.
I hope the assholes pay for both.
Itanium is nothing more than PA-RISC64 (Score:2)
Itanium is very little more than PA-RISC64. The people who needed it didn't know they wanted it and when AMD x86-64 came out they ignored Itanium and SPARC to their peril. As a result, performance has suffered dearly. PA-RISC64 and SPARC64 are the true multiprocessing performers. Intel Xeon and Pentium represent at least one decade-worth of performance setbacks when it comes to multiprocessing performance.
But, Linux runs great on x86-64, so why bother with high performance? Too bad. We still struggle
Re: (Score:2)
Not every field needs massive numbers of threads. In my field, uniprocessor performance is more important for many tools. I remember very clearly when in the early 2000's a cheap Xeon processor-based box outperformed the fastest Sun boxes we could get. Why buy an expensive Sun when cheap x86 boxes will outperform it?
Itanium was a joint Intel-HP project (Score:4, Interesting)
Itanium was a joint Intel-HP project, remember? HP might well pay Intel to keep it alive.
The idea behind Itanium was that it had lots of new, different, patentable technology, so Intel didn't have to worry about clones. The problem was that it wasn't better technology. Just different.
Classic bad CPU architecture ideas of the "build it and they will come" variety:
In the spectrum of concurrency, shared-memory mulitprocessors with synchronized caches work, and clusters of powerful machines which communicate over networks work. Those are the extremes of the concurrency range. With the notable exception of graphics processors, no machine in the middle of that range has been a success. Such machines can be built, but are so hard to program they're always behind the classical architectures. The Cell in the PS/3 is the only example ever deployed in volume, and that nearly killed Sony.
Re: (Score:3)
AMD's 64-bit extensions arguably weren't a better technology either; they just happened to be in the right place at the right time, with a solution that maintained backward compatibility with the existing x86 code base. This, more than anything else, is what kept Itanium from ever gaining traction outside of a few niche server platforms -- when Intel caved and adopted AMD's 64-bit extensions, that was the final nail in Itanium's coffin.
I hesitate to call the i860 VLIW. Yes, it could fetch and execute an int
Re: (Score:3, Insightful)
AMD's 64-bit extensions arguably weren't a better technology either; they just happened to be in the right place at the right time, with a solution that maintained backward compatibility with the existing x86 code base
AMD's 64-bit extensions were a better technology because they maintained backward compatibility. You can't just write off this very important real-world consideration as if it were meaningless.
Re: (Score:2)
This thinking risks making CPUs a commodity, a must have to build a computer but exact specification secondary to GPU and other hardware. Now is the time for Intel to wake up and apply GPU-like technologies to new areas like database queries, or lose the future to AMD, NVIDIA or any newcomer that comes along. Not that deemphasizing x86 would be a bad thing for the industry.
Two companies behaving badly (Score:2)
HP-UX / Oracle / Itanium user here. (Score:5, Interesting)
I'm in the power industry. We have some applications that are only built in Solaris, HP-UX or AIX due to the underlying Cobol code, etc.
If we want to maintain certain regs, or have access to certain markets, we have to keep this particular app.
The day Oracle crapped on Itanium, we had to get HP in to tell us what the plan was as it would take us a few years to migrate to AIX if HP was really dumping it. (there is no way in hell we're running Oracle on a (now) Oracle operating system). Talk about vendor lock in. Woof.
Since then, I have been provided HP-UX and Itanium roadmaps for a ways out. (under NDA so no more details than that)
If Oracle wins on this, and really does dump UX, then I need to bring a bunch of AIX gear in and put a team of developers to work porting our custom code which means no optimization, no rewrites, no efficiency. All of our work to improve security, and kill off bugs will be wasted as we get it barely working in a new environment before we lose support. Just in case we get a nuclear project, etc.
The thought of training hundreds of people in a new system at multiple power plants and dozens of substations alone makes me nauseous. But if we screw up the migration process and wreck compliance, we could be out of business as the fines are incredible.
I'll bet half of this could have been avoided if when Hurd was found screwing around at HP, they could have just had him executed. Then he wouldn't be at Oracle and probably influencing this situation quite a bit.
Is there any other way it happens? (Score:3)
As far as I understand, customers always pay companies to keep product alive. Are they saying HP pays more per CPU than the price of a single order in retail? My bet is on a significant volume discount.
My only wish... (Score:2)
The real problem is that HP (Score:3)
Still hasn't started transitioning their HP-UX, Nonstop, etc customers to something else. It generally takes a generation or two before all the lagards get on board a new arch. Heck, HP was still selling PA-RISC machines a couple years ago, long past the point where it was apparent the itanic was a dead end.
It seems that HP is intended to keep forcing intel to make the itanium forever, but they have to have a fallback plan. The question is, does HP want to pay for full development of a chip so complex it takes 10x the manpower to design for, or are they going to bring back something like PA-RISC or Alpha that goes fast, without to much effort. Their only other alternatives seem to be jumping on the x86 or POWER bandwagon. I might included sparc, but outside of fujitsu, that seems pretty dead too.
Re: (Score:2)
Most likely Oracle doesn't want the product but is contractually obligated to buy it. And if the product is no longer available they can stop buying it without breaching their contract. I would assume that would be because Oracle now sells SPARC servers, but I'm not familiar enough with these systems to know.