IBM to Drop Itanium

Hack Jandy writes "Xbitlabs is reporting that IBM chose not to persue Itanium in their next generation server lineup because of the "market acceptance issues" of the platform. They will still continue with new revisions of Xeon servers, however. With IBM's investments in Power, I can't help but think the writing was already on the wall. The article also hints that IBM might start using Power in their high end server products."

Comments

Itaniums are the worst chips ever

[Anonymous Coward]

They cost a fortune and don't even run Doom

Re:Itaniums are the worst chips ever

[QuickFord]

True, but they make a real nice keychain!

Re:Itaniums are the worst chips ever

[m50d]

Wrong. (Notice IA64 sitting quite happily in the architecture list)

Stick a fork in it, it's done

[drinkypoo]

No one outside of certain specialized environments that demand loads of floating point is interested in itanic. Flush the damn thing already and work on EM64T, will you intel?

WTF does "The article also hints that IBM might start using Power in their high end server products" mean anyway? The processor is called POWER, and IBM already uses it in their high-end server products, like the ones that used to be called RS/6000. As for Power, well, show me a transistor that works without it.

Re:Stick a fork in it, it's done

[Anonymous Coward]

The pSeries is not IBM's high-end server line. The zSeries (Mainframes) is IBM's high-end line.

Re:Stick a fork in it, it's done

[Anonymous Coward]

It's nearly not that simple. For people who want raw CPU performance, IBM will sell them pSeries.

Re:Stick a fork in it, it's done

[IGnatius T Foobar]

IBM already uses it in their high-end server products, like the ones that used to be called RS/6000.

Actually, that hardly does it justice. pSeries (formerly RS/6000), yes, but also iSeries (formerly AS/400) is now POWER. The new OpenPower line of systems from IBM can run AIX, i5/OS (formerly OS/400), and Linux. In fact, it can run them simultaneously thanks to IBM's really good server partitioning technology (you can partition down to 1/10 of a CPU!).

I'm currently doing some development work on one of these boxes (running Linux on POWER) and let me tell you, it just smokes. Runs circles around Itanium, even before you start parallelizing (which is usually the case, since you're always going to have a dual-core chip, maybe even several of them).

IBM has absolutely no reason to continue supporting Itanium. It doesn't buy them anything. Itanic is an architecture nobody wants. If Intel hadn't sank so much R&D into it while still being able to live off the revenue from their 32-bit processors (and now, their AMD64 clones), Itanium would have been shelved a year ago.

Re:Stick a fork in it, it's done

[Amiga Trombone]

The processor is called POWER, and IBM already uses it in their high-end server products, like the ones that used to be called RS/6000.

I think what they're talking about is using them in mainframes (zSeries), which currently use a different processor than the iSeries (AS/400) or the pSeries (RS/6000). Apparently they're going to converge the hardware of their server lines as much as possible, and differentiate them mostly through the OS.

Makes sense, if they can leverage the same technologies across all their server lines, it'll help them cut their development and manufacturing costs.

Re:Stick a fork in it, it's done

[operagost]

For better or for worse, OpenVMS now runs on Itanium so please don't flush my favorite OS, please.

It would be just wonderful if OpenVMS were to disappear just because some fools moved it from one dead-end platform to another.

Re:Stick a fork in it, it's done

[Anonymous Coward]

"The new OpenPower line of systems from IBM can run AIX, i5/OS (formerly OS/400), and Linux"

Not really. OpenPower runs Linux ONLY. You may run AIX, i5/OS and Linux only on iSeries i5 server.

Re:Stick a fork in it, it's done

[Bert64]

I doubt it would be very easy to port OpenVMS to a platform which wasn't designed to accomodate it, such as x86 or amd64.. I would be interested to see how the itanic performs when emulating alpha code too, aparrently you can run alpha binaries on openvms/itanic now, but it would most likely be horrendously slow, especially considering the fastest alphas can hold their own against itanic anyway.

Re:Stick a fork in it, it's done

[Anonymous Coward]

Alpha was a niche CPU, Itanium takes it place. The difference is that Intel have money, and that Compaq wanted out of the CPU business.

From the summary:

[Noose For A Neck]

"The article also hints that IBM might start using Power in their high end server products."

What? IBM already uses POWER in it's high-end server products. What do you think they develop it for, anyway?

Re:From the summary:

[superpulpsicle]

http://www.llnl.gov/computing/tutorials/ibm_sp/

Here's a link to the history of IBM's processors POWER. This is one of the best sites out there IMHO, and it still seems mighty confusing.

IBM never had a good history of marketing their processors like Intel and AMD. They fight competition with raw numbers.

Re:From the summary:

[drinkypoo]

IBM doesn't market IBM products. They market IBM. You buy into IBM, you do everything their way, and everything will work. These days, they add "...and you will get the best possible performance" to that, too. Nothing out there has the performance of the latest-generation POWER processors. As IBM has been busy proving building supercomputer after supercomputer, it scales pretty well too :) Granted, highly parallel opteron processors are pretty slick, but given a level playing field I know what I'd pick. Intel introduced itanium too late and at too high a cost to make enough inroads before opteron started to take off, and now itanic has no chance to proliferate enough to ever become inexpensive. IBM has been putting work specifically into making their cores as modular as possible so they can easily turn them into other versions ever since the PowerPC 601, which is why we new PowerPC cores so closely follow the release of new POWER cores.

Re:From the summary:

[Anonymous Coward]

IBM doesn't market IBM products. They market IBM. You buy into IBM, you do everything their way, and everything will work.

Do you work at IBM? Because if you don't, make it clear you're stating an opinion and let those of us who work there tell our potential customers what we sell.

Your statement sounds like something from the 1970s or earlier when IBM sold all-IBM solutions. These days we know that no company can be all things to all people, so we'll offer a solution that may or may not be entirely made up of our products.

It's great if a customer buys IBM servers with IBM software and hires our Global Services to deploy them. But if "you buy into IBM" then why do our servers let you run several Linux distros or versions of Windows? Why does our software run on Solaris, HP-UX, Windows, Linux, and in some cases even Mac OS? Why do our services consultants build networks using servers and network devices from other vendors?

Re:From the summary:

[The_Wilschon]

which is why we new PowerPC cores so closely follow the release of new POWER cores.

I didn't know IBM was making intelligent machines yet?

Re:From the summary:

[Henriok]

What? IBM already uses POWER in it's high-end server products. What do you think they develop it for, anyway?

No they don't!
the pSeries and iSeries isn't considered "high end" by IBM, they are considred low end and midrange servers. The high end is the zSeries and they doesn't use POWER/PowerPC processors just yes. Word has it that the future POWER6 processor will converge the three server lines on one processor platform. The eClipz project is tied very closely to this. "e" as in eServer, "l" as in Linux, "i" as in iSeries, "p" as in pSeries and "z" as in zSeries will.. eclips the Sun.

Re:From the summary:

[macshit]

The eClipz project is tied very closely to this. "e" as in eServer, "l" as in Linux, "i" as in iSeries, "p" as in pSeries and "z" as in zSeries will.. eclips the Sun.

I knew IBM had good research people, but I never realized their acronym technology was that far advanced! My god, with an acronym like that ... their competitors might as well just give up now and save themselves the embarrassment.

I'll miss it

[m50d]

Say what you like, but Itanium was a nice architecture. The compiler is the proper place for the optimisations, the processor should be left to do the actual processing. It's still the most efficient way I know to do raytracing or anything multimediay, and I predict there will still be a market for them for some time.

On second thought, maybe they'll start appearing cheaply on ebay. That'd be nice.

Re:I'll miss it

[Schweg]

With Itanium, Intel attempted to tackle a set of issues that isn't new, and other companies have tried (and failed at) before. It's hard to shift almost all of the burden of optimization, because it's a case of "early optimization" (the root of all evil). Optimization by the processor at run-time allows one to deal with data-dependent issues, and base decisions on statistics gathered by modern processors (such as branch history, caching behavior, etc). Intel made a good try at it, but ended up making a very power-hungry processor that exposed a lot of complexity to the programmer, and whose advantages compared to other processors on the market were not very clear.

Re:I'll miss it

[Anonymous Coward]

Yeah, but those same ideas apply to Cell, except 1000x moreso. IBM is just repositioning, trying to build some framework they can leverage into Cell dev in the next few years.

Re:I'll miss it

[Waffle Iron]

The compiler is the proper place for the optimisations, the processor should be left to do the actual processing.

On the contrary, the compiler has no insight into the actual run-time behavior of the current dataset, and compiler development can lag updates in CPU features by many years.

Nobody knows how to optimize for the exact version of the CPU that a program is being run on than the CPU itself.

Re:I'll miss it

[Noose For A Neck]

Maybe in an academic, theoretical sense it's the "proper" place for optimization, but as far as I can tell, the lack of good compilers is one of the major misfeatures that ended up stalling adoption until better, more compatible architectures like x86-64 came out.

Re:I'll miss it

[e-r00]

I'll miss it too. It was the very best CPU for our computationally-intensive applications. Is usually won with P4s and Athlons with 2 times higher clock speed... We'll miss you, Itanium :(

Re:I'll miss it

[Imposter_of_myself]

It was a helluva space heater - even if you didn't need it for computin'.

Re:I'll miss it

[... James ...]

I'd like to see Microsoft's .NET framework really take off. If we get to the point where Windows and all the apps for it are JIT compiled, I think different architectures would have a much better chance of succeeding. I wonder where Itanium would be today if you could run your normal OS (and let's be honest, Windows is the normal OS for most people) and apps on it when it was introduced a few years ago...

Re:I'll miss it

[Anonymous Coward]

Java has had that for ages, and do some very neat CPU-specific optimisations.

I don't know about the Java perfomance on Itanium in particular though.

Re:I'll miss it

[roca]

Others have already corrected this comment but I feel I have to correct it too, because the notion is widespread and quite wrong.

Often the compiler simply CANNOT KNOW much about what is going to happen at run time. You should not expect an "omniscient compiler" to be developed any more than you should expect "omniscient weather forecasting" to be developed. In the absence of good information about the future, you have to be able to dynamically adapt to changing conditions. JIT compilers can do this to some extent, but generally JIT profile analysis and recompilation won't happen as quickly or efficiently as hardware implementations of adaptive features.

Re:I'll miss it

[alext]

Here's a JVM that targets Itanium specifically.

Re:I'll miss it

[Anonymous Coward]

It's surprising considering that they had already tried in the 1980s, with the i860, which also required compiler-level optimisation, and also failed miserably (despite having been the original target architecture for Windows NT, before the switch to MIPS).

Re:I'll miss it

[johannesg]

because it's a case of "early optimization" (the root of all evil)

While you may very well be right on this issue, it is taking this quote very much out of context. Early optimization in software is bad because it tends to reduce maintainability and wastes effort on code that is likely not performance critical anyway. In contrast, the need for maintainability in compiler-generated assembly is questionable, and it doesn't really matter if the compiler spends some extra time optimizing every last statement even if it is non-critical; unless you are on unbelievably large projects it just doesn't matter. What I'm trying to say is, the quote you provided doesn't apply here.

Sorry for ranting, but I can't stand it when people take these kinds of "common wisdoms" and then display a complete lack of understanding of the actual issues behind them...

Re:I'll miss it

[Anonymous Coward]

IIRC, Microsoft released Windows 2000 for Itanium in 2001. The problem was the early Itaniums were much slower than x86 systems, and, like the later ones, extremely expensive. If Intel had produced low-cost chips for PCs, it might have had a shot, but Itanium systems have always been in the same price category as the old UNIX/RISC systems.

Re:I'll miss it

[dvdeug]

The compiler is the proper place for the optimisations, the processor should be left to do the actual processing.

The theory behind the highly-touted JIT optimizations for the JVM is that it's often better to optimize at run-time, when you know the data, then at compile time when you don't. And compilers don't usually have even the minimal knowledge the programmers have about which switches will be taken.

Intel's iAPX432 should have warned it about depending too much on the compiler. The iAPX432, the replacement for the 80286, was an intrepit chip of unique design that was sunk, in part, by a lack of compilers that could create compentent code for it. The benchmark compiler would always use the 700-cycle procedure call instead the cheaper, more specialized procedure calls available, for example.

And what happens when the underlying chip changes? You optimize for a Pentium in very different ways than for an i386 or a Pentium III, especially in the ordering that the Itanium wanted to shove to the compiler. If we wanted to recompile our code for every new chip, the x86 series would long be dead.

Re:I'll miss it

[Schweg]

Sorry for ranting, but I can't stand it when people take these kinds of "common wisdoms" and then display a complete lack of understanding of the actual issues behind them...

Only if you view the issue in the most naive sense. There are a number of reasons to avoid early optimization, and code maintainability is only one of them. The issue I was referring to is attempting to optimize before you know what conditions to optimize for. Early assumptions (e.g. a general case, "what kind of sort should I use here?") can often make performance worse instead of better.

Re:I'll miss it

[subsolar2]

Intel's iAPX432 should have warned it about depending too much on the compiler. The iAPX432, the replacement for the 80286, was an intrepit chip of unique design that was sunk, in part, by a lack of compilers that could create compentent code for it. The benchmark compiler would always use the 700-cycle procedure call instead the cheaper, more specialized procedure calls available, for example.

Actually the iAPX432 predates the 8086 ... like itanium it was an attempt by Intel to make a quantum leap. The iAPX432 was an attempt at a jump from 8bit to 32bit with support for object oriented languages.

The 8086 was about the same as the EM64T is now ... a kluge to get in the 16bit market when they realized that the iAPX432 as a too ambitous design. The 8086 was designed in 6 months and basicilly added a 16bit ALU and segment registers to addess beyond 64K to the 8080 design.

Re:I'll miss it

[Bert64]

Opensource would solve the same problem too, Once a kernel such as Linux, a libc and a compiler is ported, most of the userspace apps are quite easy to compile and run, especially since 64bit opensource os's have been around for years, so a lot of code is 64bit clean nowadays. Just look at how quickly linux supported itanium and amd64.

Re:I'll miss it

[Bert64]

And unlike the other RISC systems, Itanium hasn't been around long enough for older systems to be available cheaply on ebay... Anyone can afford an old sparc nowadays to learn the architecture or write code, and their code will run just fine on a modern system.

Re:I'll miss it

[m50d]

Compiler development can lag the CPU, but it shouldn't. The compiler knows how to optimize for the CPU better than the CPU, because the compiler is dedicated to optimizing and designed to be, wheras CPUs were originally meant to ocassionally find time to, you know, actually do the processing. Assessing run-time behaviour is overrated. The CPU is deterministic, anything you can do at run-time you can do at compile-time - just add a detection routine if you really need to, most of the time you won't. Certainly there is no way run-time optimisation can be faster than compile-time, because you can just compile to the same opcodes that would have been executed when run-time optimizing in the case that that's the fastest way to do things. But most of the time it won't be.

Re:I'll miss it

[m50d]

x86-64 is not better, it's more kludgy and horrible than even x86. There is at least 3 levels of emulation going on there, it might seem nicer from a high level perspective but working with it it's really horrible. Yes, the lack of a good compiler for a while is what killed itanium, but it really didn't deserve to die, especially now icc is working and working well.

Re:I'll miss it

[m50d]

The compiler can know pretty much. A compiler can be updated and modified much easier than a hardware optimiser. And a ridiculous proportion of the typical x86 chip is doing optimisation rather than actual processing.

Re:I'll miss it

[m50d]

There is some benefit to JIT, but a lot of what is being claimed simply isn't true. I have yet to see any real performance increase over an ordinary interpreter except in some specialised benchmarks, and the slowness of java apps belies the claims of faster performance than compiled code. Yes, Itanium was perhaps too dependent on the compiler. But there are good modern compilers, and I think if it had been launched today it would work. I don't blame intel for overestimating compiler ability, I would have done so too.

Re:I'll miss it

[Anonymous Coward]

It seems like you can grab an Itanium v1 workstation for a few hundred bucks. Not bad for a 4 year old high-end box, and considerably cheaper than a PowerMac of the same vintage.

Re:I'll miss it

[cburley]

anything you can do at run-time you can do at compile-time - just add a detection routine if you really need to, most of the time you won't

It can't be that simple, or (and I have often thought about this myself) the fastest general-purpose CPUs would not have dedicated hardware to detect cache misses, TLB misses, etc.

And in a perfect world where complexity management wasn't a problem, indeed, much of what is presently reserved to special functions in a CPU would be instead made available to more-general mechanisms that clever compilers could exploit.

But complexity management is the problem. Move complexity out of the CPU architecture and it either becomes slow (you've serialized and sequentialized everything) or practically impossible for most of its users to properly program in order to regain the performance lost due to naively using the underlying architecture.

Then there's the (apparent, at least to me) fact that dedicated logic can always (?) be "shrink-wrapped" to smaller, faster cores than can equivalent logic implemented via dynamically programmable microcode/nanocode on top of a simpler hardware micro/nano-processor.

After years of trying to teach compilers to take advantage of VLIW (and EPIC) architectures, it still seems best, overall, to leave microinstruction scheduling, dependency analysis, and so on to the hardware via its own run-time analysis.

A simpler instruction set (that leaves low-level optimization to run-time decisions by the CPU) also has the advantage of allowing "vanilla" programs to much more easily run across a wide range of compatible hardware.

(Yes, "fat binaries" or equivalents can make some accommodations, but IMO the way to go, to really take advantage of the sort of theoretical capabilities you and I are talking about, is to run an OS that is beyond GNU in terms of being designed, from the ground up, to always run "source code" at a fundamental level -- that is, treating compilation as a behind-the-scenes activity.)

Re:I'll miss it

[Bert64]

Not in the UK you can't, i've tried to get my hands on one out of curiosity, they're very rare and i imagine they're only cheap because noone wants them.

Re:I'll miss it

[m50d]

If you delay the compilation to startup, you get a delay when launching. Processors are getting faster to the stage where it might be irrelevant, but it isn't to me on my hardware - any jit type system is really noticeable. As for the dedicated hardware aspect, ime generic processors always overtake dedicated hardware you buy pretty quickly. That's why the advice to people setting up raid5 or number crunching or so on which is meant to be used for a long time is always to do it in software, because software is flexible and gets updated. Your PII ran faster thanks to its on-die optimisers, but modern compilers could probably do a better job without them by now. And I think the upgrade cycle is getting longer, so the advantage of flexibility will really start to tell.

Re:I'll miss it

[cburley]

If you delay the compilation to startup, you get a delay when launching.

I'm not suggesting anything quite that naive.

Think of it this way: when you install, say, Fedora Core 3 (FC3) on your computer, you are experiencing a delay in startup, in two important ways:

• It took the package distributor a long time to generate the binaries for some generic system, but your system is probably different enough to deserve binaries optimized specifically for it; so, to get an "optimal" install, you're looking at installation, or startup, times comparable to a Gentoo install anyway
• The supposedly-optimized binaries you're running have all sorts of special code to discover what's special about your system, and somehow communicate that (mostly ineffectively) to all the other program binaries in the distribution, so you have to wait for that to happen anyway (you might be surprised how much of the startup code your system runs each time you do something is, in fact, unnecessary)

An OS designed from the ground up around source availability would not always suffer the former penalty and could more generally reduce the latter.

Yet, there would be nothing preventing distributing "pre-digested", or pre-compiled, variants of such an OS. Such variants would be, conceptually, little different from use of prefetching or caching at lower levels.

Speaking of lower levels, I do wonder how much faster a given system might be if the CPU didn't provide constant cache-miss and TLB-miss detection, among other things, and left that to the code.

That'd certainly make code-generation more complicated, but that's why we have computers, right?

Meanwhile, inside tight loops, once the code verified that ranges of addresses pertaining to the working set of a loop were in the L1 cache, mapped, etc., the inside-loop references could be substantially faster (premapped to hardware RAM addresses, for example).

Whaddya think?

Re:I'll miss it

[m50d]

I think you would always have the first penalty to the extent that it would annoy many users. I'm a gentoo user and I just run my upgrades in the background, but I've known lots of people who say the compilation time is unacceptable. And I don't think that designing from the ground up for compile from source would speed it up that much, because optimizing is hard, plain and simple. When computers have got fast enough that the time to compile the whole OS is, say, about half an hour, then such an approach will work, and it's a good approach. But for the moment I think most users will be using precompiled things, so unfortunately all the self-optimizing gunk makes sense on the whole. Maybe we could start moving to it being optional.

Agree with you 100% on the cache-miss and similar though. That's why itanium was good - less of the cpu trying to optimize its own code before running it.

I think the problem is the average user needs optimization but isn't willing to take the time to optimize their programs. Thus programs have to pretend to have been fully compiled while optimising themselves behind the scenes. So source as a standard distribution method would make sense. But it needs to be optimised on install rather than at run-time, and it needs to be fast enough that users won't complain. If you think that removing the gunk and having an OS based around from-source would let you compile in timescales similar to installing a binary package, then you're right and let's start working on it, but I don't think processors are quite fast enough yet.

Re:I'll miss it

[cburley]

[...]on the cache-miss and similar though. That's why itanium was good - less of the cpu trying to optimize its own code before running it.

But it doesn't seem to be good enough for the marketplace.

I'm not sure whether that's because it's ultimately suboptimal in theory (no compilers could ever work around such a design) or in practice (such compilers weren't buildable within the economic and time constraints necessary to make the system, as a whole, meet its original performance goals).

If you think that removing the gunk and having an OS based around from-source would let you compile in timescales similar to installing a binary package, then you're right and let's start working on it, but I don't think processors are quite fast enough yet.

I think that it's all possible.

I think a sufficiently-well-designed system would allow a distribution to consist of not just the raw source, and not the equivalent of today's executables, but of raw source plus something in between the two that allows a final optimization stage to focus optimization resources on the known "hot spots" and, otherwise, use cheap target-specific optimizations to generate the final code upon installation.

I think this sort of system could provide all sorts of advantages over the present architecture, be faster, more reliable, easier to debug, etc.

But, since I might well be wrong, or right but ahead of my time (or possibly behind?), I don't want to start working on it until I can satisfy at least one of the following two criteria:

• Create a small-scale prototype as proof of concept, demonstrate that it would actually work and produce better "numbers" than would be predicted by the old way (perhaps actually implement the prototype's design the "old way" as a separate track for direct comparison)
• Create tools that not only automatically generate much of the code for such a system, but much of the protocols, file formats, and APIs as well

(Asking a computer to do more work for me means giving it more power to decide how best to approach the task. For example, since I'm "asking" it to allow distributions of pre-digested source leading to faster generation of optimized code upon installation, either I have to design the format of those pre-digested files myself, or I can somehow teach the system to do it itself and maybe produce a better result.)

Accordingly, I've been slowly designing a simple, scaled-down replacement for the Unix OS that is intended to allow me to meet both criteria, or at least the first.

(But I haven't really written much up on it yet -- nothing online, certainly.)

Cell ?

[polyp2000]

lets face it when Cell arrives formally theres going to be little point in ploughing resources into something thats effectively headed for obsolesence

Re:Cell ?

[Harry Balls]

Cell is very fast, but only has single precision floating point, i.e. it will not qualify for scientific applications, which demand double precision.

Cell is going to be great for gaming and rendering and such, but you won't see scientific applications running on it any time soon.

Re:Cell ?

[Anonymous Coward]

that's not quite true. While you won't get 250GFlops/cpu double precision like you can with single precision, the Cell actually can allegedly do 25 GFlops/cpu DP. That's still ~4X the DP performance of the fastest cpus in circulation right now.

In addition, there are many uses for single precision in sci/eng applications that users will be able to take advantage of (solver preconditioning, low res parameter sweep apps, etc..). What's important is that high performance single and double precision are both available on the same cpu.

Re:Cell ?

[tuffy]

Cell is going to be great for gaming and rendering and such, but you won't see scientific applications running on it any time soon.

Considering Sony's previous chips, I expect the Cell won't even be that great for gaming. More likely it'll be an overhyped, underperforming CPU that won't even match the power of other contemporary consoles, much less that of a general purpose PC CPU.

Re:Cell ?

[coder.keitaro]

Cell is very fast, but only has single precision floating point, i.e. it will not qualify for scientific applications

Take a look here.

I also remember reading a while ago on slashdot that IBM is intending to use it on high end Unix workstations.
[Although I can't remember if it mentioned for scientific of rendering applications.]
Tried searching for that post ... to no avail.

Re:Cell ?

[mrm677]

The Cell will be heavily used in game consoles, thats about it. Don't buy into the IBM marketing machine.

Sorry but having 8 SPE "coprocessors" or whatever you call them is not for general-purpose computing. Programmers can hardly deal with multhreading on a symmetric multiprocessor. What makes you think that programming a heterogeneous machine will be easier?

Re:Cell ?

[Anonymous Coward]

Except for the fact that you can compile for Itanium, but won't be able to successfully compile for the CELL SPEs for several years. Unless of course IBM wants to hire a bunch of code monkeys to write hand coded scientific libraries.

Re:Cell ?

[Anonymous Coward]

Speculations are not an engineer's tool. Unless you have concrete data to support your claim, there is no need to bark in the dark. Itanium is a great chip, some of its architectural features are better than others. Cell is also reviving data-flow architecture to exploit memeroy bandwidth applications (image processing).

Re:Cell ?

[mrm677]

This is no flamebait. The single-precision floating point numbers are on the order of 250 GFLOPS/s. However double-precision is less than 10.

Cell will be good for graphics and gaming. It won't be useful for most scientific apps because of the double-precision issue. It won't be used on the desktop because of backwards compatibility and programmability issues.

Re:Cell ?

[Flaming Death]

Considering Sony's previous chips, I expect the Cell won't even be that great for gaming. More likely it'll be an overhyped, underperforming CPU that won't even match the power of other contemporary consoles, much less that of a general purpose PC CPU.

Another classic ill-informed console fanboy comment. Where does slashddot get these quality posters.. let me guess.. XBox roks!! right?

Re:Cell ?

[Hal_Porter]

The Cell is not going to be used in general purpose machines. It has a PowerPC core, but the performance is low compared to a G5. It also has a bunch or SPEs but these are not suitable for general purpose computation either - they lack MMUs for example, and only have access to a small amount of tightly coupled SRAM. The idea is that you split your application into a software pipeline, and implement the stages in different SPEs.

All this stuff doesn't matter for games consoles where you can code all the crucial bits in SPE assembler and the whole machine runs only one application, but it would kill performance on a general purpose OS.

http://arstechnica.com/articles/paedia/cpu/cell-1. ars

Re:Cell ?

[Dwonis]

From what I understand, programming the Amiga was a lot easier than programming the PC at the time, and the Amiga was essentially a "heterogeneous machine", as you describe.

Re:Cell ?

[Bert64]

Well, the processor was a much cleaner design, no segmented memory etc, and most of the time you were programming the cpu.. You also could guarantee the hardware and os (or you could totally bypass the os), whereas on an x86 machine you could have any number of incompatible sound/graphics cards and incompatible methods of accessing memory above 640kb.

Re:Cell ?

[Anonymous Coward]

You obviously have never written parallel software on a machine in the last 10 years.

Getting leaner, IBM?

[osewa77]

First they drop a PC line tha was not making them money. Then they drop a server line that's clearly not the future of that space. I think they're making some right decisions here. If the POWER platform succeeds, as it more likely would when resources are focussed on it, and it is accepted as a viable alternative to the PC platform, the ensuing competition would probably be good for all of us.

Re:Getting leaner, IBM?

[goMac2500]

Even as a POWER supporter, I find it hard to say it will be accepted as a viable alternative to the PC platform. POWER has existed since 1994 and its failed to make a huge dent in x86, even though it has always been much faster.

Re:Getting leaner, IBM?

[drinkypoo]

POWER (or at least PowerPC) has a good chance to dominate the midrange server market as that market increasingly accepts Linux. Linux has a tendency to make the most of whatever hardware it's on (if not right away, then eventually) and IBM has already demonstrated a willingness to put engineers on GPL products where it will benefit IBM. This is an amazing reinvention of self for a company whose licenses used to state that any software developed on their system became their property.

The reason POWER never made a dent in x86 is that Microsoft saw no significant returns from implementing PPC support and dropped the platform entirely. Consequently people couldn't run what they wanted on PPC. However, if people don't want Windows any more, then there's nothing tying them to x86.

Re:Getting leaner, IBM?

[jbplou]

POWER is not going to be accepted as a viable alternative to the PC platform. It is as likely as Debian being accepted by the general business world as a viable alternative Windows

Re:Getting leaner, IBM?

[Anonymous Coward]

No, the reason that Power(PC) never made a dent in x86 is that IBM promised everyone that it would scale better and it simply did not. Furthermore, IBM themselves quashed cheap PowerPC workstations due to internal politics surrounding OS/2, never provided good chipsets to third parties, etc.

Hey "Blame Microsoft For Everything" is fun, but IBM never seriously attempted to position PowerPC in the mainstream x86 market.

Re:Getting leaner, IBM?

[Master Bait]

I doubt POWER will ever go further in market share on the desktop than it is already because nobody can get a POWER ATX motherboard on the open market. There are no available off-the-shelf chipsets for POWER, no appreciable demand for POWERs outside of Apple.

When they make me President of IBM, the first thing I'll demand is inexpensive chipsets and reference boards (with free design licensing) for the Taiwan motherboard makers.

Re:Getting leaner, IBM?

[Detritus]

I doubt POWER will ever go further in market share on the desktop than it is already because nobody can get a POWER ATX motherboard on the open market. There are no available off-the-shelf chipsets for POWER, no appreciable demand for POWERs outside of Apple.

Businesses buy systems and solutions, not Power ATX motherboards with chrome-plated heatsinks and positronic cache boosters.

When they make me President of IBM, the first thing I'll demand is inexpensive chipsets and reference boards (with free design licensing) for the Taiwan motherboard makers.

Don't hold your breath. Why is it in IBM's interest, or their customer's interest, to promote a flood of cheap, mostly compatible clones with little or no support?

Re:Getting leaner, IBM?

[Master Bait]

Because selling CPUs is profitable. And the more you sell, the cheaper each one is to make.

Re:Getting leaner, IBM?

[antime]

Not POWER, but there are several PowerPC boards available. Genesi's Pegasos 2 is based on a Marvell system controller and Mai Logic make both chipsets and motherboards. Momentum make PPC970 evaluation boards, but I don't know if they qualify for the "open market" requirement.

Re:Getting leaner, IBM?

[drinkypoo]

IBM sold CHRP or PREP-compliant (I forget which) PowerPC boards. Few people/companies bought them. There was simply not sufficient interest to get the prices down to the point where it was worth it. You know, kind of like itanic. Holding on a little longer might have made it happen, but then PPC support was dropped from Windows. It's all speculation at this point, though...

Re:Getting leaner, IBM?

[hackstraw]

you are right. Who would want to use an IBM powerpc processor anyway?

Re:Getting leaner, IBM?

[Anonymous Coward]

Yes, Apple was good for 5-10% of the mainstream market, but the original plan was to make a chip that would get 30%+ of the market, and that meant Windows.

IBM using Power based CELL CPUs

[Anonymous Coward]

IBM will be using the new Power based CELL CPUs in their new servers. Two of my friends are already working on the new architecture but unfortunately can't talk about any of the details. Both, IBM and SONY, will be using CELL CPUs in virtually all of their new products from DVD players to supercomputers. Anyone wants to make a bet with me?

Re:IBM using Power based CELL CPUs

[drinkypoo]

If next-generation media formats demand high floating-point performance, then using CELL makes perfect sense. Otherwise, not. Personally I want to see a VAIO with a Cell running Linux, that would really make my day if it wouldn't cost a hojillion dollars. Maybe if it were also a Playstation 3 it would actually sell.

Re:IBM using Power based CELL CPUs

[Anonymous Coward]

Mod parent -1 wrong.

Re:IBM using Power based CELL CPUs

[vik]

If the Cell is all it's cracked up to be, a couple of them on an AGP video board are going to trash the host CPU performance-wise. Will we see a huge twist of irony in which PCs are just reduced to a shell that supports the SMP Linux system hosted on the "graphics" card?

It's not so far-fetched. Linux runs on Cells, they're designed for good graphics, they'll be widely available, and if all else fails there's Cygwin to handle X output on a standard graphics card for die-hard Windows bunnies.

Vik :v)

Re:IBM using Power based CELL CPUs

[pmonje]

OK, I'm not supposed to talk about this, but I have two friends who work for the illuminati and all the new mind control chips are going to be based on the CELL. Unfortunately they couldn't give me to many details because of the NDAs that their reptilian overlords made them sign. Anyone wanna bet me?

seriously slashdot needs a "-1 talking out your ass rating"

Re:IBM using Power based CELL CPUs

[RzUpAnmsCwrds]

"Both, IBM and SONY, will be using CELL CPUs in virtually all of their new products from DVD players to supercomputers."

Doubtful. DVD players rarely have much of a processor at all - usually a low-end CPU combined with a custom DSP. All decoding is done in hardware, so there's no need for high FP performance like the CELL offers.

Re:IBM using Power based CELL CPUs

[rhuntley12]

A friend from an AC? And Sony may start using cell cpu's in their new stuff, but doesnt mean they still will in 2 years or that it'll even be as good as they said. Sony is good at really hyping stuff up. Most people that have just read in magazines and what not think that the cell processesor in PS3 will be the most powerful processor on earth, just like people didn't want the PS2 sold to china because it could become a supercomputer.... I'll wait until it's really out.

Re:IBM using Power based CELL CPUs

[Bert64]

If the cell is mass produced it may become cheaper to produce than a custom dsp, also this will allow dvd players to be upgraded more easily in the future to handle new formats, tho companies nowadays dont want you to upgrade your existing hardware, they would prefer you to throw it away and replace it.

AMD64

[bstadil]

work on EM64T, will you intel?

baloney, call it by its proper name AMD64

Re:AMD64

[drinkypoo]

If I'm addressing anyone but intel, I will. I don't think intel will listen to me if I rub it in their face :)

Re:AMD64

[Anarke_Incarnate]

AMD64 is the real deal. EM64T is a kludge that is mostly compatible with AMD64. AMD64 has better performance and better handling of >4GB RAM.

Re:AMD64

[Hal_Porter]

AMD64 has better performance and better handling of >4GB RAM.

In what way exactly?

Re:AMD64

[mczak]

In what way exactly?

I suspect the parent poster is refering to the IOMMU of the Athlon64/Opteron chips, which intels EMT64 chips lack. This might have consequences for some PCI (or other I/O) devices, the OS might need to use copy buffers if those devices want to do dma transfers to/from address space above 4GB.

Re:AMD64

[Anonymous Coward]

If you're "addressing" Intel on Slashdot, then I think calling it AMD64 is the least of your worries.

Re:AMD64

[Hal_Porter]

Yeah, I read about that, it's a cool hack, in the most positive sense of using something in a way that it was not designed for. Intel failed to spot this extra use for the AGP GART when they cloned the chip I suppose.

It's an interesting question though. AMD got to design the chip and the architecture at the same time. Intel had to retrofit AMD's 64 bit stuff to the P4. There are all kinds of reasons why this would be hard - the P4 had a dual speed ALU which needed to be widened to 64 bit for instance.

I wouldn't be surprised at all if the resulting chip had some performance issues, but I haven't seen any benchmarks as to comparative 64 bit performance though. I don't particular like P4's even for 32 bit stuff - it looks like they ultrapipelined the CPU to get higher clock frequencies in a way that reduces performance compared to a similar priced AMD part for instance, whereas AMD seems to have worried about real benchmarks like Spec and ( Doom3 :-) )

Did retrofitting AMD^H^H^HEMT64 make it worse?

Re:AMD64

[DigiShaman]

Granted, the AMD64 bit platform is the best stepping stone to migrate the industry into the 64bit world (the majority excluding Apple). But I though that EM64T was designed by Intel from the ground up. So how you can call it a kludge is beyond me. :-/

Re:AMD64

[Anarke_Incarnate]

intel did not design it from the ground up. They copied AMD's spec and tried to graft it onto a P4. They did not do a great job. Anybody who tells you that intel designed it from the ground up is just plain wrong

In related news...

[cocoa moe]

... Apple drops acid ;-)

Re:Grammar nazis unite

[garyday]

Tosser! Since when do modern operating systems come with a spell checker, i believe my OFFICE platform comes with a spell checker, i don't really recall a spell checker being a pre-req for a base O/S .. Therefore, :) is it too much to ask for people to engage brain before putting hand to keyboard. On a side note, I think we should go back to ZX81 architecture, i mean those guys really had something going there, that 1k ram pack was really ace, maybe i can hack it into my current PC, the extra ram would do wonders for me

IBM's High end

[MagnusDredd]

Article has some strange ideas about what constitutes a High-end server. I'd imagine a IBM P595 which supports up to 64 Processors would be high end... IBM seems to think so too. But then again what do they know about high end. I mean, they are only #2 in the High end server market (over $1,000,000 per server), and #1 in the mid-range server market (between $100,00 and $,1,00,000 per server).

Re:IBM's High end

[Anonymous Coward]

When he talks about high-end he means big iron, you know those mainframe thingies with OS/390 or z/OS or whatever it's called today. Capisc?

Re:IBM's High end

[ScrewMaster]

I assume that Riker Microsystems is Number One.

Re:IBM's High end

[SunFan]

Riker Microsystems is #1 in Eunics sales, not servers, specifically.

Not accurate

[Anonymous Coward]

TFA and moreso the summary and headline is not fully accurate according to what I've heard.

What IBM has decided not to do is support the Montecito IA64 chips. Apparently Intel initially approached IBM about licensing the X3 technology for an chipset to support Montecito, IBM agreed and shut down their own program to develop a chipset and redeployed the resources, Intel came back a few weeks later and said they had changed their mind, would IBM build an X3 chipset for Montecito but by this point they had also announced that the next post-Montecito Itanium chip would be plug compatible with Xeon. Hence the market opportunity for Montecito is about 18 months so it's not worth IBM's effort to build a chipset for only that time.

IBM has therefore decided to continue to sell the existing x455 servers through this year, skip Montecito and support Itanium again with X3 when it becomes plug compatible with Xeon. That means that for about a year they will have no server that will support Itanium.

Two years is a long time in this business so who knows if anyone outside of the HP/UX install base will care about Itanium by then but IBM does have a plan for continued IA64 support if current trends continue.

This is not good news for Itanium but it's also not a complete cancellation.

Re:Not accurate

[Anonymous Coward]

IBM has therefore decided to continue to sell the existing x455 servers through this year, skip Montecito and support Itanium again with X3 when it becomes plug compatible with Xeon. That means that for about a year they will have no server that will support Itanium.

Can you elaborate on what the timelines are for there to be a Xeon-compatible Itanium? And does that really mean it will be X3 compatible, or does it mean a future Xeon and future Itanium, neither of which are strictly speaking compatible with existing mobos/chipsets/whatever, but are indeed compatible with each other?

...and given Intel's track record and the industry's complete lack of urgency in taking up anything Itanic (other than HP, of course), could you give your level of confidence that this is really only a year away?

Re:Not accurate

[raxx7]

There will also be a Montecito using the same FSB as Madison, so they'll be compatible with existing chipsets. That's another reason of IBM to skip it, although the current FSB is lagging performance.

Re:Not accurate

[raxx7]

Second option: a future Xeon and a future Itanium will be using the same chipsets -- Common System Infrastructure.
Orginally, it was planned to 2007 but it might show up earlier.

Re:Not accurate

[Anonymous Coward]

a future Xeon and a future Itanium will be using the same chipsets -- Common System Infrastructure.

OK, so that's different from what the original post in this thread intimated, which was that a future Itanium might show up that works with X3. So that leaves X4, which is still 18 months out, for any IBM revival of Itanium ...which as others have pointed out, would seem to be contrary to their interests in the POWER architecure.

Orginally, it was planned to 2007 but it might show up earlier.

...because Intel has such a great track record in pulling in Itanium-related projects?

Re:Not accurate

[Anonymous Coward]

Correct, from what I have heard inside.

Re:spell

[Anonymous Coward]

Hack Jandy needs to stop trying to use words he has only heard spoken aloud

Expanding your written vocabulary with words you've heard spoken is a good thing. Pointing out mistakes can also be helpful. Telling people not to bother trying, however, is just retarded. Try growing up, Dun Malg. You might find you like it.

Re:Grammar nazis unite

[fdobbie]

Mac OS X comes with a built-in spell checker.

Power vs Itanium vs Xeon vs Opteron

[SpamMonkey]

Ok, so we all know the various CPU names and who makes them etc but do we actually know how they compare? Me and the team I work for have total ownership of 7 SAP Application servers and 1 database server, total ram in the DB server is 48GB and the App servers have been 4 and 12GB's each (normal compared to batch processing). They're all running on either IBM P630 to P670's. What does that mean? I have NO idea except that they are able to comfortable deal with 1200 active users at any given time.

now, if someone can tell me that Itanium will give us better performance for more we'll look into it, if it's Xeon then it's Xeon (pah but you get the idea). What I fail to see is why it's important what hardware is being used as long as it does the job it needs to do!

Thanks.

Re:Power vs Itanium vs Xeon vs Opteron

[ScrewMaster]

I think the winner goes to the processor with the coolest-sounding name. Of course, "Itanium" sounds like someone at Intel misspelled titanium, and "Xeon" sounds like the chip contains a mixture of xenon and neon gases. Maybe it does. At least "Pentium" vaguely suggested "fiveness" (as opposed to "fourness" in the 80486) but Intel's latest naming schemes could just as easily apply to automobiles or copy machines.

Re:Power vs Itanium vs Xeon vs Opteron

[progr]

In Italy the word "Tinanium" sound very weird but the word "Centrino" take the trophy cause a "Centrino" in italian is the name of a little silk handmade work. You can find it over a round table behind a flower pot. I think that in future amd and intel will be very attracted by biblic or greek name like: "Thanatos", "Strategos", "Chronos" and so on.

Re:Power vs Itanium vs Xeon vs Opteron

[Anonymous Coward]

TBH Itanium and Opteron are the best names for a CPU. Celeron also isn't such a bad name.

The most annoyingly childish names are Sempron and Cell.

Pentium is also a cool name but you get bored with it too quickly.

Then there are also those fuckedup names that sound like some special version of some type of a posion/gas. Yes, I mean Duron and Athlon.

Re:Power vs Itanium vs Xeon vs Opteron

[Anonymous Coward]

Here are some COOL names AMD and Intel could consider for their next product:

AMD Fasteron
AMD Speedron
AMD Numberone-on

Intel Suckium
Intel Notcheapium
Intel Expensium

and my favorite AMD Intelsmasher+

Re:Power vs Itanium vs Xeon vs Opteron

[Shadowlore]

What I fail to see is why it's important what hardware is being used as long as it does the job it needs to do!

It depends on how broadly you define "needs to do". If you can get the "job done" in terms of performance (i.e. concurrent users) yet get it for significantly less money on different hardware, and no software/UI changes for the end users, would you consider it? Especially if the other hardware had a longer life expectancy?

Life expectancy and cost are factors included in "gets the job done" IMO. Cost is more than just price as well. There is hardware support which includes parts and maintenance, as well as power and floor space consumption. Anyone renting space at a co-lo facility will attest to that, as will anyone running a singificant sized data center.

These are but some of the reasons why which hardware is being used is important.

Re:Power vs Itanium vs Xeon vs Opteron

[Anonymous Coward]

"Itanium" sounds like someone at Intel misspelled titanium

From my sources inside Intel's marketing department, here are the top ten candidate names for their next processor architecture:

10. Hodium
9. Ridium
8. Rypton
7. Candium
6. Ickel
5. Rancium
4. Ron
3. Odium
2. Assium
1. Old

Re:Power vs Itanium vs Xeon vs Opteron

[chthon]

Too late! I use them already at home!

Re:Ahem

[ScrewMaster]

I think he meant "purees Itanium in their next generation server lineup."

TPC-C

[Anonymous Coward]

What Intel did was to add cache, and then more cache on chip. This is similar to the P4 EE (Emergency Edition).

Just using 64 CPUs, the IBM Power5 gets 3 times the performance in TPC-C than the 64-CPU HP-Itanium.

With 16-CPUs, the IBM Power5 570 is around 80% the performance of the 64-CPU HP-Itanium Superdome.

Hmm, the Itanic performance is really bad if the workload doesn't fit in the cache. Also, in order to get good performance, the code can't have too much branches.

http://www.tpc.org/tpcc/results/tpcc_perf_result s. asp?resulttype=noncluster

Re:spell

[Dun Malg]

Hack Jandy needs to stop trying to use words he has only heard spoken aloud

Expanding your written vocabulary with words you've heard spoken is a good thing. Pointing out mistakes can also be helpful. Telling people not to bother trying, however, is just retarded. Try growing up, Dun Malg. You might find you like it.

Grow up? Here? Never! Slashdot is so full of assclowns and jerkoffs that it's the only place I feel comfortable being a Total Fuckwad. As for expanding your written vocabulary with overheard words, it's often wise to check the spelling first. In this day of spell checkers, there's no excuse not to (other than perhaps laziness or stupidity) and it's particularly foolish to submit articles for publication without checking. I generally don't care if comments contain proper spelling, for what it's worth.

Gee, reminds me...

[Anonymous Coward]

...of the IAPX-432 debacle. (And if you don't know what that is, google it. It was the Itanic LONG before the name Itanic was cool.)

Re:Gee, reminds me...

[Hal_Porter]

Yup, both IAPX-432 and IA64 had architects like a dog has fleas.

Re: Sig

[Anonymous Coward]

What's with the sig, it makes perfect sense to me?

Re:spell

[Anonymous Coward]

How is this not off-topic? No one cares about your petty spelling/vocabulary arguments.

Re:GRAMMAR NAZI!

[YU Nicks NE Way]

The line "Gonna take inappropriate comma usage down to zero" doesn't scan. It would work better as "Gonna take comma splicing, down to zero".

High-end servers?

[rdean400]

They already use POWER in their iSeries and pSeries servers, which are the highest-end single servers on the market.

As far as the decision goes...I think the Itanium wasn't a profitable platform for IBM in the first place, which made it easy to scapegoat marketshare. :-)

They could have been popular

[extra the woos]

If Intel released them at a low price and with desktop motherboards that were affordable. If an average geek could build the latest itanium system for $200 more than the latest athlon system, well, people would buy it because it's something different, it performs well, and because they want to mess with the architecture. It shuld have been marketed like the P-Pro. Too much for the desktop user, but if you want one you can afford it and you can build it/buy it!

Re:They could have been popular

[Anonymous Coward]

Yep. Transmeta made the same mistake.

Re:They could have been popular

[mabhatter654]

percisely, Intel wanted the midrange market of 5-10K servers so it could create it all to itself....rather than evolving the community as before. Itanium came out not compatible with ANYTHING...the 32bit support was "hacked" on only in the second version of the chip.

Intel's real problem was relying on Microsoft as the sole support for the chip. They pretty much blew off Linux as an also-ran until MS screwed them over with an overpriced and undersupported version... funny how that happens!

Re:They could have been popular

[Anonymous Coward]

percisely, Intel wanted the midrange market of 5-10K servers so it could create it all to itself....rather than evolving the community as before. Itanium came out not compatible with ANYTHING...the 32bit support was "hacked" on only in the second version of the chip.

I wrote code on early v1 Itaniums, and they implemented x86 compatibility in hardware. It was much slower than actual x86 CPUs, but so was native IA64 code (on v1). The problem was performance, not compatibility. Itanium2 was a lot faster, but not nearly fast enough to justify the price premium over AMD64.

Intel's real problem was relying on Microsoft as the sole support for the chip. They pretty much blew off Linux as an also-ran until MS screwed them over with an overpriced and undersupported version... funny how that happens!

Oh rubbish. The problem with acceptance of Itanium has always been high hardware prices and low price/performance compared to x86, and now AMD64. If Intel had produced CPUs and chipsets for desktop-class Itanium systems, and performance had matched the promises, everyone would be moving to Itanium instead of AMD64, and Microsoft would be selling a mass-market version of Windows for Itanium.

Speaking as an Itanium Architect

[Anonymous Coward]

I think that Intel buying the HP team means they are going to do one last Itanic, then just cut them off. Perhaps they just need to fulfill prior obligations before they kill it off? Working at Alpha was so much better than Itanic.

Re:Grammar nazis unite

[KontinMonet]

Grammar? It was spelling. In any case, is it too much to ask an editor to be able to spell correctly?

Consolidating on one hw-platform ?

[Anonymous Coward]

IBM uses Power in the I-series (old nams AS/400), P-series (old name RS/6000) and have just put some new small servers on the market intended primary for Linux called OpenPower... btw. high-end IBM Intel-servers (X-series) has an Power as hw-management unit (this is quite funny IMHO). I suspect that Z-series also is powered by a special version of Power... Itanic didn't offer IBM any special advantage...

Re:Consolidating on one hw-platform ?

[chez69]

actually, the zseries runs a CISC based processor that's been updated to 64 bit, and can still run programs from the 60s.

The high end kit for speed is the P series kit, not the xseries.

(I don't work for IBM, but my employer spends ass loads of money on IBM kit)

Re:Consolidating on one hw-platform ?

[Anonymous Coward]

While writing all the following, I thought about the PBS show on the Concorde a few weeks back -

http://www.pbs.org/wgbh/nova/concorde/

I thought it summed up quite well how having the technically superior answer, but one that once it was delivered answered the wrong problem (especially given that it was significantly delayed, way over budget, and competing with the unexpected introduction of a game-changing competitor / the 747) shows how a big technical bet can wind up being an expensive lesson.

For long-time IBM watchers, it's always about taking someone else's idea - "Windows
Everywhere" - and doing the CPU version; add in Linux to help solve the "which OS" question. Not original, but IBM has a history of succeeding at ideas others have thought up. After all, what was the basis for the PC in the first place?

"Now we're gonna go back, way back - way back - back into the past..." (Jimmy Castor, anyone?). We were once told that we would eventually see MS Windows dominate on every possible platform: embedded devices, cell phones, PDAs, "thin clients", Tablet PCs, desktops & laptops, small servers, mid-size servers, "Windows Mainframes" (with some, er, side branches like X-box and Windows Multimedia). "Windows uber alles!"

I do suppose deployment of Windows across this wide assortment of platforms actually happened, but today it seems clear that Windows won't dominate the future of computing to the same degree it did the last 15 or so years; on almost all the platforms listed, there are viable competitors or alternatives. And with the availability of Linux, Power, and power-derived platforms (Mac Gx, Sony PS3, MS Xbox2 and the other Cell-driven game engines, etc.) what do you expect from IBM?? There is more to life than Windows. In fact, Linux and OSS along with "Macintosh-The NexT Generation" (aka OSX) have sort of detoured this whole Windows dream of total domination over the past few years. (And I sure hope PalmSource's recent selection of Linux helps it stay alive as the alternative to PalmPC OS from MS, based on the same sort of strategy.)

As a consequence of Linux's popularity, the space for Power to prosper has gotten much larger than say back in 1998, when AMD was on the ropes, Itanium (then just plain old Merced) was supposed to do to the 64-bit RISC CPUs what the P6 (Pentium Pro) had done to 32-bit x86 and RISC ones, and Wintel (and SUN - don't forget those "we're the dot in .com" commercials) were going to render IBM to the same dead end as DEC, Prime, DG, as well as Hitachi, Amdahl, ICL, Fujitsu and the other plug-compatible Mainframe makers. (Don't kill me on accuracy here - becoming hard to even remember there were "plug-compatible" mainframes). Just think - if Itanium had lived up to the hype and Linux (and BSD, OSX, etc.) had not shown up, we would have a choice of Windows, Windows, and - oh yes, WINDOWS for all but legacy platforms (which would mean UNIX variants at this point).

Yes, this becomes a "virtuous circle" discussion - the combination of Linux on Power (LinPow?) serves as a viable counter to the Wintel (or mostly Windows) strategy espoused above, where Power availablity enables Linux, which further enables Power... Power-derived architectures now show up in high volumes for games (Cell), cars (Moto and embedded Power), Macs (PowerPC), pSeries (AIX boxen), and their "evil twin Skippy" siblings iSeries (OK, maybe AS/400s make sense to some; but I like the ability to now mix and match OS versions on the same box lots better). Many of these are the same platforms where we are seeing Linux being pushed, at least as an alternative to Windows, by IBM. Yes, there even have been rumors of Power for zSeries (while this might be a complete replacement like the VAX to Alpha transition for VMS, I think joint processors are more likely in the same frame under a single master OS for the next while, or perhaps highly-available Grid-based alternatives to Mainframes based - once again - on Power; hey it might be named "Cell" for

"IBM to Drop Itanium" is proof!!!

[Kymermosst]

Proof that the best way to accelerate an Itanium is at 9.8 m/s^2.

(That's 32 ft/s^2 in ye olde units.)

Re:Grammar nazis unite

[Anonymous Coward]

Not to be pedantic, but... persue? C'mon... don't most modern operating systems now come with spell checkers installed? Is it too much to ask that people posting use them before putting something online for thousands to read?

"C'mon" is a colloquialism, and should not be used in writing except between quotation marks. Technically, it isn't even a correct contraction because the 'e' has been removed without replacing it with a tick mark.

"nazis" should be capitalized.

don't most modern operating systems now come with spell checkers installed?

No. Usually it's the Word Processor or office suite that comes with it. Given the fact that this is Slashdot, he/she probably posted his/her comment with Firefox. The last time I looked (right now), Firefox had no way to spell check what is typed into an edit-box. (There may be an extension that does - I did not check. If there is, it does not come with Firefox.)

From dictionary.com

[Anonymous Coward]

pursue v.intr.

1. To follow in an effort to overtake or capture; chase.
2. To carry on; continue.

peruse

1. To read or examine, typically with great care.

Persue = no word

It's all about price

[Stonefish]

If intel want to succeed in this space they have to slash their prices on the chip.

It must be a much better buy (bang/buck) than the existing CPU families.

Promote early adoption in the geek/research space and as those folks move into industry applications will follow.

At one stage the organisation that I worked for was approached to utilized the hardware beta for scientific analysis and we were very interested however the cost of the equipment never fell to a level that would allow it to compete with the X8 arch.

Intel recoup your losses, promote the chip at a very low price point. Ignore you marketing/sales people trying pleading that a many different model. It is a short term strategy that is applicable when you have a monopoly.

Why do they build Xeons if theirs PS3s are better?

[Anonymous Coward]

I don't understand them!

I believe the Power PC Cell processor will be used

[MrJerryNormandinSir]

When you think about it, the Power PC Cell processor is much more advanced than the Itanium.
The porcessor architecture is very scalable and I
believe will not only be used in the playstation

Check ou this blurb I found on the net:
ell provides a breakthrough solution by adopting flexible parallel and distributed computing architecture consisting of independent, multi-core floating point processors for rich media processing. With the capability to support multiple operating systems, Cell can perform both PC/WS operating systems as well as real-time CE/Game operating systems at the same time. Scalability offered by Cell can be utilized for broader applications, from small digital CE systems within the home to other entertainment applications for rendering movies, and to the big science applications as supercomputers.

A team of engineers from IBM, Sony Group and Toshiba are collaborating on the design and implementation of Cell which is expected to deliver vast floating point capabilities, massive data bandwidth and scalable, supercomputer-like performance. The design work is taking place at a joint development lab the three companies established in Austin, Texas, after the project was announced in 2001.

IBM plans to begin pilot production of Cell microprocessors at its 300mm wafer fabrication facility in East Fishkill, NY during the first half of 2005. The first computing application IBM plans for Cell is the Cell processor-based workstation it is developing with SCEI.

I believe this processor will be used in IBM's new
design. It just makes sense.

I like ....

[zogger]

..."Nostradamos" as a name. The first psychic CPU that predicts what tasks you will be doing.

Unfortunately, all the apps and any kernel for it have to be programmed in quatrains...

Drops Itanium because of hardful JIT-Java-to-IA64.

[Anonymous Coward]

open4free © : Sun is devil for IBM.

Visionary

[SunFan]

Whoever said that the ISAs would condense down to only x86, PowerPC, and SPARC appears to have been correct. Alpha is gone, mostly. MIPS is gone in the desktop/server, mostly. Itanium kinda came and went, it appears. PA-RISC is still popular...but but HP wanted Itanium.

Re:Visionary

[Bert64]

Alpha is popular too, and still outselling itanium, even tho HP is jacking up the prices on Alpha hardware to try and discourage sales..

Re:Visionary

[SunFan]

"Alpha is popular too, and still outselling itanium..."

Irony, thy name is Hewlett Packard. PA-RISC is clearly out-selling Itanium, and if Alpha is too...LOL

Re:Visionary

[Anonymous Coward]

No, Itanium outsell Alpha, but PA-RISC still for some time outsell Itanium.

Re:Grammar nazis unite

[Anonymous Coward]

Given the fact that this is Slashdot, he/she probably posted his/her comment with Firefox. The last time I looked (right now), Firefox had no way to spell check what is typed into an edit-box.

And also, since this is Slashdot, the inevitable Macintosh user will add the comment "Safari will spell check what I'm typing in the edit-box"

Re: Sig

[Dun Malg]

What's with the sig, it makes perfect sense to me?

Sure, if you know what the guy saying it is talking about. Imagine it's Day 1 of SIGINT analyst school and
"The caveat SECRET-SPOKE is classified CONFIDENTIAL-HVCCO-- which is itself, UNCLASSIFIED"
was the second sentence out of the instructor's mouth, the first being "All paperwork must be properly labelled with the proper classification and caveat." The reaction of the entire class was universally "huh?"

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:From the summary: hogwash

[kurt555gs]

1st. unlike machinery or chemicals software costs almost nothing to reproduce.

If you want to license you software under a comercial license and it is really really good, and you price is reasonable, then people will BUY it, just like anything else in this whole world.

If you want to gouge people, then you can hardly blame them for using it and not paying you.

Remember, because it costs almost nothing to make, softwares REAL value diminishes with every copy made, paid for or not.

rant off

New stickers

[Proc6]

I want my next server to be IBM and the sticker on it to say "Powered by Power".

Re:New stickers

[Hymer]

...look then at the IBM OpenServer 710 or 720...

If it doesn't have the sticker I'll make one for you...

Itanium is dying

[Megane]

Re:Itanium is dying

[Anonymous Coward]

Itanium is so dead that not even NetBSD runs on it!

speling (yes, dodo, I know that's misspelled)

[prairiedock]

Can't the editors at least fix spelling mistakes in the article headers? Regardless of whether the mistakes are due to ignorance (most on /. are) or are just typos, isn't one of the functions of editors to edit? Having "persue" in the headline and in a million links out there in webland is just an embarrassment to every /. reader who might hope to persuade others to take viewpoints expressed on /. seriously....

Why Itanium Failed

[bani]

what happened with itanium is intel made a number of huge gambles on technology.

in order for itanium to be successful, every single one of them had to pan out.

what happened is virtually none of them panned out.

intel blew their load on a high risk gamble, and lost. they still can't quite come to grips with the fact and are still sinking billions of dollars into a doomed architecture -- despite the fact that just about every original itanium partner has already given up on it (err.. "jumped ship", hence the itanic joke)

intel has been beating on itanium for nearly a decade and it still hasn't lived up to a single design goal.

and before the itanium defenders go "no, itanium was only ever intended for rackmount servers", that is 100% contrary to intel's own marketing literature which states that "workstation" is one of the target markets of the itanium.

Re:Why Itanium Failed

[Alioth]

But was Itanic really a total failure?

No - not really. The mere existence of Itanic caused MIPS, PA-Risc and Alpha to simply roll over and fold before the competition even got underway. Intel has just essentially destroyed three competing architectures.

So now that MIPS, PA-RISC and Alpha are dead, those vendors were all supposed to move to Xeon.

But poor old Intel didn't see Opteron coming.

Re:Why Itanium Failed

[jhagman]

It's very sad that the Merced -project sank two very nice processor architechtures, namely Alpha and MIPS. I have no knowledge of pa-risc or itanium, which are also becoming dead because of Merced.

Re:Why Itanium Failed

[bani]

er, mips is hardly dead. cisco would be somewhat shocked to hear the chips they use in most of their routers dont actually exist.

A Geekish Note

[Jouster]

[BLOCKQUOTE]With IBM's investments in Power...[/BLOCKQUOTE]
Aw, heck, so now my Black Lotuses and Moxes are even more expensive? I mean, diversifying a portfoilio's good and all, but collectible cards games might be stretching it a bit far....

Jouster

Re:Grammar nazis unite

[Hack Jandy]

Woe is me. Firefox 1.0.1 it was :(

Re:GRAMMAR NAZI!

[woah]

Louie: They's throwing robots!
Linguo: They are throwing robots.
Legs: He's disrespecting us. Shuttupa you face!
Linguo: Shut up your face!
Legs: Wassamatta you?
Louie: You aint so big.
Legs: Me and him are gonna whack you in the Labonza!
Linguo: Bad... grammar... overload. Error! Error!

Re:Grammar nazis unite

[amerinese]

C'mon is obviously a WRITTEN colloquialism (that reflects oral usage but it's not one to one), given that it does not have the 'e'. You yourself recognize that it is normal colloquial usage, being that no one writes "C'meon".

Another religious 'Mac' twist

[Blitzenn]

Why would they go with an chip like the Power when it also has no basis for acceptance as a high end server platform. That would seem like just as stupid a move, if not worse, than the Itanium. Granted it's their own technology, but if that's the precurser, then the Power chip is already out of the question. They have much better stuff behind their doors than that old thing. Have you read anything about cell processing lately? If they want to go with a chip that already has a base of millions of shipped units, then perhaps they should go AMD Opteron. That would be a better choice for battling the acceptance problem they perceive (And even Microsoft is specifically wrapping their 64 bit OS around it). I don't think you are going to see IBM try to drive the market with their Power chip, that's EXACTLY what happened to Intel and the Itanium. Besides that's a huge move from Cisc to Risc that 99% of their exisiting Intel based customers are simply not going to do. I have to believe IBM is smarter than that.

What's left

[Stu Charlton]

If Itanium dies, what do we have left for a 64-bit chip? POWER5 and Opteron? SPARC still really lags, and while they're going down the crazy multi-core/hyper-threaded route, I'm not sure it will make up for the horsepower gap.

HP is the only major manufacturer left of the Itanium line - their Superdome Integrity SMP boxes are wonderfully fast machines. After using both a Superdome and an IBM pSeries SMP(though, it was POWER4 based), I have to say I'm supremely impressed with both. But now, POWER5 is out and Itanium is still where it was last year. And POWER5 absolutely smokes. I don't think data centres are too keen on jumping onto the Itanic.... HP needs an ace-in-the-hole to get out of this one.

Re:What's left

[Anonymous Coward]

Although Montecito may not exceed Power5 in performance, I think it will come very close.

Re:Grammar nazis unite

[Anonymous Coward]

Hi! Inevitable Macintosh user here!

It's not Safari, it's the modern operating system. Every text box in a native application gets spell check 'for free'.

Thanks for listening! Let's all have happy!

Buzzwords

[tommyth]

"Market Acceptance Issues" (n): It sucks.

Stop your lies.

[DigiShaman]

This is bullshit. Just how did Intel copy AMD? In fact, Itanium was designed from the ground up and that is a bloody FACT! Hell, Itanium was being worked on far longer then when AMD had plans for a 64 bit chip. I can site proof of this, can you back up your claim. How you got modded informative is beyond me. And yes, I will flame anyone who post such BS without the facts to back it up.

http://www.geek.com/procspec/features/itanium

Re:Stop your lies.

[Anarke_Incarnate]

Hey dipshit, someone should teach you the difference between EPIC (Intel's IA64 instruction set) and EM64T, which is the kludge they strapped onto an overheated overpriced Pentium 4 (Called a Xeon). That is a bloody fact. Nobody was talking about Itanium. In fact, next to nobody is BUYING Itanium. This is why intel had to eat some of its own lunch and make a 64bit Xeon line. Go flame yourself, you crybaby.

Re:Stop your lies.

[Anonymous Coward]

How cool you are... Take a walk and calm down.

Re:Stop your lies.

[Anarke_Incarnate]

Hey, I did not hide behind an AC post. When someone goes out of their way to make themselves look like an idiot, frothing at the mouth about how they are right and are clearly wrong, they deserve a smack to get them back in order.

Re:I believe the Power PC Cell processor will be u

[Anonymous Coward]

I am sorry, but you are somewhat clueless here.

Re:From the summary: hogwash

[mcrbids]

Remember, because it costs almost nothing to make

Oops. Typo? Freudian slip?

Software has a very high cost to make. It requires highly skilled laborers, and lots of time to make. Making high-quality software is difficult, and by the nature of things it's difficult for a programmer to make software that's easy to use.

Notice I never argued against open licenses. Also, I never argued that software costs little to reproduce. It costs about $0.50 to make a CD-R, if you don't mind a vanilla, paper case.

I argued (and still argue) that software piracy reduces the actual saleability of soft wares. It essentially makes a vendor compete against himself, and therefore should legitimitely be called "theft", because doing so takes value away from the goods that highly qualified personnel spent lots of time to create.

If you want to gouge people, then you can hardly blame them for using it and not paying you.

Don't like the price? Don't buy it. Think it's wrong? Write your own open-licensed product and give it away. There's certainly plenty of people who are.

The ONLY time your arguments might have weight is in the case of a monopoly, and in these United States, we have protections for that, too.

Why do you feel the need to justify theft? Your arguments cast a pall from those of us who are dedicated to using and supporting truly open, free licenses on software!