Slashdot Mirror
Why Doesn't the Itanium Get the Respect It's Due?
happycorp wonders: "As in recent years the Itanium does well, easily beating x86 processors even at its low clockspeed (1.4Ghz). The supercomputer people are serious about benchmarking (no easily tricked microbenchmarks or reliance on closed-source
commercial apps), so the discrepancy between the performance and perception of this chip is serious.
With a single-CPU Itanium2 system at
around $2000 their price is already reasonable, and the price would come down
(and software would be ported) if the Itanium ever became a mass market chip. Having an affordable chip one step above a Xeon or Opteron in floating-point performance would not be such a bad thing for gaming enthusiasts (or 3D artists). So, the recent
article
on the
Top 500 supercomputers list brings up a question I've been meaning to ask:
Why do we see so many disparaging opinions of the Itanium processor (all those 'Itanic' jokes, etc.)?"
"It seems computing enthusiasts' sentiment is set against this processor, and its likely that it's going to be abandoned sooner or later. We'll be paying for x86 compatibility indefinitely (recall the Xeon has roughly
three times the number of transistors of the ppc970 for example; but we hardly get three times the performance).
These are a couple scores from the top 20, with the total gigaflops divided by the number of processors to obtain a per-processor speed:
rank processor ghz (gflops / #procs) speed #5 ppc970 2.2 (27910 / 4800) 5.81 #7 itanium2 1.4 (19940 / 4096) 4.86 #10 opteron 2.0 (15250 / 5000) 3.05 #20 xeon 3.06 (9819 / 2500) 3.92
Given this, consider what a 2 or 3 Ghz Itanium could do.
(fine print: I am not affiliated with the Itanium or the top500 list in any way)."
These are a couple scores from the top 20, with the total gigaflops divided by the number of processors to obtain a per-processor speed:
rank processor ghz (gflops / #procs) speed #5 ppc970 2.2 (27910 / 4800) 5.81 #7 itanium2 1.4 (19940 / 4096) 4.86 #10 opteron 2.0 (15250 / 5000) 3.05 #20 xeon 3.06 (9819 / 2500) 3.92
Given this, consider what a 2 or 3 Ghz Itanium could do.
(fine print: I am not affiliated with the Itanium or the top500 list in any way)."
They should have called it the "Dangerfield".
Who is going to buy a 2,000$+ graphics card to play games?
"It felt almost as good as stealing cars from grandma." -- Margaret Thatcher, probably.
The chipmaker has released two new Itaniums for two-processor servers as part of its effort to eliminate price premiums on the chip.
Intel announced on Monday two new Itanium processors for two-processor servers, another step in the company's efforts to eliminate price as a barrier to Itanium acceptance.
The 1.4GHz Itanium 2 with 3MB of cache is designed for servers in clusters. The new chip will provide about 25 percent more performance and cost much less than the initial Itanium optimised for clusters, which came out last year, said Jason Waxman, director of multiprocessor platform marketing at Intel.
The second new chip, a 1.6GHz Itanium 2 with 3MB of cache, is optimised for higher performance in general-use two-processor servers, he said.
Waxman reiterated that Intel is working on several technologies that will eliminate any price premium on Itanium by 2007 and thereby allow its performance advantages to, hopefully, blossom.
"The price/performance balance will be heavily in favour of Itanium," Waxman said.
With the focus on price, the Itanium melodrama is once again reaching a turning point. After several years of delays, the chip family debuted in 2001 to poor reviews and negligible customer acceptance. A second version of the chip that appeared in 2002 dramatically improved performance but failed to spark the market.
Itanium finally began to gain acceptance in 2003 with Madison, a new version of Itanium 2 that substantially improved performance again and lowered the cost. Intel shipped about 100,000 Itaniums in 2003, compared with only around a few thousand for the first two years. Itanium volume is expected to double this year, chief executive Craig Barrett said in February.
But in 2004, Intel announced that it would come out with a version of its Xeon chip that runs both 32- and 64-bit code. Xeon and Pentium chips typically run 32-bit code. Itanium runs 64-bit code, which, among other advantages, lets a computer maker pack far more memory into a computer.
Itanium, however, requires completely different software to work well, a factor that has hindered adoption. Part of the appeal of the Opteron chip is that it can handle larger memory loads in 64-bit mode on essentially the same software base.
Lowering the cost of Itanium servers won't eliminate the software issue, but it will begin to create an environment in which greater acceptance could occur, which in turn could cause software developers to gravitate to Itanium. Analysts and PC makers have viewed this theory with various doses of scepticism, but the range of opinion is generally substantially less negative than it was 18 months ago.
Price drops have already had some effect. In 2002, a two-processor Itanium server cost about $18,000 (£9,859). With the new chips, a similarly configured system can sell for less than $8,000, while basic one-processor Itanium servers will go for just more than $2,000.
Some of these price cuts have come as a result of Moore's Law, which predicts that the number of transistors on a chip will double every 18 months. But Intel has also expanded its product line to better suit the economic realities of two-processor servers. The company also designs and partly manufacturers many of the Itanium servers on the market, which cuts independent engineering costs.
To lower the price further, Intel will begin to create products and add features to Itanium so that Itanium servers can be made out of many of the same components as Xeon servers. In 2005 and 2006, Itanium servers will be able to use the same memory or other components of Xeon servers, Waxman said.
In 2005, Intel will also come out with two different chipsets for Montecito, the next major version of the chip. One chipset will wring maximum performance out of the chip, Waxman said, while the other will allow server makers to insert Montecito into their Madison-based servers, thereby cutting down independent design efforts.
By 2007, Intel will
Because Intel tried to force everyone to jump on the 64bit bandwagon at once, while windows didn't even support it yet, without backwork compatibility to existing 32bit software. It's a good design, just doesn't (didn't ?) fit well with the mass market at the time of the release.
the dead ones were always much better :)
Wondering why i am doing so strange posts? I am trying to get a "+5,Flamebait" or "-1,Insightful" rating.
I have certainly noticed a general move away from Intel in the past few years. I think they may have had a run of bad press and serious competition from other manufacturers lately.
They just aren't the juggernaut they used to be. There was a time when they built it and people came. I presume choice is what's keeping the sales down.
Why do we see so many disparaging opinions of the Itanium processor (all those 'Itanic' jokes, etc.)?
Because people repeat what they hear. Many people here only know what has been said on Slashdot about the Itanium. They've never used one. MrDicker64 said it was crap, so it must be!
I had to study the chip in one of my EE class. The technology in it is really really impressive. I love the memory architecture provisions!
What we need for gaming is a Physics chip more than anything else right now, not some general purpose crap.
Its the next big thing, not a bigger horse, rather a few horses and some that specalize in Phyics, Multiple cores, and Vector processing for graphics and DSP for Audio.
Multi core GFX chips, Multi core CPUs, and Physics processing. Offload that general processor.
Its easier to harness the power of many horses than grow one 100 times as powerful.
I think the big problem is that it cannot run x86 software very quickly. Most software that people want to run in the mass market is precompiled, binary x86 software. That stuff just does not run well on the Itanic. That, combined with the fact that the mass market still doesn't really benefit from a 64-bit address space means that the Itanium was a more expensive, slower processor. It's no wonder that it didn't sell.
Early versions also had problems with heat. Where I work we have some Itanic workstations and in the winter, if we were chilly, we literally turned them on to help warm up our offices.
One, it gets no respect because nobody uses it. Where is the kudos for the transputer? Why does nobody love the Apple ///?
Second, yes it beats the x86 into the ground. I'm not surprised. Now show me how it compares against a real CPU. We've already seen that the Itanium is competing in a different space (supercomputers), so show me how it compares with the MIPS that SGI have ditched in its favour. I wouldn't be surprised if an n GHz MIPS stuffs an n GHz Itanic into the floor.
Probably because when it mattered a single CPU Itanic was more like $12,000 and not $2,000. After fucking up all their marketing and delivering strategies no one wants one anymore.
I'm Rick James with mod points biatch!
Hundreds and hundreds of products have been killed or permanently crippled because their first versions were terrible. Itanium is the same thing. With the public perception of the Itanium still the same as it was for the first (pathetic) iteration of it, how are you going to convince your manager to spend the money to get it? Benchmarks only go so far.
-Daniel
a few reasons.
Itanium was a huge project jointly developed with many partners, most of the significant ones have long since abandoned the effort.
It was supposed to be the future of Intel - shipping units on the order of the pentium line. A redesign from scratch of how processors "should" be designed.
It's taken far longer, cost far more, and yielded far less than promised.
That's basically it.
Also, I'd be willing to bet Intel staked a bigger part of its decision on the availablity of platform independent binaries making serious inroads, which hasn't really materialized. Platform independence of the major OSS and commerical apps is obtained through porting and source-level compatability.
http://techworthy.com/PCUpgrade/SeptOct2004/64-Bit -Gaming.htm
Because for Itanium compatibility they'd have to port everything over to the Itanium proprietary instruction set. You can see how eager they've been to do that for Macs, so guess how likely they are to port it for Itanium.
$8.95/mo web hosting
32-bit processors are shipping with 64-bit extension which allow them to break the 4GB memory limit.
This was main reason people wanted to see 64 bit chips, and since it has been resolved cheeply, there isn't nearly the market for it.
Inertia, would be my answer to this question: Inertia of the technological kind keeps x86 on the desktop, even with the 64 bit extensions.
Inertia keeps Microsoft on the desktop, even though it being low hanging fruit for crackers.
Inertia can be a good thing... in this case, it's a bummer. I can safely say that my next game rig will be A64 powered, simply because of... inertia.
EveryDNS. Use it. It works.
AC's need not reply
but my understanding, from the rumor mill, says that the Itanium was too little, too late and was partially aborted in an effort to get it out of the lab. It was a joint HP/Intel effort that was supposed to be the "next big thing" in processors, but dragged on so long in the lab (more than 10 years) that, by the time it was released, contemporary competitors already had nearly comparable horsepower and an established mindshare.
Because it's cool to hate Intel.
I love Intel. I wouldn't use another chip because I've had bad experiences with them, and never a single (notable) one with Intel. But for some reason, people love to bash Intel, simply because it's the big behemoth.
Less costly doesn't necessarily imply better - I'm an avid hater of Microsoft, definitely, but it's not solely because they're huge.
AccountKiller
I may be entirely wrong, but I believe the dislike for the Itanium stems from the fact that you can't compile any decently optimized code for it. Apparently, even Intel can't create a good compiler/linker and toolkit for creating machine code that makes good use of EPIC. Even though the processor itself is more efficient and faster, the same thing compiled to machine code running side by side with an Opteron or any other x86-64 chip will see the x86 win. If somebody could come up with a decent compiler/linker that provided full EPIC optimizations, they would be bangin, but they don't have it so we don't use it.
Wouldn't surprise me to see this processor showing up in powermacs in some time.
The people who work on scientific applications take performance seriously. They put a lot of effort into optimization. The itanium architecture is hard to optimize for, and the compilers just aren't there yet for the general case. So you wind up with a disparity between the performance in scientific applications and general purpose applications.
Other reasons itanium can't compete:
1) Compare the performance of itanium with xeon/opteron in running native x86 code.
2) Compare the costs of building real end user systems.
3) Compare the availability of windows xp drivers.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
Wonder how much of their performance comes from their R&D investment in compilers
One, market penetration. Windows *kind of* works on Itaniums. Code has to be compiled specifically for the platform - they're not very good at x86 code through WoW.
The BIOS replacement they use is not functional. It's very difficult to set up disks for use, and if you lose the disk that the BIOS data is kept on, you're screwed. As far as I know, there is no way to make that fault-tolerant short of manually storing the contents of that partition on another drive.
Support for the Itaniums has been terrible. The HP systems are riddled with hardware problems, and their support personnel (at the enterprise level) have no idea how to comprehend that they don't operate quite like any other workstation.
- It was aimed at a saturated market that wasn't about to make a big-time investment in changing architecture.
- The slow momentum gave it a bad name.
Of the two, I believe the latter is the more damaging.People don't want a processor whose main purpose in life was to artificially refresh Intels control on much of the Intellectual Property associated with the processors. AMD is getting too close, so they change everything and hope to charge royalties.
Digital is, by definition, imperfect. Analog is the way to go.
I google'd "Itanium Problems" and found several reasons as to why the Itanium doesn't gain much respect. Rather than post tons of links and issues, I'd advise you to click your new google toolbar and search away :D
While, the IA64 has always had great floating point performance, there's an awful lot of us out here that don't need fast FPUs -- e.g. code development, database, web serving, network i/o etc. Sure, IA64 is a winner for the teraflop oriented supercomputing community, but for the rest of us, integer performance matters more. And for price/performance, x86 and x86_64 beat ia64.
Dvorak speculated in 2003 that Apple would indeed switch to Intel and that they would use the Itanium line rather than x86. That does make sense, because it would mean a high quality Intel processor without much risk of homebrew Macs.
to compile for Itanium. Speaking as a compiler researcher, Itanium is great for generating research papers because there are all sorts of things that you can do from a compiler perspective. The problem is, outside a research environment, someone has to implement a lot of the ideas in an Itanium compiler to make it useful. Unfortunately, most of the stuff in the Itanium research papers isn't easy to implement and most of what gets put into commercial compilers are the easily implementable ideas.
Why anything doesn't get the respect that it is due. It is because people don't want to give it respect. The Unix People go Well Sun Ultra Sparc (Or any other of the 64 bit Unix platforms) has be 64 bit for many years before the Itanium. The Apple crowd went well the Power PC is now 64 bit (although this is changing, and may possibly give Itanium some respect). The windows users are afraid of Itanium because it may break a lot of compatibility in their legacy apps. The Linux users are afraid of a complete Intel Dominance and put their development efforts to AMD 64bit chips. It is a state where you see the old king dieing and this is your only opportunity to get a change in government before the kings son gets in power. Why doesn't FreeBSD get the respect it deserves, or why doesn't Python get the respect it deservers. The winner is not always the best or even close to the best, the winner is often the one that people feel good about.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
With a single-CPU Itanium2 system at around $2000 their price is already reasonable, and the price would come down (and software would be ported) if the Itanium ever became a mass market chip. Well, it's sort of like Linux. The only way it people will use it is if it's already a mass market chip, but it will never become a mass market chip if people don't use it. Sort of an infinite loop.
Software is like sex. It's better when it's free. -Linus Torvalds
May as well as why Linux/Mac/*BSD/etc. doesn't get the "respect it deserves." There is no real answer.
My personal thought is that price:performance was not in line with other choices available to the end consumer.
Do really dense people warp space more than others?
Itanium is broken by design. Its asm syntax etc is different from amd64 on *purpose* which was to create to incompatible 64bit system where Intel would keep AMD out of the market. Oh well, they failed badly.
Never learn by your mistakes, if you do you may never dare to try again
The first Itanium was too late, too slow, and too costly. This led to a bias being developed against it based purely on misinformation. Don't forget to factor in the typical hypocritical resentment of Intel...
... wait for it ... poor legacy compatibility. Legacy support is certainly a valid and very important issue but why must people constantly hammer away at this when considering the Itanium at face value: A very capable high-performance processor which would be ideal for UNIX workstations, servers, and supercomputing applications.
People are always complaining about X86 for its legacy cruft yet when Intel designs a bold new architecture based on an extension of the VLIW approach, they get slammed for
The biggest drawback of Itanium is the fact that it's a proprietary architecture. But then again, so were Alpha and PA-RISC. POWER and SPARC might not technically be so but if you want workstation-class POWER CPUs, you've got no choice but IBM (I don't see anyone else jumping on the PowerPC bandwagon and producing competing compatible chips) -- with SPARC there's just Fujitsu and Sun's inferior offerings.
But looking at the architecture for what it is, people should be excited about Itanium. It's new, it's fresh, it's fast, and it's interesting. Certainly far more interesting (and challenging) than RISC with a lot of unexploited potential.
I guess because (for some moronic reason) AMD are "good guys" and Intel are "bad guys" we just have to get all giggly and rub their noses in it. Shit happens. Sometimes you win, sometimes you lose. Some products take off, some don't. Itanium looks like a good architecture for transaction processing, at least on paper. Turns out the market was more interested in backwards compatibility.
My systems professor told us that they chose to create a very complicated assembly language, that while may be efficient, makes programming un-nesceissarly difficult. If people don't want to program on your platform, you have a problem.
where AMD rulz, and Intel is teh sux!
Linking to an article from over 1 year ago (April 19, 2004) while discussing pricing doesn't make me think the submitter went through any trouble to look up more current information on the Itanium series.
A quick trip even to Google News
While I understand they were trying to make a point about entry level pricing, lets at least have some current information to work with.
The Itanium was designed to change the way processors worked. Most processors today are some sort of dymically scheduled behemoth that are capable of detecting instruction collisions on the fly, and reordering instructions for optimal parallelism and thus performance in the light of those collisions. Itanium takes a completely different approach. It is an extremely wide processor that has absolutely no collision detection or reordering. All of the work in this respect is placed on the compiler's shoulders. In theory, a good compiler could make this chip very, very fast, and in reality, as you see, this can be the case. So why did it fail? Intel hyped the hell out of this processor, and then missed their release date by a full two years. That is microprocessor suicide in the land of Moore's law. So, when Intel delivered a chip too late that failed to perform the way they marketed it to, the chip died. In recent years, Itanium has really come around, but it's hard to escape your past in this industry.
Other relevant problems for adoption are tied to this need for a good compiler. Making a compiler as smart as it needs to be for Itanium to live up to its potential is not cheap, and Intel is not known for just giving away such technology. I'm sure the fees to license Intel's compiler are nontrivial, and that does not encourage development. Realistically, Itanium will never become a desktop chip just because of the massive adoption effort that would go into such a switch.
One thing to note, however, is that other chips aren't that far away. You suggest that a 2ghz or 3ghz Itanium would be incredibly fast, and I agree, but I seriously doubt Intel can ramp it that fast. Also, the Opteron specs you show are for 2.0ghz, and I believe Opteron is up around 2.6 or 2.8 ghz nowadays.
Ultimately, Itanium is a great design, but wrapped in a poorly executed initial implementation. It does teach a good lesson that compilers can really help improve chip performance, and down the road, architectures that take this into account may reign supreme. But I wouldn't look to Itanium to do any more than instruct us for the future. She is not a desktop chip.
I'd rather be cycling.
Peter.
Why do we see so many disparaging opinions of the Itanium processor (all those 'Itanic' jokes, etc.)?
Uhh, maybe because of the really stupid sounding name? It always sounds to me like something's been left off the name (ie the "Ti.." part). But of course they couldn't trademark "Titanium".
Naming is important - IMHO, a big reason why many open source programs do not get used in corporate environments is due to odd/wierd names. Many otherwise worthy programs are never considered or evaluated, due to "silly" names, I believe.
So if the name's a joke...maybe the processor is seen as a joke, too...I remember it was seen as the "next big thing" and there was lots of excitement & buzz about it, before it got the silly name...then it sort've dropped off the radar...
I'm hazy on what you mean by "platform independent binaries". Are you talking about
-Peter
Microsoft apps are nonexistent, and open-source apps tend to have crappy performance due to the fact that IA-64 depends overwhelmingly on compiler optimization. Developers can use Intel's compiler, but it requires work to use with most Linux systems (the only other platform that supports IA-64 besides MS, AFAIK).
Net result: no applications => no uptake, QED.
Egg, chicken, all that.
Lacking <sarcasm> tags,
You have floating-point listed there, which is great for science I'm sure, but where are the integer numbers?
I have seen the future, and it is inconvenient.
I asked the same question at RealWorldTech.Com about 6 months ago and got a fair number of replies - several of which I thought were fairly insightful.
i on=detail&PostNum=2978&Thread=1&entryID=43600&room ID=11
http://www.realworldtech.com/forums/index.cfm?act
You can see the full thread at the bottom of the linked page.
Is that some new Itanic chip?
I'd say the Itanium is getting a bit of respect in the highly available/high performance arena, with the HP/Tandem NonStop Platform moving to Itanium
The Itanium is the hardware equivalent of this Joel on Software parable, and a close relative to IBM's biggest blunder.
AMD64/EM64T provides a solid upgrade path from x86, whereas the horrible x86 perfomance of the Itanium (first in hardware, then in software) makes it a clean break. In the real world, where we depend on cheap, commodity x86 server hardware and applications, a clean break to a more expensive, incompatible, unproven platform was just not attractive to customers.
And don't discount the AMD factor. Even Intel-only shops benefit from the price and performance pressure any measure of competition in the x86 market provides. Moving to an Intel-only platform would remove the last barrier to permanent hardware lock-in.
The itanium is an amazing architecture with so many performance boosting upgrades that it would have blown everything out of the water.
If it came out on time.
It was so late that by the time it came out it was still better than existing processors, but not by a large enough margin to justify its cost.
As the clock speed goes up, and as the other processors find their limitations and drop out of the race, the Itanium will look better and better. There is, however, a large investment in time and software that must be made before it becomes truly useful. It is unlikely that MS is going to support more than one architecture simultaneously for the desktop or server as it tried to do for x86/alpha.
The big marketing push and the number of companies signing on to the good ship itanic coupled with the constant pushback of the release date caused Intel to lost a lot of the press attention they should have received when it did come out.
It'll be interesting to see what happens over time, especially as Intel wants it to be a server chip.
Of course, this could all be a big leadup to the announcement that Apple is going with the Itanium.
-Adam
Why do we see so many disparaging opinions of the Itanium processor (all those 'Itanic' jokes, etc.)?
The jokes are not about Intel's processor so much as they are about Intel the company. Intel nearly bet the farm on the Itanium and for a while, it looked like that bet might sink the company. Too many folks have too much invested in x86 software and the Itanic had a bad rep for running those x86 apps compared with AMD's 64 bit x86 style CPUs. A new processor family can't expect to succeed based only on its speed. You have to consider the (warning, buzzword) ecosystem based around that processor. Itanium has the failed to become a mass market CPU for the same reason that sparc, mips, alpha, and others have. PPC seems to have done a better job by grabbing a large chunk of the game console market to complement or even overshadow its use in servers.
FreeSpeech.org
The SGI Altix at #3 is Itanium 2 also.
When Itanium started, Intel was absolutely nowhere in 64 bit and high-end computing. Thanks to Itanium, over half Intel's competitors simply walked away from the market with little more than a few press releases from Intel.
Consider that at the time, you had Alpha (Dec), PA-RISC (HP), MIPS (SGI), and Sparc as leading 64-bit computing platforms.
HP in it's infinite wisdom was suckered the worst - giving up their own leadership position just to be strung along for many years in Intel's PR bluff. However Wall Street loved the "ooh, intel's story's so aWsUM that even HP is giving up" that SGI spun off and MIPS gave up on the high-end space; and Dec->Compaq->HP undervalued Alpha and it went away.
This has to be the most successful come-from-zero-to-wipe-out-half-the-market story in the history of computing. How can it be considered a failure.
And I am NOT talking about performance
or price/performance.
I am not even talking about adoption rates.
Look at what Merced (the real intentionally forgotten name of Itanium) was supposed to be:
1. A complete replacement for x86 processors
2. A complete replacement processor for HP, MIPS, Compaq/Digital (even before HP took them over), systems at the high and low end.
It was to be a FORCED technological direction that rubbed a great number of people, including myself, in many different tech-influential places in the industry to HATE the new architecture.
So when the prices stayed up, the performance delivered nothing special, and AMD came out with a brilliant alternative...
No one was there to keep the faith.
Just when Intel was to make it into the mainstream with it, its remaining leg was cut out from under it (picture the knight thrashing about with his limbs cut off one at a time, still trying to fight in Monty Python and the Holy Grail).
It was doomed from the beginning.
Itanium is like the "Edsel" of cpu chips. It's too non-mainstream and while the geeks may ooo and ahh over its esoteric technical details, nobody really cares to actually buy one.
I've seen some nice tricks in the architecture of the Itanium. They can really improve performance on most code. One of the tricks is register-rotation, combined with rotation of the computation-flags. But I think the actual chip-level-design they used is a lot slower than the Pentium series. Using these tricks the Itanium should outperform 5x and not just 2x. So probably the cause of slow performance can be found in the design of the chips. I personally think that the designers were hindered by the cooperation with HP (?), because of patent-reasons, bad communication, office-policies etc..
I was associated with porting a Java appserver onto Itanium a while ago (3+ years). Intel invested in this effort (free servers, access to hardware/compiler experts etc).
The problem was that, the JVM on this was very slow. It lacked JIT support and just crawled. Practically, speaking we would not have sold a single license on this platform. So, we decided to move away from being an early adopter to a market follower.
If not for Intel's premature marketing effort in engaging partners this early, we would have waited until a more stable environment was available, resulting in successful product shipping.
I talked to several presenters at an Itanium conference when I went to demo our product, I heard a similar story.
Well the reasons may not always be technical for such debacles.
First check this graphics: http://blogs.sun.com/roller/resources/JeffV/itaniu m_rev_ext.gif
Itanium processor was pushed so badly in the past, there are 9 IDC estimates for the processor reenues. All those 9 estimates were terribly wrong. how come IDC makes 9 consecutive wrong estimates for a product? Itanium deserves al kind of jokes and bad remarks IMHO.
- it is still an expensive product. Maybe CPU's are relatively cheap now but systems designed for it are at least double the price of an equally performing Opteron systems. Motherboards, chipsets are expensive.
- PErformance numbers you presented from super computers does not reflect realities. For super computers performance metrics are depends on so many parameters (inter connection, application type etc). Nobody can claim itanium is the best bang for buck.
- it is a power sucker.
- Still no good native software support, no good OS support, people are reluctant to use Virtual machines either (Java etc.)
according to geek.com "Itanium 2 processors currently consume 130 watts of power during operation, double that of any existing desktop or server processor made by Intel." beyond that they are up against a already strong market in 64 bit processors. SUN and AMD are kicking their ass in that realm. SGI's recent slippage makes them a do-or-die target for Intel to snap up.
True, Intel needed to make the change at sometime to 64 bit, their Pentium lines only doing 32bit will eventually be dropped by everybody as 64bit becomes a standard. SUN is eyeing the entry into the larger mid to upper end business market thru Linux on Ultrasparc and AMD Opteron making a nice growth path for businesses into their bigger box server market.
Sorry about the writing. Robot fingers, you know? Cliff Steele in DOOM PATROL #23
Intel figured it was big enough to set the trend by making a radical change. It was wrong and paid the price when the market didn't follow. IBM thought it was big enough to set the trend by making a radical change with Micro Channel Architecture (replacement for the ISA Bus). It went nowhere and helped kill IBM's dominance of the X86 PC world it created. The fact that Intel didn't bet the farm and loose everything is either good planning or dumb luck on thier part.
*** Sigs are a stupid waste of bandwidth.
Intel spent over a decade spending billions to convince everyone that it it didn't have "Intel Inside" it was crap. Ok, they won that battle and it is now costing them the war.
What they didn't consider is just HOW the average person would perceive "Intel Inside", i.e. as an Intel processor running the Intel ISA. So when Intel itself tried to shift gears and say "no, we meant buy anything we throw out the front door" customers balked. The sad fact is that AMD understood the implications of what Intel's marketing dept had done to the marketplace better than Intel, hence AMD64.
Plus there is the practical problem. Unless you are doing scientific computing you don't see a benefit from Itanic because there are few applications running native and emulated performance sucks. The scientific computing set tends to be writing and compiling their own code so they are in a position to benefit, hence their interest in buying clusters of Itaniums. The DEC Alpha died a similar niche market death and unless Intel manages a major marketing miracle Itanium is fated for the same sad end.
Or has everyone forgotten Alpha? A totally kick ass CPU design killed by lack of natice apps? When Microsoft tried to push off the ia32 version of Office via emulation that pretty much killed it as everyone else decided emulation was good enough for a market that hadn't proved itself yet. So Alpha died in the mass market from underperformance and languished in large racks doing scientific computing until it could no longer sell enough volume to justify the R&D to continue ramping performance. Also pretty much the same story as Sparc.
Democrat delenda est
Why doesn't the Titanic get the respect it deserves? I mean, c'mon it was really really impressive technology for the time.
And why all the Intel bashing? Well, they seem to screw up a lot these days. First Itanium, then clockspeed over core efficiency (giving us the awful pentium 4), lackluster dual core performance, etc..
The problem with Itanium is that it was hyped as the greatest thing since sliced bread. But when Merced ("Itanium 1" so to speak) was finally released to the market, two years after its originally planned introduction date, it was already outdated and being outperformed by competing CPUs including x86-based ones. Also, Merced was overpriced, was over-designed, produced more heat than the sum of all its competitors, and the market was still not ready for Itanium because specifications were released so late (even to NDA signers).
With over-designed, I mean it had way too many features that nobody would have cared about if Itanium would have been a success in its own right. Things like supporting ix86 on the chip; supporting HP-PA; supporting both big endian and little endian; supporting 32 bits pointers and 64 bits pointers (OK, the last one is perhaps useful sometimes, but still...).
To illustrate how bad the heat was: At the company I work for, in the south-east of Germany, we had an entire floor that didn't need heating in the winters -- each room had an Itanium work station and that was enough. In the summer those machines were turned off whenever possible.
Anyway, within a year after Merced, McKinley was released, and it was a huge jump forward compared to Merced when comparing heat consumption and performance. It was almost a competitive product, except for the price and the lack of easy hardware support for virtualization. Hell, it is not clear to me to this day what market Intel and HP were trying to attack, but it definitely was already a declining market at the time: Work stations (i.e. that market where Sun and SGI were performing so well in even then) and Big Iron, which is hardly a market large enough to sustain the huge development effort of Itanium.
By now, AMD64 was coming to the market as well, and it was cheaper, it was 64 bits, it had a great memory controler, and as a bonus it was backward compatible with the mainstream architecture of the day.
Now you make the business decision: What do you want to sell? Hot, underperforming, expensive Itanium-based machines for which no software exists? Or machines that are cheaper and which can run all the software on the market?
I know it has been said that Itanium is too difficult to compile for, and that GCC is to blame for the failure of linux-on-ia64. Or that Microsoft pulled the plug on Itanium when they decided to drop Windows-on-ia64. But if you see the market hype of the late 1990s and early 2000, 2001, Itanium was a hot topic and everyone was on that boat. In the end, it was Intel and HP trying too hard to make Itanium compatible with everything (which was never really possible, so what were they thinking?!) and initially not living up to their words about the performance potential of Itanium.
It is sad, because really if you look at it, there are some pretty good ideas in the Itanium.
2. where is the performance data of $2000 itanium?
3. quoting speed per mhz is meaningless. absolute speet matters.
4. all spec, jbb2000, tpc-c, tpc-h etc favor amd for upto 8 processors. above 8 proc, itanium just competes sometimes up and sometimes down with sparc, power processors.
5. with binary compatibility with windows not there and commonly available linux distros and apps not there, what is the appeal for itanium (other than forced migration of hp-ux)?
yes, itanium has lots of potential, perhaps same as ia432 had in 1980s, but right now, nothing to be excited about.
Itanium was killed by intel's megahertz marketing. Why get an expensive 1.4Ghz itanium when you can get 2 3.0ghz xeons for less? The amd-intel 1Ghz race hit it even harder, since intel had to totally sell out itanium's higher ipc for the p3's higher frequency, and meant the p3 could be brute-forced to equal or greater performance as the new, non-mainstream itanium architecture.
In my opinion the p4 was the worst thing ever to come out of any microprocessor house in the last 20 years, as it not only comprimised microprocessor design for the horrible and blind-sighted goal of mainstream marketing, but essentially caused a large part of the current TDP crises the industry is in now, and reinforced our mentally handicapped reliance on single-threaded programming.
The humor in the itanic label has nothing to do with the chips, it has to do with intel trying to have it both ways: intel chips are the most powerful, with the only metric that matters, frequency, and ipc and design efficiency matter little, but also that "oh yeah and we have this amazing chip that is so powerful but runs at half the clock speed." It was a blatent contradiction in marketing messages.
For f*cks sake, they called their double-clocked alu "NetBurst"... seriously, why not add an onboard memory controller and claim it's "SuperBandwithMaker", which uses it's amazing technology to increase the speed of your dial-up connection...
Yes, if you market to customers by treating them as idiots, expect them to choose the stupid product, and ignore you when you claim to offer another product that "no really this is a good chip, not like that other one which we said was the fastest", which is actually better for you in the long run, because you can set a new foundation for improvement.
When amd came out with the opteron at 64-bit, and with surprisingly competitive performance while still running legacy apps at faster speeds, how do you compete with that?
Here's hoping they do manage to resurrect the alpha lines, Ibm even went a little over to the marketing darkside with the g5, trading frequency scaling for TDP, but they usually manage to rebalance the two after a few years of revisions.
The first rule of USENET is you do not talk about USENET.
Java. Mainly.
Java is handy for many, many things. One thing it's not going to help is a large scale migration from one platform to the next. Years ago I am sure Intel thought it was going to be bigger, but it's not. Meaning more trouble migrating to new hardware platforms!
You are comparing a $2000 Itanium processor with a $500 Opteron. Now if you multiply the benchmark result of the opteron times 4, you will have the Itanium easily displaced for an Opteron solution of the same price.
2. x86 is bad/ugly/dirty/whatever, however Itanium is not exactly clean either. The stacked register file is a good example of that. I personally prefer x86-64, which takes the evolutionary approach: fixes quite a few of the problems of x86, while still retaining the core features.
3. x86 chips do out-of-order execution; Itanium, OTOH relies on the compiler to schedule instructions and bundle them together. The main problem here is that doing instruction scheduling statically is much, much harder than doing it dynamically. An average program has a basic block size that is less than 10 instructions. It's very hard to find parallelism within such small basic blocks, so to be efficient at all, you need to do profiling to build traces/hyperblocks. In fact, profiling on the Itanium can give you a performance boost of 30%. However, profiling is hardly desirable from a software developer's perspective
The Raven
Didn't the same thing happen with the Pentium Pro and it eventually became the P2 and P3. I wouldn't be surprised if Intel is becoming quiet about the Itanium to weasel it into the mainstream.
Intel was so late in delivery that all the high performance workstation people abandoned the Itanic.
I can get a TigerDirect reconditioned AMD 3.0 GHz laptop with 2 Firewire ports, USB, and top of the line wireless with 512MB RAM for about 25 percent less than the Intel version that runs apps at about the same speed (accounting for differences).
So the reason noone respects the Intel chip you mentioned is it's still too expensive.
-- Tigger warning: This post may contain tiggers! --
Certainly it has not been a do-all chip like it was originally said to be, but for the people who need performace, the chip is respected. Everyone who sells 16x and bigger boxes understand what the chip can do, and there is no shortage of buyers for 16x and up Itanium machines.
I'm sure that neither the server admin nor the software vendor wants to deal with the headache of finding the right versions of ever piece of software. Does Windows come with Itanium 1 optimizations, or Itanium 2 optimizations? How about HP-UX?
I realize that all procrssors suffer the same optimization problems across generations, but the dynamic execution capabilities of most chips go a long way towards minimizing this effect.
Never eat more than you can lift -- Miss Piggy
Could Apple be planning on utilizing the Itanium and become the saving grace for this processor from Intel?
Hrmmm....
* Increasing pipeline stages to boost clockspeed. Doing this confused customers into thinking P4s were that much faster than AMD's offerings. AMD had to come up with the PR ratings to keep mindshare.
* Killing Alpha in favor of Itanium. I like how the story poster didn't bother to factor in the Earth Simulator...
* Processor ID. Consumer-hostile.
* RAMBUS. Worse than DDR and was only because Intel had an invested stake in the company.
* Thinking USB2 is better than FireWire and pushing it to sell more CPUs when FireWire would be better for just about all of the applications that require the high speed interface. USB2 requires a "master" (aka intel CPU) and FireWire doesn't.
I'm sure I could come up with some more....
>1) Compare the performance of itanium with xeon/opteron in running native x86 code.
Well for some apps Itanium is a faster solution.
>2) Compare the costs of building real end user systems.
Correct.
If I'm not mistaken, next year we should see motherboards and systems that will work with either Xeon or Itanium CPUs. Then the processor cost will be the differntiating factor (and single CPU Itanium systems could cost about the same as dual CPU Xeon systems).
> 3) Compare the availability of windows xp drivers.
Yes. And the same goes for AMD64, tough to a much lesser extent. I think the answer to this lies in my comments on your second comment - once the chipsets and other infrastructure is unified, the Itanium drivers will become commonly available.
How is GCC support for Itanium? If the chip leaves so much up to the compiler, then the easiest road to success would be for Intel to build support in GCC. Recompile your Gentoo, and you have a fully optimised OS to show off. Does anybody know how good Itanium does on GCC?
10 ?"Hello World" life was simple then
Itaniums also have up to 9 Megabytes of cache. And they come with super heavy-duty FPU units. It's unfortunate that we can't directly compare the core architecture of the Itanium vs. the core architecure of the x86 without influence of the extra cache and FPU resources that the Itanium target market supports.
If somebody would build an x86 with that much cache, and increase the proportion of transistors in the x86 FPUs to match the Itanium's target market, then we could compare the relative benefits of x86 core vs. EPIC core. Until then, when you compare the two processors, you're most likely looking at differences in the logic units outside the core, which are determined by target market price/performance tradeoffs and which are largely independent of the core architecture.
Not necessarily true. What about the Pentium 4 (willamette)?
;)
When it was first introduced, not only did it get badly beaten by the Athlon in the performance stakes, the 1.4 - 1.6Ghz versions were even slower than the previous generation Pentium 3 coppermine! Plus you had no choice but to purchase extremely expensive and high-latency RAMBUS memory (in pairs).
Was the Pentium4 a failure? No. Because Intel really went to town on it turning up the Ghz and adding loads of cache to make up for a rather slow chip design. And being a monopoly over AMD sure helped, although the Athlon definitely hurt them badly...
There is of course the Pentium PRO, but that does kinda fit in with what you said though
I spent a while working on building cluster of IA64 machines in the HPC space. I don't anymore, and I can't say as I miss them.
First, as noted elsewhere, at the time when opinions were being formed, the cost of a decent dual processor machine with a useful amount of memory was about $30K (a typical HP cluster machine) while the comparable x86 box was about $3000. Today those prices are lower, but the x86 is still cheaper by several times.
Second, the HPC folks were perfectly happy running Linux, while no one else seemed to be. And Linux works on the machines. But in general purpose computing the advantages of the IA64 aren't seen.
Especially now when we have x86 with 64 bit memory addressing.
Next, the hardware (when you're willing to pay for it) was lousy. One box from HP had 15 fans for the 2 processors, in a 6 or 7U box. I think it would take off if you aligned it properly before turning it on. I can't imagine a rack full of these, much less a whole machine room (we had 2, and wouldn't allow them both on at the same time, but we had these in office space).
The EFI BIOS is truly an amazing piece of work. It's hard to work with on the console if you're a PC user expecting normal menu traversal with arrow keys (left and right arrow keys worked, but to move up or down you needed to use 'u' or 'd'). But we were building clusters, and expected to do things with the machine via serial interfaces. But the UI was even worse over a serial line. What happened to nice, easy to program command line firmware? (Note to HP Fort Collins: you're supposed to take the BEST of both worlds when you try to merge paradigms).
Yes, there was also a network interface which would take the place of the serial connection. But this wouldn't use DHCP, but rather manual network configuration, which required this funky hydra cable to connect to some other machine in order to set up properly. And each person connecting to this system would get write access to the console. There are decent ways to share a console across multiple logged in users (c.f. http://conserver.com/ for one example).
For a set of machines designed for cluster computing, they were about as unfriendly as you can get for us infrastructure people.
(Note, these hardware obvservations are only based on about 6 classes of HP machines that I've put my fingers onto; I have no knowledge of any non-HP IA64 machines).
Once up and running, they're not bad systems. But they're not so much better than x86 systems to win over most people. There's always a class of user who is willing to pay for the highest performance. But the larger part of the market pays attention to flops/dollar, and building a thousand node x86 system is still cheaper than 500 nodes of IA64.
(Disclaimer: I haven't had to price out IA64 systems for a year and change now; I don't believe my claims are invalid, but if someone can show me I'm wrong, I'd love to know it).
I have no interest in working on IA64 systems anymore, unless you're paying me to do it.
The Itanium is one of the few good things to come out of Intel. x86 is inelegant and Intel knows it. Massive amounts of engineering effort goes into keeping x86 performing the best, and a more elegant RISCish CPU architecture such as the Itanium would require alot less to go alot further. Personally i hate AMD for extending x86 with AMD64, now the bilnd masses will be clinging onto x86 for many more years. Shame on AMD for their inelegant choice. Yeah yeah i know they're a company so it's socially acceptable for them to be brutally pratical to make money and not do nice things but it is unfair to reap the benefits of their raw praticality (teh monies!) and still be "allowed" to be immune to criticism of their lack of elegance. Intels marketing revolts me but i despise x86 and I'd like to have an Itanium but definately not an Athlon 64. And i wouldn't be seen dead with a Pentium 4. I won't get started on Netburst, suffice to say that even x86 looks elegant compared to it.
This guy are sick.
The whole industry was preparing for the arrival of Intel on the 64 bit server processor market. Intel was so huge in the 90's (and if even bigger now) that everyone thought: "This is it!" HP, Compaq, Novell, ... everyone. They thought it would have such a good price perfomance issue that they threw out all their own stuff and only sell Itaniums a couple years later.
Wasn't that part of the reason why they stopped developement on the Alpha? Could someone please comment?
Also Tru64, Unix for 64bit by a consortium of industry giants was to be developed only for the Itanium.
So the anticipation was huge. And compared to the anticipation it was such a huge failure. And it still is on many applications, because nothing got ported when it came out because of the failure to meet anticipation.
The Itanium now rocks on some supercomputing tasks, but everyone and their dog is currently switching to the x86 Unix called Linux, not because it is free (they tend to buy expensive service contracts) but because the cost/performance ratio is so good mainly because of competition. If there would be a really good, stable and fast port of some Unix variant to the Itanium and some big company would support it the Itanium could still come around.
But at the moment it is chicken and egg. No cheap Itanium means no port and no port means no mass sale that leads to expensive Itaniums.
Why do we see so many disparaging opinions of the Itanium processor (all those 'Itanic' jokes, etc.)?"
Because there are about 1,000 Itanium servers that have been sold in total. I remember reading one HP quarterly report that said they sold less than a dozen in that quarter.
Even if Itanic was the best thing since sliced bread, it has no market share (probably because it's so expensive and so incompatible with x86), therefore almost nobody will waste their time developing for that platform.
It isn't just performance that matters, it's performance per dollar.
No, Intel bashing is perfectly justified. Here is a comprehensive list of reasons why you should hate Intel:l oadableAssets/AMD-Intel_Full_Complaint.pdf
http://www.amd.com/us-en/assets/content_type/Down
contains some amazing revelations
People don't buy Itanium's powered computers just because they don't need it.
You know, a computer is made out of hardware AND software.
Ok, you provide me a faster hardware, but put me in the need to find/try/buy/learn new software or, even worse, wait for the monopolist in charge to recompile and distribuite his.
Who needs so?
For LINPACK type calculations Itanium2 is probably the best what you can get and is price competitive. For a lot of other tasks it is not.
Save the bandwidth. Don't use sigs!
I will give it rave reviews.
When I first moved to the SF Peninsula in the early '90s, a friend of mine took me by a shopping center called Fashion Island. This was a classic enclosed megamall, built maybe a decade before and located at the junction of two major freeways; what made it noteworthy was that it was almost completely empty. At that point there was an operating movie theater and an ice rink; the rest was an enormous dark, abandoned hulk. Practically new shopping center, decent location in a densely populated, well-heeled area -- no tenants, and thus no customers.
What had happened to cause this? A few years into the life of the mall, in a stroke of unfortunate timing a couple of the anchor department stores simultaneously merged or went out of business. Without those stores in place, the smaller stores in the mall saw reduced traffic, and many left. This in turn made it hard to attract replacement anchor stores, and inevitably more and more of the small shops faded away. This continued until there was no way to pull out of the death spiral. The ultimate solution? They bulldozed the entire thing and replaced it with a new mall with a new name, "Bridgepointe," which appears to be doing just fine.
Maybe it's not a perfect analogy, but the point here is you had a perfectly fine mall, with maybe one or two minor problems -- and the stench of failure all over it. Once that stench settles in to stay, you often have no choice but to call in the bulldozers, or finally add 64-bit extensions to your existing processor line as the case may be.
Oh, and it would help if Intel could make decent motherboards for its own chip.
It could be my information is out of date, but I don't believe Intel has made their own motherboards for years. It's just not cost-effective. Rather, they give specifications for boards and 3rd-party manufaturers then fabtricate their own varieties. Or at least that was the way it was when I co-oped with them. *wry grin* I always wondered how much they saved after the costs of bad publicity due to bad implementations. Amazingly often, a company would ignore corrections to their motherboard diagrams and of course Intel would get blamed when the motherboards would malfunction...
This sig has absolutely no significance and serves only to take up screen space and waste the time of the reader.
Forgive me if this has been commented previously.
Intel's problems with the Itanium began with its first iteration: they expected Google to be a prime customer. However, the latter was unimpressed with the excessive power consumption. That is, compared to simpler chips and custom mother board boxes they already had the Itanium was much more costly. There are two costs: purchase and maintenance; the Itanium was more expense in each.
We've got 6 of them at work. And I've had many a sleepless night trying to get them up to par. We're running oracle and bea weblogic on them and it's been a nightmare. Coupled with the poor EMC support for them and the random init_tty lockups, I rue the day some idiot up high said go with the itaniums. We don't get the performace we see on the regular old xeons and now with the new x86_64 xeons, there's no need for them. The only reason we even have these is because of the high ram capibilities. Each machine has 16 gigs of ram, but since you can do that with xeons now, I'd love to replace them with more reliable machines. I call our itaniums our little space heaters...$30000 space heaters. *sigh*
Is the Itanium2 system over 3 times faster than the 2850? The servers my group runs are web servers, file servers, print servers, database servers, etc. I would have to be smoking crack to decide to move from Xeons to the Itanium2.
Now, having said that, our Computer Science department just recently purchased some Itaniums for a small cluster (I think about 30 nodes). They chose the Itanium because of its floating point performance.
Itanium has its place, but that place isn't in my server room...
Well, there are many reasons the Itanic failed. It was a great architecture, a neat idea. Shift all of the intelligence in the chip up to the compiler, execute in-order, optimised code, get rid of deep bypassing, etc. Generally, get rid of the extra 50% of the chip that's dedicated to turning an instruction stream into a series of vectors.
Note, it *was* a neat architecture.
Then, everybody got involved. Imagine a roomfull of architecture, compiler, and systems PhD's, each with their own pet idea. And this chip had them ALL in it. Anybody remember the i432? In a way, this was the i433.
BUT. This meant a complete break with the current codebase, and in the final analysis intel didn't have the guts for it. Especially once their hopes for compilers weren't being borne out (once, Intel was a HUGE player in the market for compilers PhD's). So the guys at Intel decided to add x86 hardware compatilbility to this. Then, since their compiler plans weren't working out, they added out-of-order execution.
Now, all of these things had crazy interactions. Suddenly, who knew what it was doing? Then the power... all those units, executing all those dead instructions - it ran HOT. Then the fact that x86 compat and o-o-o were a gigantic boat anchor in terms of chip real estate, driving the cost through the roof pretty much sealed its fate. It became a "server processor". And if you get 7 or 8 P4's for the price of one Itanium... well, your cluster is better served with those 7 or 8 P4's.
Pride goeth.
Just about the very very cheapest Itanium system you can get is $2K before you add processors or RAM.0 6123L-iR&hl=en&lr=&safe=off&sa=N&tab=wf
http://froogle.google.com/froogle?q=SuperServer%2
Come July 13th when Intel ships its x86 dual-core server support, a dual core x86 machine will be less than $1000, including CPU and base RAM.
There are few tasks where 1 Itanium system > 3+ x86 systems.
And thus Itanium has no real market other than very specialized tasks where cost is no object. For beyond the basic system cost, you need training in VLIW, special compilers, and highly specialized performance optimization technology to get any real performance from Itanium.
All in all, there are many good reasons why Itanium = Itanic. The cost/benefit scenarios are dismal at best. I would not expect Itanium to ever be successful in the mainstream.
The decision to move instruction-level parallelization from runtime (in the CPU, hardware, expensive) to compile-time (software, cheap on a marginal cost basis) ended up being a poor one for general-purpose computing. You save silicon not having all the fancy instruction scheduling, reordering, etc., but you lose the knowledge of the runtime environment the hardware has when you move it into the compiler.
.NET bytecode to native code and Transmeta x86 to native VLIW) can do a better job because they can profile the running code and get a better handle on likely execution paths. These would be a good match to the VLIW Itaniums to compensate for them lacking that "complex" hardware to keep the execution units supplied.
Sure, there's a lot more processing you can do off-line in the compiler, but you also have a lot less information about how the code is actually going to be executed at compile time.
Theoretically, JIT compilers (Java and
The Itanium2 makes a good supercomputer chip because you can optimize your code very carefully and you've got a good idea what the data looks like and what branches will be taken, etc. at compile time.
Most amusing to me was that the early versions had the chip serial numbers on the area covered with the heatsink. Removing the heatsink voided your warranty. You needed that serial number to get warranty work done on the processor.
This sig has absolutely no significance and serves only to take up screen space and waste the time of the reader.
I thought the official name WAS Itanic. I figured they just liked the movie, but were missing the obvious bad omen. Oh well.
You see? You see? Your stupid minds! Stupid! Stupid!
The computer-architecture blog "realworldtech" has been slogging this one out in extreme detail (particularly as regards integer performance, where the Itanium has seemed to lag); see http://www.realworldtech.com/forums/index.cfm?acti on=detail&PostNum=3510&Thread=1&entryID=52549&room ID=11
"My opinions are my own, and I've got *lots* of them!"
Maybe this is a topic where the pathetic dotnet would make sense - as a transition platform from the dinosaur x86 architecture to something new, better, where you only have to execute fast the IL language code, and not the native x86 instructions. Then PPC and x86 and ia64 could all really go at it. Unfortunately, while everything is x86, dotnet sucks because it's slow, bloated, and adds yet another layer or black magic between you and your hard earned cpu cycles. In any case, more power is not what we need anymore, but less power - I'm glad to see the other news about Transmeta. I'd love to see a chip that uses 5Watts and gives me 1.5 GHz P4/Athlon performance. I'd be content.
The Itanium isn't making any headway in the marketplace for the same reason the PowerPC didn't. It doesn't run X86 code, not natively at full speed.
In the early years of PowerPC, the plan was for it to replace both the 680x0 (in Macs) and the 80x86 (in PC), thus bringing the platforms together on common hardware. The lure was better performance.
The PowerPC processors in that era cranked out roughly 80% better performance than X86. Apple went for it, but PC makers weren't interested. "What, you expect us to break compatibility for a measly 80% increase in performance? It is to laugh! HAHAHA!!"
Now they are laughing at Itanium for the same reason. I suspect if you could get Itanium to run 300% faster than X86 at the same price, then the PC makers would sit up and take notice. But they simply aren't going to be bothered with it unless the gain is really huge and dramatic.
Let me tell y'all a little story.
Back in '94-'95 i was doing the third grade of the Computer Science course at the Royal Institute of Technology, which meant I had to choose a specialization. I chose "Computer Systems", ie. processors, busses, caches and what-not.
This was a very exiting time to be studying processors since (for a fleeting moment) Intel processors where the absolutely worst processors among the serious combatants.
Yes, you read that right. The Alpha was (of course) and unstoppable juggernaut, but through a freak act of development schedules the new MIPS had managed to outstrip the latest Alpha.
After MIPS and Alpha we had PA-RISC, SPARC, PPC and then finally the pathetic, lowly Intel x86.
Alpha had strong plans of totatlly replacing the x86 by offering Alpha based x86 emulations that were faster than the fastest x86 in running x86 code.
But now, Intel announced the Itanium.
Apparently, all the CPU makers sat down and discussed this, and agreed that "They may be last right now, but they have piles of cash. They could do this. They really could."
So, what did the competiton do?
Because of aquisitions, they also happened to be saddled with the best processor ever made, the Alpha.
Stick with dying Intel... Develop best processor. Hmm...
Well, you all know where HP is going.
And then what happend?
Intel didn't deliver... and didn't deliver... and didn't deliver some more.
Year after year passes...
When the Itanium was finally delivered, it was obvious that every other platform could have kept up, if they would just have kept developing their processors!
But they didn't and now they sleep with the fishes.
Conclusion: By making their Itanium announcement, Intel slew four out five serious competitor. It doesn't relly matter if the Itanium sucks. In fact, the Itanium would be Intels greatest success even if they had never delivered it.
I choose to remain celibate, like my father and his father before him.
Itanium was expensive, had very poor backwards compatibility, had poor performance on existing software, required a complex compiler and intricate coding, and was notoriously finicky to get up and running. Even now, the only advantage that the article offers for Itanium is its gigaflops performance running a heavily-optimized floating point benchmark on a 4,096-processor system. I doubt if a line if forming to buy one, based on that.
Based on what it offered to the IT industry, the Itanium got way more respect than it really deserved. If anyone other than Intel had tried to sell the world on the Itanium, they would have been laughed at for decades.
It's heavily targeted towards highly hand-optimised, floating-point code. Ideal for scientific applications.
But it runs hot, it can't be ramped up very much higher, and its average-case performance with the best compilers for its platforms are MUCH worse than best-case, and much closer to worst-case.
The EPIC architecture positively demanded a much smarter compiler, and no-one outside a lab has been able to deliver on that.
Couple in poor integer and branch performance, memory bandwidth that is outclassed by the Opteron in practice, and what is still a stupidly high price reflecting a niche processor for specialised scientific computing, and you realise that the Itanium2 is a good chip for a supercomputer which is running a hand-optimised scientfic application -- but absolutely bloody awful for anything else.
Even Intel realised that for conventional servers and desktops, Itanium was an awful choice, which is why they begrudgingly adopted AMD's x86-64 as EM64T. Microsoft has abandoned future ia64 support in the Windows branch, leaving Linux as the only viable platform (which makes sense because that's what the supercomputers these days run anyway). Itanium has no general-purpose computing future, and never really had a general-purpose computing past.
Intel were heavily over-optimistic, and overspecialised the processor without a compiler two-generations ahead that it needed to back it up. Since it's a bona-fide failure in everything but a highly-specialised niche (at which it is now only as good as the competitor POWER5, but more expensive), but Intel marketed it as an unsinkable powerhouse, the future of computing; the "Itanic" label is rather well deserved. Let it die; use an Opteron or POWER5 instead.
sounds like the original poster hasn't actually used an itanic. they suck balls. big hairy moose balls.
One of the biggest problems: You have to re-write your code to get decent speeds. People have code that is quite speedy on i386, macppc, sparc, mips, etc, but is dog slow on the itanic (both the itanic, and the itanic2).
another problem: they have fast giggleboops, but they aren't really fast. a friend of mine works for an animation movie studio, and intel gave them a 32-machine cluster to try out for the render farm. so my friend put the itanic cluster against a 32-machine cluster of what they were currently using: 3 year old alphas. The alphas were 1.7 times faster. Even after an intel engineer came in to tune their machines, and massage the code.
Consider that Apple wanted to go with the Alpha for their Macs, and Digital wouldn't return their phone calls. The PowerPC was definitely a "second choice" for them. If there had been a major market for Alpha, and if DEC^H^H^HCompaq^H^H^H^H^H^HHP had had competent management, the computing world would be a very different place today.
"My opinions are my own, and I've got *lots* of them!"
Itanium was dead at the starting gate for the same reason PowerPC and Sparc are dying: Not x86 compatible; doesn't run x86 software.
Sure, you can get the poorly tested Itanium versions of Linux, suffer through all the bugs that only express themselves on that architecture, recompile your apps, argue with the commercial vendors to get them to recompile their apps, etc. But why would you do that? At the end of the day the manpower costs far more than buying some extra servers.
If Intel (or anyone else for that matter) wants to create a commodity non-x86 chip, they'll have to solve the chicken-and-egg problem with the software. Since that problem is nearly insoluble with the direct approach, they'll have to find another way around it.
Moderating "-1, Disagree" is simple censorship. Have the guts to post your opinion.
I just read a whole write up in the Sunday paper this weekend about a local (Colorado) company that is writing some new and cool stuff specifically for the Itanium processor. Matter of fact they used to work for HP / Intel I believe and set off to start this company. So they seem to think it's still got tons of potential and will still be the next big thing.
Exactly, take a look at SCO for example. Despite their insane ramblings in public, the main point of their argument with IBM is as follows:
1. They made an agreement with IBM for Project Monterey. SCO brought their UnixWare to the table, both sides brought engineering resources, and IBM promised $1mil in marketing.
2. Around the time Monterey was done, IBM said that there was no market for 64-bit Unix and didn't do any of the promised marketing.
3. SCO is so cash strapped, they have to sell out to Caldera, who renames itself "The SCO Group" and proceeds to go insane.
If Santa Cruz Operation and IBM weren't so concerned about the Itanium, SCO (or maybe The SCO Group) would be in much better shape right now.
Out of curiosity, I just checked itanic prices at dell. The cheapest configuration for a single (dual capable) 1.5GHz itanic with 2GB RAM and 36GB SCSI HD is over $17K. For comparison, a similarly configured 3.6GHz Xeon (also dual capable, 2GB RAM) is just over 5K.
The article poster is simply trolling. Where the fuck can you get an itanic for $2000? The cpu *alone* costs that much! The article that the moron linked to confirms this: "The 1.4GHz Itanium 2 comes out Monday for $1,172 in 1,000-unit quantities. A 1.6GHz version comes out in May for $2,408 in similar quantities." (last paragraph)
Need I give any more reasons for why it's not popular?
___
If you think big enough, you'll never have to do it.
real easy: change the name. It's not "itanium 2". It's "iridium" or whatever their newest name will be.
In my corp it's not due any respect. Despite assurances of a stable state the processors led to numerous core dumps and app/os level crashes of various sorts. The vendor support for them sucks too. In short for serious prime time use when millions ride on uptime, these chips SUCK. I'm generally pro-Intel but in the 64 bit world AMD is currently alot better (assuming we're talking a Windows based OS...I don't have any *nix Itanium experience)
This is not the point. Chips that perform very well per MHz normally have longer (that is, more complex) pipeline stages that can't be clocked as high as chips that clock higher with more, shorter pipeline stages.
It is only fair to compare chips delivered performance on the day you take your measurement, and then perhaps adjust for cost. You can't adjust for MHz or you'll get completely misleading results.
This sort of compiler technology goes back to the early days of RISC (MIPS, etc). The compiler has to schedule instructions, manage speculative execution directly, etc.
If you think Itanium is complex, you should see some modern DSPs. 8 functional units/8 instructions per cycle, all sorts of dependencies on cross data paths, etc. Not to mention huge numbers of specialized instructions for working on multiple data (like "add 4 8-bit values to 4 others and saturate the results"). The compilers will not equal a good human, but they do a pretty darn good job.
I see it reported as a server processor but not for the desktop.
Why is this?
Thanks
Your nick is quite appropriate for your post content.
Couldn't (alpha) have (alpha) something (alpha) to (alpha) do (alpha) with (alpha) HP (alpha) stabbing (alpha) another (alpha) processor (alpha) in (alpha) the (alpha) back... one that didn't need heroic measures to get top of the line performance...
Beating x86 is like kissing your sister.
That question answers itself: You think differently from most people. Highly specialized, hand optimized massively parallel predictable crunching seems to matter to you. It doesn't to most people. You're in a minority. Get used to it.
BTW, i860 and Alpha suffered from basically the same problem.
Wow that G4's fast! It is if you code in altivec. How many do? Very few. Wow that Itanium's fast! It is if you code for IA-64. How many do? Very few. Hey guys, Lets build an Itanium PowerMac!
Nothing sucks like a Vax, nothing blows like a PowerMac G4
Itanic has its reputation because its a more fundamentally broken design than the i432 was...
It's the instruction set. It originally didn't use x86 (I have not kept up to know otherwise).
It is actually a VLIW (Very Long Instruction Word) chip instead of CISC. Yup, it's a RISC chip but don't say that to Intel! It takes RISC instructions and strings them together to make a VLIW.
That's why when vendors wanted to go 64bit they went with AMD chips, no porting. It can run 32bit and 64bit and it does it in x86 chip talk.
So if you go with Itanium;
1: replace x86 boxen with Itanium boxen.
2: rewrite all your stuff for Itanium
3: your boxen do nothing until step 2 is complete.
4: ???
5:Profit!!
or you go with AMD Opterons;
1: replace/upgrade boxen with Opterons
2: keep running what you have while all your SW is updated to 64bits.
3:Profit!!
A couple of points that seem to have been missed when looking at why the itanium less widespread:
- each CPU is quite large, having a square surface area for the unit about 2" x 5" and it's about 2" high
- That area includes a voltage regulater and the passive cooling fans
- It doesn't include any of the necessary active cooling
If you add these physical factors to the points already made about heat, power and EFI bios, it's obvious to say that Itanium won't run in your mini-ATX destop or laptop. This isn't a slam on the design, as it was never designed to run in those form factors, but it's hard to see how any cpu today is going to have a wide use if it isn't available for dual use for destop and servers. Once you eliminate the desktop market, (and I'm going to lump the workstation market in with the servers) the number of places you can sell these processors drops considerably.Once you start adding in the lack of Windows support for itanium, the strides that the 86_64 architechture has made in capability, and the low numbers of current adopters, it's not looking like Itanium will ever gain widespread acceptance.
The Internet has no garbage collection
Most comments are from teenagers who parrot crap spouted by bitter, old fans of Alpha.
Even the link to the article about how much cheaper it has become, says this:
That price is less outrageous than past prices, but it's still not competitive in the real world.As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Hundreds and hundreds of products have been killed or permanently crippled because their first versions were terrible.
There's the answer! Now it's clear what has to be done to make this processor a success!
The first version was terrible, you say? Well, then simply apply the one and only strategy that always guarantees that an absolutely horrible first version becomes a great market success.
Put a sticker on it with the name "Microsoft".
--
Terrorists can't threaten a country's freedom and democracy. Only lawmakers and voters can do that.
Price, heat, and power.
The Itanium is a very good performing 64-bit processor, BUT it's expensive (mostly because of the outrageous cache sizes), it's hot (mostly because of the outrageous cache sizes), and it draws a lot of juice (mostly beca...well you get the picture).
It's also a bear to write apps for. Intel did a lot of work on the compilers, and they are very nice, but there are a limited number of OS options for it and they are randomly stable depending on from whom you bought the hardware. I guess if you're going to setup a small server that's not going to do much that would really require a 64-bit processor you can make statements about OS options, but if you're going to use the machine for scientific/engineering work there aren't a whole lot of options (SUSE and RHAS).
I do agree, it's a great processor IF you can afford it (and I don't mean the initial sticker price). Let's also not forget the small production volumes and the near decade of R&D to get a working chip.
Why this chip is not for me are two reasons:
1: I'm not buying one before the software is ported to it -- and at a comparable price to its PC equivalent!
2: It may be a step above an Opteron for floating point, but is it still that step about a dual processor Opteron that I can buy today for less money than a mono-processor Itantium?
As for the "Itanic" jokes (all of which are way off-base, since heat output of any H.M.S. Itanic would melt any iceberg long before it could do any damage), blame The Register. I saw them use the term long before anyone else.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
well, from my experiences with this chip...
The first versions were so terrible that it soured alot of people on the whole line.
The Itanium2 is a big improvement over the first generation, and if you are willing to hand tune all your code in assembler you can get some very respectable performance. The reality is that if you compile your code with any current compiler and test the Itanium against the opteron the opteron will win. There are exceptions but this is what I have seen.
Namaste,
Here's a reason: P4's are like DDR667, Itanic's are like DDR266.
Incidently, an IT guy might find a performance increase for some applications when migrated from 64-bit big-iron HP/IBM to big-aluminum like x86 w/ the memory extensions---due to faster memory speeds. FP perfs aren't so import when you got WSL sloshing around in 32G of RAM. What might be more import is how fast can you read that 100M java object.
The big deal with Itanium is that it's a change in ISA (Instruction Set Architecture). The 32-bit to 64-bit jump on Intel/AMD isn't a fundamental shift in ISA like the Itanic, hence migration is less complicated.
It's like moving from the United States to Canada. There are cultural differences that make for some adjustments, but there are enough folks speaking English to get along. Unlike, moving from India to the United States, where hardly anybody can speak Sanskrit :)
AUM, Shanti
Your quoting FP performance. The "integer" (aka general purpose) performance isn't nearly as competitive. This is because its a static VLIW machine, and its hard to write a good VLIW compiler. Writing fast FP code is simpler. Then there is the fact that the Itanic is 3x the hardware of the machines your comparing it to. Bigger caches, and all that. Your misunderstanding of clock rate is also simplistic. In order to get the Itanic faster they would have to create a longer pipeline, this would more than likely decrease the IPC and keep the processor from scaling lineraly.
Basically it was pointless. we don't need yet another processor targeted into the same market the POWER64/SPARC64/PARISC and now the X86-64 etc are in.
The whole arch is a mess in my opinion its accually probably worse than the x86, this is evident in how long it took to get the thing out the door. For a processor based on the idea that superscaler wasn't easy and wouldn't perform its beginning to look like the itanic is accually in that boat. Its a dead arch, there are orders of magnitude more x86-64 machines out there even though the itanic had a two year lead. Why should I use itanic when there is a larger software base for PPC/POWER and its multivendor?
POWER is cheaper,faster and more mature and it can barely compete with x86 in the desktop area. ARM has pretty much taken over the smaller chores (cellphones, PDA's MP3 players etc..) and smaller chips like the 8051 clones sit below that.
Give it up, it was stupid, Intel was wrong. My opionion is that itanic was a marking plan to lock up the processor market. If we were all forced to run itaniums back in 96-98 then we would all be buying intel chips for everything. Instead intel had to release the P-Pro to keep ahead of Cyrix/AMD, only they never got far enough ahead to kill AMD to release the pressure and transition everyone to Itanic, where theyhold all kinds of patents and copyrights on the instruction set. Plus they couldn't make the thing work and it slipped for 5 years.
Melt?
(Aiming for the PITHY +1 Moderation)
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
It should be obvious that the quality of an architecture is now totally unrelated to its popularity: the x86 heap-of-shit is now so popular that it's being used by people like Sun and Apple in place of superior architectures. "Compatibility" and "low risk" are seen, rightly or wrongly, as more important than the technical attributes of the processor. Other processor vendors have found this out in turn over the last decade. Somehow, Intel were the only ones who hadn't noticed.
To drive this point home even more: we were told by Intel that to do our own code generation was not recommended. They recommended that we use their C compiler backend. For us, this was not an option, and we never did the port.
No gamer is going to buy a 1.4ghz chip over a 2.4 regardless of the actual speed.
What are the fast chips? DX20-133mhz now? They would prefer a dx50-60mhz chip if intel could make it.
itanium (itanic) is a poor design for anything other than numbercrunching. It is a relic of theoretical supercomputer designs that were popular in the late 1970s. itanic shines on floating-point benchmarks, and is mediocre at best on everything else.
Since the late 1970s, we have had RISC and then superscalar RISC, some now with elements of VLIW. This provides better real-world (general-purpose) performance using substantially less power and fewer transistors than itanic.
Modern RISC processors (including x86 which are RISC internally) can reschedule execution of instructions dynamically (i.e. at run time). itanic can not. It relies on the compiler to schdule the code. It is only possible to schedule code well at compile time for very well-defined problem sets i.e. floating-point maths intensive programs like numerical simulations. NASA currently owns 5% of the world's itanic processors (in a single machine).
itanic was intel's attempt to kill the 64-bit RISC market, putting all of its competitors out of business. Like all great megalomaniacal plans, it has failed. It was a marketing-driven processor, and a failure.
It can't compete with clunky old UltraSPARC IV on server-oriented workloads. Even that market, which isn't big enough to sustain Sun and its processors, is orders of magnitude bigger than the market in which itanic has any relevance.
For big servers nowadays, you have a choice between Opteron and POWER.
In science and engineering, you're often better with something like Opteron, POWER or something fancy from Cray, NEC or Fujitsu. itanic runs hot and consumes too much electricity.
Has anyone ever seen one? I haven't. There was one at a show once on the Red Hat stand, but they wouldn't let me performance test it... and they wouldn't even let me see it because it had over-heated.
itanic is about the most expensive turkey in computing history.
Stick Men
Thanks for answering. Double thanks for not being snarky!
It's like a whole new slashdot.
-Peter
Pure and simple.
I mean really what idiot thought we were stupid enough to buy this garbage let alone eat it????
I guess Cliff does.
Tell you what cliff why don't you go jump off a cl....
No sorry it's to easy and you're not worth it.
But i mean God seriously itanium? just what were you smoking anyway.
Coward? Coward! Thems fighten words!!
Well, from what I've seen, in #1, itanium is rarely the faster solution. Particularly in, say, desktop productivity suites. But even if it is close or marginally faster, it still has to contend with #2, which may well improve with motherboards that support xeon/itanium, but my guess is that pentium optimized motherboards will continue to rule the desktop. You certainly may start to see more itaniums in servers if the price is more competitive. As to #3, AMD64 retains the enormous advantage that even if you can't get 64-bit drivers, you can sit in 32-bit mode and still get a great price/performance deal.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
... it kept me warm in the winter.
The Pentium Pro was not really a flop. The PPro was faster and better then the Pentium. Operating systems such as Linux, FreeBSD, OS/2 and WindowsNT all worked very well on the PPro. The problem was that the PPro was optomized for 32bit opperations. Software that used 16bit opperations took a performance hit (25% slower IIRC). This was a big problem for Microsoft as Windows 95 still used a lot of 16bit code (as did 98 and ME). At the time Microsoft had no plans for a consumer based 32bit operating system. Intel was basically forced to redesign the PPro for better 16bit performance and the Pentium 2 was born. In the process some of the more server oriented features of the PPro were stripped such as SMP and 36bit addressing.
Was your company database software compiled for the Itanium.. or were you guys using the x86 emulation mode? (i sure it was the latter... otherwise your results would be quite different). the parent (by jarich) needs a psuedo-mod of uninformed
1) compare the performance of itanium with xeon/opteron in running native IA64 code.
2) and 3) don't match, since "real end-user systems" won't run Winblows XP. Windows Datacenter 2003, maybe - but not XP.
Compatibility may be an issue with volume sales, but the Alpha was definitely respected in a way the Itanium is not.
And the reason is that just keeping up with x86 is not good enough. What Intel was saying was exactly what had to be true to make it worthwhile to introduce a new processor: we will leapfrog the x86 and enter a whole new world of gigaflop goodness.
If they had managed to deliver the performance - meaning performance far in excess of x86 processors - then there would have been respect, and they could have worked on the price. But just keeping up with the x86 is uninspiring.
Add that to the initial wild promises, and you get even more cynicism.
Currently, Itanium is pretty much exclusively used in really high-end machinery such as SGI supercomputers, which are not really within the reach of the ordinary computer enthusiast. So, since there are more urban legends about the chip than actual knowledge, it's easy for such myths to spread.
But there are some trends that seem to push this chip outside the exclusive markets - e.g. the aforementioned SGI is working hard to translate its knowledge in the field of supercomputers into things more accessible, such as high-performance enterprise servers. If Intel makes concessions to the price, those high-performance servers might actually use Itanium.
Of course, if Intel keeps the price high, then those "small supercomputers" will just use some other CPU, to keep the total price down.
Bottom line: it's a perception thing. It's easier to ridicule what is not known, hence the "Itanic" myth.
First off, remember that the current Itanium is the Itanium 2. The original Itanium was a failure in almost every way.
A reason why the DEC Alpha and other processors died wasn't because of a bad design, it was because the software support wasn't there. Think about it, if you could run all your Windows apps under Linux, more people would run it. The difference in what software is available is a big issue. Sure you can recompile many things to run on a given piece of hardware, but for endusers and even when it comes to servers, if a company needs to spend a lot of time making a program run well on a new architecture, and it doesn't sell well, it's a waste of resources to do it.
So, look at the current market. The Itanium 2 is horrible when it comes to performance under Windows 2003 or Windows XP. If you have native support, then the performance comes up to speed a bit better, but compare that to the performance of an Athlon 64/Opteron and the P4/Xeon chips.
Then you have the issue of how fast does it do in the real world. Theoretical performance in benchmarks is one thing, but how well does it work in a real-world environment? Can you say that the Xeon can hold it's own against an Opteron with a multi-processor system? In some applications it can, but as time goes on, AMD has tweaked the Opteron and Athlon 64 design so even in areas where Intel has dominated in the past, AMD is getting much closer in terms of performance.
Why go to a new architecture, which means all new software when you can stick to a tested architecture that runs all your current applications faster?
The Itanium 1 is like the Pentium Pro. In theory it was better because it dropped the legacy support that plagues the x86 world, but since it failed to be better at running the current/older software for the x86 world, it didn't do well. Eventually the Pentium Pro design was used in the Pentium 2, and as a result it did well, but the original implementation was horrible.
The Itanium 2 design may eventually make it's way into the current x86 world(compatability and such), but very few would be willing to switch to it.
The Opteron/Athlon 64 design is up to 2.8GHz, and is expected to eventually make it up to the 3.8GHz mark which is where the highest end Pentium 4 processor is. All things considered, I'd stick with the Opteron for multi-processor systems, or Athlon 64 FX 57 for single-processor. You get the best of all worlds without the drawbacks.
Intel & HP could give for a upgrade from Alpha to Itanium was/is: "Easy migration to MS Windows Server and MS SQL server", what bloody respect do they deserve ??
The whole Itanic story is based on this Windows compatibility and migration to Windows. The power Itanic deliver is not better than what the target market allready have... and the Windows aspect is not a argument in the UNIX world.
They don't deserve our respect... they do everything they can to insult us...
I am not affiliated with the Itanium or the top500 list in any way.
Sounds like PR Bilge to me. And if only FP performance mattered, then the Cell processor would Rule Them All.
As in recent years the Itanium does well, easily beating x86 processors even at its low clockspeed (1.4Ghz).
See note above about FP performance not being the be and end all of what many of us use our processors for.
We'll be paying for x86 compatibility indefinitely.
Don't know about you, but I think we'll pretty much all be running under the quite-a-bit-nicer AMD64 a lot sooner than indefinitely. Even holdout Apple has now blessed the Intel ISA future roadmap.
So it's statements like the above leading me to call it Bilge.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Java, and most other byte compiled languages, perform quite poorly on processors with heavy branch prediction penalities. If Intel thought Java was going to be the key to Itanic's success, then Intel's left hand and right hand were having a serious failure to communicate.
No code for the chip has really been written to take advantage of its security features. Therefore it acts just like an avaerage chip. Take a look at this article from The Denver Post on the subject http://denverpost.com/business/ci_2848039
Indecision may, or may not be my problem! -- Jimmy Buffett
Boy, that second processor for only $6000 over the single-processor system is a real killer. Definitely a Beowolf cluster of single-processor boxes is the way to go.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
The pc-centrix /. crowd who thinks a 2x is big have no clue what an itanium is for.
In Soviet Russia, the Itanium...sells!
Intel has the same problem as Microsoft... Product development over the last decade has been driven by marketing, not by developing good technology.... They've spent millions adding features and bloat to their respective products for the purpose of marketing, and the results are Itanium, Pentium 4, and (for Microsoft) Windows XP. Both companies' 'product segmentation' also is to blame... Itanium -could- be a mass market chip -if- Intel didn't need to keep it's prices high in order to prop it's gross margins.
And why are AMD and Apple technically 'better' than Intel and Microsoft respectively yet have low(er) market share? Because their focus has been on technology. To compete, they HAVE TO be better than Intel and Microsoft, so their money is better spent on true R & D, not marketing. Since Microsoft and Intel still enjoy the benefits of economies of scale IBM handed them in the early 80's.
Might turns some heads at that remark, but think about it like this.
The Pentium Pro, the True P6 chip, that was developed during the early 90's was nothing to anyone, except in business / mainframe environments. The P6, with its true cores and integrated memory, was hands down the champ, when compared to the P5. Amazingly enough, every incarnation of the Intel X86 processor is STILL based on the P6. Yes, there have been improvements (MMC, SSI, HT, DC, etc) but the core design principles themselves have remained unchanged.
The Itanium (which sucked ass when I was developing on it) was a new start, at a new core. Itanium2 (which is what became available shortly after Intel realized how bad the I1 was) was a push to market.
Now, a few years have passed, and AMD got the great idea of releasing the x86_64, to act as a bridge between 32bit and 64bit. Intel IA64 requires wholly new programs to really take advantage of the 64bit, whereas Opteron/64 don't.
Intel then realized its mistake, in not creating a 64bit bridge. Tadaa! EMT64 comes out.
But what is happening on the back side? Itanium going away? No. The ia64 will continue, just at the p6 did. Give it a few years, and you will find that x86ia64 chips are starting to be developed. That being, a single chip with a front side x86_64 implementation, and a backside ia64 implementation. These chips will act as a branch point between the two architectures.
Intel will continue to develop, until the market has reached the necessary 51% saturation, and then start phasing out x86.
Processor development is near linear. If clock speeds can't be increased, you increase the core count. When core counts can't be increased, you increase to bus width. Lather, Rinse, Repeat.
1: Both are monopolies (Intel case is being proven now).
2: Both want to continue acting as if they not monopolies.
3: Both are taking until Version 3 to get it mostly right.
Case closed.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Given your little performance comparison chart:
rank processor ghz (gflops / #procs) speed
#5 ppc970 2.2 (27910 / 4800) 5.81
#7 itanium2 1.4 (19940 / 4096) 4.86
#10 opteron 2.0 (15250 / 5000) 3.05
#20 xeon 3.06 (9819 / 2500) 3.92
Maybe the question should be, why doesn't the ppc970 get the respect it deserves? I suspect that the ppc970 has a much smaller die than the itanic. Sure the clock speed of the ppc is 0.8 GHz higher, but who cares if the ppc costs 1/2 to 1/4 as much? Also, it would be interesting to know how much power each of them uses.
I'll admit, I've fallen for the "Itaniums suck" line, but clock-for-clock they appear to be speedy, based on the numbers given. Now if only I had a beowulf cluster of 3.06 GHz Itaniums...
"Murphy was an optimist" - O'Toole's commentary on Murphy's Law
Itanium is (or used to be) very power hungry. When Itanium released, someone from Google remarked that they had no interest in Itanium because electricity was a very large operational cost.
You might have guess form the price that it's not a desktop chip. It's designed for high performance FP calcualtions almost exclusively. There's a market for that, we have an Itanium cluster here for just that reason.
I think that's part of the reason people hate on it on Slashdot, is they think it could and should be deisgned for their app, which seems to be webserving and scrolling text for the large majority. It's not, for that, Intel's solution is the P3 and P4. The Iatanium is if you need to crunch massive amounts of FP data for research or modeling or whatever.
I think it would be like hating on ARM (Xscale in Intel lingo) chips because they don't run at 1ghz+ speeds and have no FP unit. Well the reason for that is they are designed to be low power embedded chips, and they do so brilliantly. That they don't complete performance wise with a P4 isn't relivant, they aren't designed to.
In the market that it was designed, the Itanium seems to compete fairly well. Like I said, a research group here got a cluster of them. Reason was they calculated it to give them the most bang for their buck for the kind of work they do. Now their desktops are all P4s, wouldn't want to waste money on an Itainum there, but that doesn't mean it's worthless.
I was recently involved in a build up of a clustered system. We evaluated several dell based Itanium 2 servers. At the end of the day we had issues with stability when we started to MOSIX the itaniums, It seemed as though there may have been less development in resolving bugs. That maybe because of the cost being so dam expensive originally. In the end we got better performance / £ to use an Opteron based system. In the next few weeks we will be taking the nodes down to upgrade to dual core processors and they are STILL cheaper than the Itanium processors from Dell. Total cost of ownership made it better to NOT use the itanium and buy an extra nodes and get better redundancy etc.
I recall the time Intel tried to introduce a non-X86 compatible CPU in the 1980s, it flopped because it didnt have good compilers and applications. It was called the i986(?). I recall it had special CISC intrustion to optimize object-oriented computing. I cant find it on Google, probably because this was pre-web and people dont write much about failures.
I imagine the 3GHz Itanium does a great job power Santa's Desktop.
It makes as much sense as imagine what a 6GHz G5 would do or an 8GHz Pentium IV.
They all do nothing because they don't exist. I don't really know much on the subject, but my understanding is designs can be made to clock real high but do less per a clock (P4) or clock real low and do more (Itanium) or somewhere inbetween (G5 Athlon).
I know there is more to it then that, but part of the reason the P4 is so inneficient is it was designed that way. The reason the P4 clocks so high is that it needs too, the reason it needs to is it was designed that way, and the reason it was designed that way was so it could clock high.
Wow, sent an e-mail as suggested when clicking on "use classic" banner, and got a fast response that addressed my msg
Actually I think they chose the Itanium specifically for its ability to handle large memorysets.
The Altix is the machine you're looking for I think.
You are not entitled to your opinion. You are entitled to your informed opinion. -- Harlan Ellison
You'd need approximately 7967 2GHz opteron processors to perform as well as 4096 Itaniums.
Opteron 846 (2.0GHz) which was used in this system.... $675 OEM or $775 Retail.
Itanium 2 (1.4GHz) used in the compared system is approximately $2050.
$8.4 MEEELION for Itanium 2 system
$6.2 MEEELION for Opteron 846 system
Opteron runs standard 32 bit x86 stuff (read: compatability)... is easier to support (read: very common architecture).. and is cheaper...
Here's a $12M budget for the next few years to buy new hardware.... do you want 1 Itanium 2 system or 2 Opteron systems (twice the power, and all the other stuff).
Here is someone elses great explanation for why Itanium/Intel suck...
"Itanium however, is completely based on the future. It has no real history. Only 2-3 processor releases, and all the big server sellers are getting behind it. And if you were to judge Itanium by Intel's history, you know that when Intel pushes something, it happens. I know they too have had failures, but seeing the big server oems get behind this should be a good clue to where this is going. Alpha engineers were acquired in 2001. Merger took place a year later. Not sure if Intel knew about the merger when they were making the deal. I think they were just trying to buy Alpha IP, but I could be wrong about that.
Intel pushed Rambus in a huge way and was flatly rejected by most consumers as being too expensive at the time. This lead to AMD picking up a few crucial market share points as i recall from history. So much for Intel's strong arming there. Itanium's history is what, 10 years or so at this point? Over 10 billion dollars invested in R&D. It is the love child of Intel desperately wanting a big chunk of the big iron market and HPQ's dying PA-RISC. As it stands right now Intel has almost ZERO market share of the big iron market, that is still ruled by SGI, IBM and SUN. Yes, lets discuss why Intel was buying Alpha IP shall we? Could it have anything to do with them blatently ripping off DEC Alpha tech and getting caught and rather than facing an incredible lawsuit they decided it would be much more cost effective to just buy the company?
You like to clamor about how Itanium adoption is a given for all of these leaders of big iron technology. Lets look a little closer as to why this is. SGI is on the ropes and arm chair quarter backs such as myself wonder how much longer they will be withering on the vine before they are bought out or file for Chapter 11 protection. MIPS for high end computing is quickly becoming passe and has been that way for years. The only reason SGI can still use the chip is because of their NUMA architecture and they leverage massive amounts of cpus to work in tandem with one another.
HPQ as already mentioned is in the love nest with Chipzilla. Some consider the Itanium to be the greatest quo ever pulled off in corporate history. Basically Intel has footed a 10 billion dollar bill for HPQ to redesign a brand new architecture that no one really knows or is comfortable with from a software writing point of view. Adoption at this point in time has been abysmal at best. It certainly doesn't help the Intel cause when a bug is discovered with this supposed "enterprise" level chip and the only solution to it is to scale back significantly on its clock speed so that it doesn't rear is ugly head."
Silly question -- I believe that Itanium is getting all the respect it is due. It's just not due as much as you seem to think.
Ian Ameline
I have to laugh at this statement.
"The supercomputer people are serious about benchmarking (no easily tricked microbenchmarks or reliance on closed-source commercial apps)"
What pray tell do you base that assertion on? I work in the so called super computing field and I can tell you that statement would be false, if it had any meaning whatsoever. Benchmarking is good for one thing, advertising for funding through press releases. When it comes down to it what you want is to actually get work done. The itanium makes a big noise but comparitively little actually gets done on it. First because it never lives up to its performance hype running real code, and two because writing code for it (or porting code) is far to costly and time consuming. For all that extra effort all I need to do is go buy more x86 and still save a bundle. Unless you want to argue that the itanium helps to employ more programmers and thus stimulate the IT industry I just do not see the attraction.
Check out some of the newest Terra Grid members. I will give you a hint, they are not running itanium. You know the terra grid, one of the biggest Itanium booster projects?
When it comes to actual science it is far better to have an app that is actually running, analyzing or producing data than some theoretical ability to do what you are doing today 15% faster. Oh and of course you will only get that performance if you spend 200% or more on software development than you already do.
Also not all science is done with floating point. Floating point is all nice and good for generating things, but there is a massive amount of computing work that is all integer. FLOPS is not everything and x86 and x86_64 is perfectly good for that work, if not the best solution, certainly price performance and installed software base.
Oh and that $2000 number? I can buy a dual processor Xeon or Opteron node at the close to highest Core frequency and fasted FSB in addition to disk, motherboard, ram and 1U case for less than $2000. Why would I pay $2000 for a single CPU machine? That is easily twice as much for less performance when you factor in that I get to have twice the processors or more.
3.92 * 2 vs 4.86
Is there even a contest? That and your numbers use obsolete processors, my newest xeons are 3.2Ghz with 800Mhz FSB, both features very valuable for my application. Compare that $2000 Itanium against a dual socket dual core x86. never stacks up.
Itaniums are not popular because the numbers just never add up when you look at practical deployments. That is even before you factor in the massive redesign of every application you run to even get close to that theoretical capacity of the itanium.
Anyway I have no idea why this was even posted, the itanium ship has come and gone. Dual and multi core X86_64 is the future for the next few years in the computing I am involved in. Beyond that, unfortunately my crystal ball is out of batteries so I will have to get back to you.
Well, after you take the "Compilers" course maybe your love for IA-64 will have, uh, dimished a bit.
Maybe the "C" student think so, those just taking the class because they are required to do so. The "A" students and those with an interest in compilers often think the opposite. Here is a challenge, here is an open area for research, parallelization is the future and here is a CPU that fits in with that concept.
Maybe it's not a new idea but since virtually all articles have the ICOS post, does it make any sense to give links to the original site?
IMHO we could just as easily CTRL+C CTRL+V whole articles here. Slashdotting would be an yesterday's nightmare.
(Yes, I know, in the Free World we could highlight the text with the mouse and press the middle button just as easily)
The Itanium, had it been an x86, would have been the next logical step from the Pentium. Microsoft's weak support and developer reluctance to switch instruction sets, have killed all hopes of application support. Only a select few are adventurous enough to support Itanium running on esoteric UNIX distributions.
I've been *HOPING* that Apple would use Itanium on their high-end Power Macintosh. This would have given a second wind to the chip's lack luster success. Apple really is the only company that can give Itanium application support. If they decided to support IA64 *AND* x86, they could offer a high performance option along with their budget Macs.
I'm saddened by the end of the G5 line, but I'd shut up entirely if it were replaced by the Itanium. That chip is just fast... and elegantly redesigned. As I see it, it's the best chip out there.
Apple is great at supporting odd architectures. Their O/S already supports multi-platform binaries... c'mon... somebody petition Apple to add an Itanium build option in X-Code... PPC, IA64, and x86 all at once. I'd be so happy.
even if the ia64 system is faster then the x86_64 system it's not nearly enough faster to account for it's higher price.
especially since most of the target users that the poster talked about are going to be useing a lot of systems in a cluster anyway the single unit performance is far less important then the price/performance ratio.
Chips aren't really the limiting factor in supercomputer performance overall. You could custom design a chip to get crazy performance for the linpack benchmark that the top500 list is based on, but that doesn't mean it'd be worth squat outside of that.
The interconnect is probably at least as important to getting good performance (with today's software at least) as the processor is. Disk subsystems are probably about half that important. (and at least here got neglected in the initial setup)
How many itanium2s are in the top list? Are they making waves with their performance/price ratio? What about their heat when you're talking trying to build ultra-dense supercomputers? (Hint: spreading out is very bad for your fast interconnect) What does all that heat equate to in terms of cooling and then total power required?
Slashdot Patriotism: We Support our Dupes!
This is a very good point. While the WWDC demo was done on a Macintel with a 3.6Ghz P4 (x86), the developer machines from Apple are also x86 boxes, and the first wave of machines next year are supposed to be Pentium M devices (x86), it's curious that the compiler options in XCode are PPC/Intel and not PPC/x86.
What caught my eye in the article linked at the top of this post is that the price premium for Itanium is supposed to disappear sometime in 2007, about the same time that Intel processors will be appearing in PowerMacs and XServers in lieu of PPC970's. Maybe it's just a coincidence and the PowerMacs and XServes will have 64bit x86 processors. Or maybe they will run Itanium?
And then there's the expectation that iPods will be getting XScale processors which will be more than fast enough to decode and play video... which means that if the iPod at that point would play audio and video, show pictures, and include a cell phone, you've basically got a handheld computer in your hand, similar to Compaq's iPaq or Dell's Axim. Perhaps we'll see "OS X Embedded" and the requisite SDK's before long???
Maybe not, but in lieu of real announcements from Apple I will GUARANTEE that we'll see millions of "maybe if" scenarios all over the web...
the mighty Alpha would fair against current CPUs, provided that it used modern technology (i.e. 90 nm silicon fabrication, 3 MB SDRAM for cache, a 333 MHz and beyond bus...)
Where have YOU been? This patent law has been big news, and covered here before. The entire EU was out to enforce software patents, and basically hurt open source software from being developed freely in Europe. I guess blogs/opinions and random screenshots are now 'News for Nerds. Stuff that matters' as the /. editors sure accept them as news. Slashdot used to be a place for good news a long time ago.
If not I think you better go watch The Life of Brian and come back to remove the egg off your face.
My particular love for this joke is irony of looking into a room of Monty Python fans and they are all sitting there chuckling and repeating that very same line.
Yes, yes we are all individuals!
Classic!
What sound does an Itanium make when dropped?
DUD!
...we do the same thing with the Catalyst 4000 in our Cisco lab.
:p
This sig rocks the casbah.
I think the biggest thing that doomed the Micro Channel Architecture (MCA) was the fact that IBM did not bother to license the technology at very low cost.
If IBM had done a proper job of licensing MCA at a low cost then not only would MCA have replaced the old ISA bus, but alternative bus connection architectures like EISA, VL-Bus, PCI, AGP and PCI Express would have never happened! This is because we know now that MCA could be easily expanded all the way to 64-bit bus connections and support very fast bus speeds. Indeed, if you look at PCI and PCI Express connectors they strongly resemble the old MCA connectors in physical design.
"The reason the P4 clocks so high is that it needs too, the reason it needs to is it was designed that way, and the reason it was designed that way was so it could clock high."
Oh sweet vertigo!
HP came into our large site and told us all that RISC is dead. It has gone as far as it can go, but no more. Their own PA-RISC will be phased out, but its okay, because nobody would want it in a few years. SPARC is completely worthless because it is an old fashioned RISC processor. Embrace the new Itanic CPU like our (then, before she was fired) CEO has! (Followed by the parade of 100 slides.)
We thought it was silly that HP was throwing away a perfectly good platform and throwing all their customers over to the Itanic commodity CPU. Which, of course, wasn't a straightforward transition because there was no binary compatibility. But there were a lot of vendor promises. (KEYWORD ALERT! "vendor promises" is a tip-off that something is wrong.) And frankly, we really didn't believe them. It sounded more like marketing / sales / VPs were driving the change and the technical crew were taking a back seat, but being called to duty to try to come up with justifications that their customers could buy into.
That was about five years ago. Fast forward to today. The skeptical audience was right, and the sales droids and their vendor promises were wrong. And the sales numbers go to show that you still can't sell a technical audience with just marketing propaganda.
I think my summary (epitath?) for Itanium would be, "A processor platform designed for computer company VPs but (un)sold to a technical world."
OH FOR MOD POINTS
Thats interesting take on the market.
Single processor itaniums for under 3000$v er.php?cat=352
http://www.siliconmechanics.com/i2190/itanium-ser
Montecito, coming in 6 months. Running 2 cores >2Ghz frequency, AND adds multithreading to micture at sametime, takes the itaniums worst performance hurdle in those integer apps that itanium cares about by making a dedicated 1MB L2 instruction cache for each cores.
Now I'm doing some code generation stuff. Itanium is great for code generation target it has lots of facilities to get great performance out of it. However, there exists a LOT of ugly spagetti code that is given to compiler and say compile this and give us performance. The ugly spagetti is slower that properly designed code on *ALL* platforms however it hurts itanium most by not being able to guess and do out of order execution, and lower clockspeed. And based on what I've seen on university students coding I can say that the problem is growing all the time...
Now I've found that Itanium is great target for the stuff that I make a code generator for.
Emacs is good operating system, but it has one flaw: Its text editor could be better.
few people like paying $1000+ for a cpu alone, for example.
No one WANTS to spend that much on a CPU, except for those who want bragging rights. I think the point that the parent poster makes is valid. Very few people would complain about getting the top of the line AMD64 FX-57 chips - there would be more "oohs" and "aahs" amongst the /. crowd. In fact, those chips go for over $1100 each at Amazon or New Egg
FYI, the Athlon 64 X2 4800+ is also going for over $100 as well at New Egg
For anyone with the need or desire to have a fast top-end CPU (or plenty of cash to burn), $1000 isn't a big deal.
Why Doesn't the Itanium Get the Respect It's Due?
It does
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
When Itanium was up and coming it was expensive. This in itself is _not_ a problem for Intel as up till then they'd been making a bucket of cash on every x86 CPU they sold. I imagine they thought they could do the same with Itanium.
However, AMD began to challenge in their cash cow (x86) and the clock rate race started. Itanium would never be able to compete on clock speed but the battle lines were drawn. Ever since Intel has pumped oodles of cash into getting the P4 as fast as possible. Unfortuantely the P4 really isn't that great as a CPU, but it keeps up (with a bit of kicking) and I guess that will do.
Also AMD realised there was no immediate reason to push the 64bit desktop (no software!) and so just didn't bother to compete on that ground at all, this probably saved them a fortune and gave them a few years break to re-equip x86 with 64bit addressing. Which TBH isn't that hard as there are so many operation modes on x86 anyway! 1 more isn't gonna kill it! They did do it very elegantly though. AMD have built the worlds first truely defacto 64bit CPU... well... for the next 10 years anyway.
Intel already have the Pentium M but don't seem to be sure on how to sell it. It has very good IPC stats but a low clock rate. Oh well. Everyone's gone to some sort of PR value... even Intel... kinda... dual and quad core are going to help confuse the PR ratings considerably.
Time flies like an arrow. Fruit flies like a banana.
We chose *never* to move to an Itanium-powered Tandem. We're now an AIX shop instead. IBM's HACMP is just as rock-solid reliable as Tandem/Himalaya/NonStop/Whatever...
1. Power consumption,
2. Power consumption, and of course,
3. Power consumption.
Replying anonymously because I already modded in this story. Glad you brought up points 1 and 3, because I work for a company that builds custom Linux clusters for Los Alamos National Laboratories and other government agenCIes who shAll remain nameless. They use Opterons exclusively, for reasons 1 and 3. When you figure out how much it takes to power and cool 100+ node clusters and how many more compute nodes you could buy for that money, the reasons are obvious. We had our Intel rep come to us a few months back, asking what they could do to get us back to selling more Intel chips like we used to, and we said, "Cut power consumption." Number one concern, not FLOPS or anything else. Cut power consumption. Cheaper would be nice, too.
The P2 and P3 (and even the Pentium M) are largely unchanged from the original PPro.
P2 = PPro + MMX
P3 = P2 + SSE
There are other differences (cache structure and speed, bus signaling, addt. transistors for branch prediction, die shrinks) but it's all the same block diagrams and stuff.
P4 was different... the whole netburst thing. Wow did that suck.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
rank processor ghz (gflops / #procs) speed
#5 ppc970 2.2 (27910 / 4800) 5.81
#7 itanium2 1.4 (19940 / 4096) 4.86
#10 opteron 2.0 (15250 / 5000) 3.05
#20 xeon 3.06 (9819 / 2500) 3.92
Given this, consider what a 2 or 3 Ghz Itanium could do.
Your comparison doesn't show the whole picture. You are listing clusters with 3 different interconnects. If you want a true comparison you need to compare clusters with all the same interconnects configured the same way. Also you are listing different architecture styles. The Opteron solution is a Cray with their own interconnect and specialized motherboard hardware.
Maybe the XT3 (or whatever it is called) sucks in comparison to myrinet or quadrics. (I wouldn't know)
Also with Xeon and newer Opteron processors proper cooling is an issue. If you aren't cooling the cpu properly it will slow itself down limiting performance.
that $1000 is the _bottom_ of the itanium line. $1000 barely gets your foot in the door with itanium. $1000 is _top_ of the amd64 line.
the difference is like night and day.
Just speaking from experience. I ran 4 dual itanium 2's with 16 gigs of ram each. It took my group months to optimize the code, and in the end they were slower than a 2.5G pentium IV, and the only advantage was the extra memory addressing ability. We trashed the itaniums in the end, and purchased 2 quad opterons with twice the memory for half the price. THE OPTERONS RUN ORDERS OF MAGNITUDE FATSER. I have benchmarks that prove it. My group was doing all floating point operations. This is all done in 'c' code compiled directly on the platform that it's used on. This was using the same version compilers and redhat linux. The itanium looks good on paper. That's about it. The only people that I know personally that speak highly of the itaniums are the people that have never wrestled with them.
I could get a single-processor Itanium system for "around $2000". Or I could get a system with dualie 2GHz G5s for $2000 (less, if I'm a student).
:)
According to your stats from the "top 500", a single G5 2.2GHz has a bit more power than a single Itanium 1.4GHz. So the 2.0 G5 x2 should be almost twice the processor, for the same price. Yeah, the Powermac is a desktop machine -- but it's a kicking one -- and the article you linked said that a "basic" 1x itanium machine "will go for just over $2000".
Other advantages: while the future of the PPC is not entirely certain right now, I'd venture to say that Itanium's future is even shakier. Dualie 2.7GHz G5 systems are already available (at a $3,000 pricepoint from Apple). I see the power/price ratio continuing to look better for PPC than Itanium for a while.
Also, there's considerably more software out there for PPC than Itanium, and probably more developers familiar with the platform.
If you want more reasons why Itanium deserves almost every slam it's ever gotten, ring me back -- or read some of the other posts here
Yeah, what kind of assholes wouldn't report on that?
This is exactly what I'm investigating at the moment. I'm trying to find out how to get the best performance for an I/O bound database. We have tables with 10s of millions of rows that must be joined with other 10s of millions of row tables. The datbases vary between 50GB and 150GB.
...it's just that your system lacks the bandwidth.
I'm guessing you have a 754-pin A64... you've got a crippled bus from your cache to memory.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
It was way too expensive when it shipped, and is still too expensive. Even if I wanted to buy one, how do I know it'll work well and reliably for our needs? Most people have never seen an Itanium. Commodity processors get a lot more public use, run cooler, and are better supported by most software. After reading some other comments, I'd sleep easier with Celerons powering our servers. Besides, most performance problems can't be fixed by throwing a faster cpu at them.
I remember the first Itanium demo I went to. It was in conjunction with a software vendor deminstrating their SQL server software for the chip.
You could not hear the presenter because of the noise the CPU fan was making. He had to move closer to the audience so we could hear him!
And that is a big part of the problem. Coming out with a power hungry CPU just when power became a huge issue was not a clever thing to do. AMD decided not to follow itanium for precisely that reason and said so at the time. AMD64 was their choice and the rest was history.
Tim Josling
Furthermore, having just tried to order a pair of Itanium 2 servers for a SQL cluster:
These supply issues quickly gathered the chip a new moniker - Unobtanium.
that $1000 is the _bottom_ of the itanium line. $1000 barely gets your foot in the door with itanium. $1000 is _top_ of the amd64 line.
True, but the Itanium is pretty much stuck as a server/workstation chip these days. It's still pretty pricey, but compared to the price of a decently laid out workstation or server, it's expensive, but not TOO bad.
I'm not too surprised that the chips cost that much, considering how much the projected market has shrunk.
Personally, I don't disagree with you - I wouldn't want to spend more than $500 for ANY chip. It's just that for certain specialized fields, that $1000 chip might be worth it.
Main reason the opteron is getting so much press? its x86 compatable, end of story.
People would love faster proccessors but don't want to have to pay a pack of patroniseing programmers to convert all their code to it (or in most cases beg and plead to their closed source developer to please make it for the new design)
...
If it was anything like my EE class (1998 era) they were also handing out Itanium architecture manuals, Itanium platform reference guides, a book about Processor Architecture written by a guy from Intel, taught by a professor who drank out of an "intel" mug, and the cute female grad student had an "intel inside" t-shirt on with an arrow pointing down. Yeah, I had that class.
If you substitute 1996 for 1998, Alpha for Itanium, DEC for Intel, and change the T-Shirt on the cute female grad student it sounds like my graduate architecture class. Are you suggesting there is something odd here? Well, other than the girls.
look at the "available" date, those systems don't ship yet! hahaha, vaporhardware! I have two 1.5GHz Itanium2 boxes at work, they run RedHat Enterprise Linux & Oracle about 15% slower than my home 1.5GHz Xeon box. Why someone would pay 2.5 times the money for same performance in most business/serving apps is beyond me.
isn't the cell processor similar to the itanium2 (please make corrections)? just find it weird that everybody seems to think that cell next generation of processors for everyone but suffers things that are the same itanium 2 which is criticized by the same people. anyway, just a thought.
Live your life each day as if it was your last.
...and businesses don't like to get locked into a single vendor. Intel got too greedy and anti-competitive, and that's a big reason why businesses aren't embracing the Itanium.
I've read a fair bit about the architecture, and I've done a decent amount of programming & optimization for a VLIW CPU. So, let me correct something: THE ITANIUM IS NOT VLIW. It is VLIW-like, in that dependencies between instructions are explicitly encoded. However, the scheduling of the instructions is not static, so it doesn't place as much a burden on the compiler as a traditional VLIW.
Other benefits include that code doesn't have to be recompiled from one generation to the next, yet it can take advantage of different amounts of parallelism, across the family. Also, it allows you to do branch prediction & speculative execution, or interleave instructions from different threads (i.e. hyperthreading).
On the down side, dynamic scheduling forces the chip to contain scheduling logic. The ever-increasing size of this logic, for wider & more deeply-pipelined superscalar CPUs is one of the main points in favor of VLIW.
So, Intel's EPIC (Explicitly Parallel Instruction Computer... or something like that) is really closer to CISC/RISC than VLIW.
It's or its? I think it's its. Duh.
1) "It's an Intel processor, so it has to have great x86 compatibility and performance. If it doesn't run that well, it's a failure."
Why doesn't anybody rap an IBM Power processor (or a HP PA or Sun SPARC or...) for having NO (not just slow) x86 compatibility? Itanium is the only non-x86 processor (one with a different native instruction set) that does ANY native emulation of x86 -so does this win praise from people who like Power but wish it supported x86 as well as it's native instruction set? STRANGELY, NO.
2) "It uses an instruction set (EPIC) that only Intel makes so no competing processors can use it. This is very bad from a company with as big a market share as Intel."
Well, Itanium is not (not yet anyway) a big market share leader in the big enterprise space that it's targeted at...like IBM Power. Of course IBM would sue your ass off if you tried to independently build and market a Power-compatible processor. The truth is that in the high end of the enterprise computing market, proprietary is the norm and Intel is no worse or better than any of the other players in this space. If you don't think Intel SHOULD NOT be trying to make processors that compete at the high end, fine, but that's what IS where Intel aims this processor at. (Anybody think that AMD WOULDN'T do their OWN exclusive high end processor and unique instruction set if they thought they could sell enough of it to make it work?).
3) "Itanium is hard to work with, being so different with the VLIW instructions, so dependent on compilers and so darned parallel the compilers excrete binaries in structures called 'bundles'. Why didn't Intel just build a better x86 instruction set processor?"
It always facinates me that you can sometimes hear this comment voiced in the same conversation that somebody else claims there are no major innovations happening in computer architecture. For better for worse, Intel stuck it's neck out and did a very elegant highly parallel processor design that is not CISC or RISC. And does Intel get credit for spending a ton of money to bring to market something that ISN'T just a tarted up decendent of the 4004 cpu from the 70s? You answer.
(There's lot's more to say here, but I'm out of time...)
It gets no respect hey?
"Back to school!
FalconShould there be a Law?
A better question is why doesn't the Alpha processor get the respect it deserves. Yes, HP is trying to kill it, but it still trounces the Itanium 2 even today. Alpha is the fastest, flat out. Ever wonder why you can't find Alpha vs. Itanium 2 benchmarks on hp.com? Because Alpha is faster and hp is trying to bury it, so it doesn't publish the numbers. I would know, I used to work in the High Performance Technical Computing division at HP.
HP is currently "giving" away an Itanium server for attendees of a $2000 HP and Intel Developer Workshop - of course only in the USA (always the same pattern, do not call it fascism). http://www.theinquirer.net/?article=24502 The regular price for such a server is more than $3000. What you get is a low-end server without the extras for a workstation. Oh, and HP recommends Microsoft Windows XP. Do you recommend Microsoft Windows? We have seen such deals before. A few months later, let's say one VMS revision later, and the free server is no longer supported. The quotes at pricewatch show why a regular low-end Itanium system costs more than $3000. Did I mention that not all of these Itanium systems run all operating systems? Do not expect that all Itanium systems are compatible. And which operating system would be most affected? VMS, of course. But the biggest problem is the lack of CHEAP low-end systems. Who is seriously going to spend money on a $3000, or maybe $2000 Itanium PC? Nobody. A PC costs $500, no more. Why would anybody want to invest into two different platforms for workstations and servers? Performance? Just tripple the number of low-cost x86 servers and you still end up with lower costs than for any Itanium system. The bottom line is that without cheap workstations and notebooks there will be no growing Itanium market. Intel made it pretty clear that this will not happen. Look for the AMD64 series instead!
I think the only thing that really mattered was/is cost. Until you're making IA-64 desktop workstations for $700, it's just not that interesting to the masses. Compilers aren't as strong as they could be because compiler writers aren't working on IA-64 compilers because they aren't selling because of cost. 32bit IA-32 support is nice, but when it's all said and done, you want IA64 code. Because it's expensive and nobody buys them, MS doesn't spend the several hundred million dollars to keep Windows running well on it and get the Office Apps ported.
It's kind of interesting, Intel timed it really well to kill off sparc, alpha, PA-RISC, Mips and potentially even POWER. They essentially did kill off most of them, Sparc is kind of limping along for now and IBM's still in the ring. On the other hand, they also made kind of a big blunder and they hurt themselves by hurting a lot of that competition. Why would I spend $100k on an IA64 solution that hasn't be proven in the field rather than a Sparc or POWER solution? Or even better, spend $30 on Xeons and Linux.
Really to make it happen, HP is going to have to move a lot and start to hurt IBM with them and the cost will have to drop substantially. As is the x86-64 market is building up, being 64bit alone isn't a selling feature, people don't care about RISC vs. CISC vs EPIC and there is simply no way that small and midsized companies are going to pay for the relatively small performance improvement.
As an avid necrophiliac, I concur.
I have a hard time taking the Itanium serious for two reasons:
1. I have heard there are thirty-six megs of cache on one version of the chip
2. When the original Itanium was released all those years ago, its hardware x86 emulation was so terrible that Intel wrote a software emulator that was faster than the hardware solution.
Don't get me wrong, the first isn't a bad thing(I just find it a tad rediculous), and the second no longer applies. More seriously I think that its completely incompatible instruction set combined with high cost keep it from more widespread adoption. Even if the supercoputer people are getting all hot and bothered about it, that still leaves a few fractions of a precent below one hundred of the world's population that the chip has to do something for.
Oh, a lesson in history from Mr. I'm my own grandpa.
Intel initially did a superb job of marketing the ia64 processor -- they scared the bejeezes out of DEC, HP, and SGI. So much hype from Intel, in fact, that each of these companies which already had 64-bit processors and a market for them, has abandoned their processor in favor of Intel's marketing muscle.
Intel then pulled the plug on much of their ia64 development "roadmap", apparently having created the desired disarray amongst their 64-bit competitors. To top this off, Intel's own marketing bullshit over processor clock speed meaning more than system thoughput shot themselves in the foot.
With no backward compatability to their ia32 architecture, little regard for seeding low cost development platforms to gain market share, and their own internal battle over "processor clock speed rules all", it is really no wonder that the ia64 has not gained widespread acceptance.
Reuse of code from previous generation processors is important -- IBM's PowerPC processor, SGI's MIPS processor, and SUN's UltraSPARC processor all offered backward compatability.
Intel's ia64 processor doesn't even garner respect from Intel. Why would any other computer OEM offer more respect than that?
should have been "mid to upper end small business market" Numbers wise Intel/Microsoft still dominates. If SUN can take a bigger bite out of M$'s SQL Server market then they will be set. Now, Linux/Mysql is starting to show big there. Why not, do it on Ultrasparc?
Sorry about the writing. Robot fingers, you know? Cliff Steele in DOOM PATROL #23
Ten years ago you could buy an Alpha that had the same bragging rights - 64-bit registers & addressing, wider bus, better atomic primitives, etc. *YAWN* It solves a serious problem that confronts maybe .05% of the world, and runs about the same amount of sw compatibly. A technological masterpiece/kludge in the tradition of IPv6.
You can't fight in here - this is the war room!
Sad part is x86 is still cheap, and MS still can't handle 1bit of competition. Can't we just have Alpha back plz?
The 286 was the last 16-bit processor; 386 and 486 were both 32-bit. The crippled 386SX had a 16-bit interface, though it was internally 32-bit.
Somewhere in the late seventies, a company had a bright idea. Instead of asking the developers to recompile their source code for the new processor, they would give the new processor at least a mode wherein it would simply execute the binary code of the old processor.
This idea made them big. This idea made them the biggest. And they are still called Intel.
We were part of a pilot program at our school on the Itanium. I think the Itanium tanked because Intel didn't plan the release correctly.
.03$, from hanging around the Itanium team at school.
Remember the Itanium is a brand new architecture. It's NOT x86. It's like jumping from x86 to PPC if you will... That means you have to recompile everything. Of course, I'm not talking about compatibility mode.
Of course, introducing a new architecture is a *huge* endeavor. Intel wanted to fast-hand it and their marketing department went nuts over it. Problem is, the Itanium *wasn't ready*. You can't introduce a new architecture and expect the market to gulp it in. They were trying to push x86 out with the new chip. That couldn't happen. How long has x86 been around? How many *billions* of cash rest on that chip?
Anyhow, the first Itanium was only a technical preview, in a way. It was the first public diffusion of the chip, to try to get developpers used to it, slowly.
However, there came another problem: Itanium is an EPIC (Explicitly Parallel Instruction Computing) architecture. Not RISC or CISC. Do *not* listen to their marketing BS about EPIC. Technically, it kicks ass. In practice, you need a good compiler to take advantage of that instruction set.
That was the core of their problem. The compiler. Compiler gurus will tell you it's devilishly difficult to optimize parallel processing correctly. The cool thing about PPC chips is that it manages some parallelization on its pipelines internally. The compiler just doesn't have to care. It's much more difficult on EPIC, because of the E: the compiler has to manage the parallelization itself.
The chip arrived, but the compiler wasn't there. Intel relied on universities to develop powerful compilers. Universities are great for research. This is fine, if you have the time. Intel did not have the time.
At one point, I think, they tried to push the GCC guys to work a bit on optimisation for the Itanium. The answer was no. They had spent too much effort on optimisations for the x86 platform. x86 has been out for so long, and they're *still* optimizing. If someone wanted to do some Itanium optimisations in GCC, go ahead, but not them.
One important problem with the compilers, as another poster has pointed out his experience, was memory access. The Itanium is supposed to have a *killer* memory subsystem. It just wasn't used correctly.
So in the end, it boils down to this: the Itanium arrived before its time. Its time will be when good optimizing compilers will exist. I believe the NCSA is working on one that could fit: OpenMP, though its applications are primarly supercomputing.
Intel tried to push the processor too hard too quickly. It tanked in trying to replace the x86 market. It is now releguated to niche applications, like supercomputing. Maybe its descendants will come back to the front of the scene when the compilers will be good enough: when GCC will have optimisations for it.
Just my
Misleading titles? Inflammatory blurbs? Keep in mind that Slashdot is a tabloid.
Itanium gets no respect mostly because of the attitude of HP...
Forcing their customers to drop Alpha or PA-RISC for itanium, which can be a costly transition... And expecting Alpha users to migrate to HP-UX which lacks many of the features Tru64 had..
Had HP put money in to developing the Alpha instead of itanium, then the alpha would be at 2-3 ghz nowadays, and would still be beating the power chips from ibm in raw performance..
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
HP has no where to go *but* Itanium for its small-SMP, Superdome and NonStop lines. That article noted they're investing $3 billion in using that chip over the next few years.
And Microsoft does support Windows Server 2003 on Itanium, that article was referring to client XP (which makes sense -- there is little/no need for an Itanium workstation at this point.)
I can get HP-UX, Linux, or Windows Server 2003 on HP's Itanium2 machines (I've had several running HP-UX and Windows in a corporate lab).
-Stu
I've built designed databases and/or application server farms for coporations in telecom and finance.
If you need 64-bit addressing (i.e. large shared memory sizes, or individual process heap sizes over 2GB), you basically have 4 choices: Opteron, SPARC, POWER5, and Itanium2.
If you take a quick look at the SPECcpu2000, SPARC is a condenter if you need throughput, but not if you need raw CPU power. POWER5, Opteron and Itanium2 are about on-par, with minor (~10%) differentials between each other, in both power and throughput.
So really, it comes down to what hardware vendor do you want to do business with, and what software support is on each chip/OS, and that determines the chip you'll buy. Most corps deal with the big 3 (IBM, HP, Sun). HP chose Itanium2, Sun chose SPARC or Opteron, and IBM chose POWER5. Some systems are only tier-2 support on AIX POWER5, for example, but are tier-1 on Itanium2 HP-UX. Some are the other way around. Go figure...
-Stu
Are you saying that those applications specifically are running slower, or that YOUR application that uses Oracle and runs on REL is 15% slower. If it is your application, did you build it with the Intel ICC compiler, or did you use GCC? If GCC, then there's your performance hit right there, and then some.
Actually, I did some research, and I realize I wasn't completely on the mark with my comments. I think the big two chip companies are moving somewhat in the right direction, with dual-core chips, though there is certainly a need for low power Transmeta chips for reading Slashdot, or at least the big players should drop the clockspeed to 1 Mhz when the CPU is idle, and only crank it to 5GHz when needed.
As far as off the mark goes, did you know that the human brain consumes humongous amounts of power? More exactly 20 Watts out of a 100 Watts resting pace, so even though it makes up only 2% of the body's weight, it consumes 20% of total energy/oxygen needs? So even this ever-perfected life runs the cpu "hot" relative to the rest of the system.
Also, by 2020-2050 we can expect real artificial intelligence if things keep up they way they do. We better start thinking about the consequences of what happens when machines are smarter, more intelligent, can hold wittier conversations, and make better supreme court judges than human beings. Is that something we can deal with, or trust?
For anybody in the need, I source my Itanium 2 1.3's from a guy (packageshop@hotmail.com) for only $699! Where else can you get 3mb on chip cache for that kind of money at that price...! I've only noticed a marginal speed bump with the 1.4 and 1.5's which are priced astronomically much more, so for price/performance tricking out a box with the 1.3's are the way to go, and their price is now very affordable. Now you can get them for a steal and these little babies are screamers. I run a bunch of tricked out servers maxed out with four 1.3's each and the main cool deal with the Itanium 2's, was you can blow right on past the 4gb barrier and it executes code in big chunks. Working with any computer with less than 4gb and 4 processors and you don't know what you are missing. These badboys have 28 dimm slots or 64 dimm slots and go up to 56gb and 64gb respectablly, node scalable up to 224 gb. Show me that motherboard of yours again with 2 dimm slots that goes up to 2gb??? Um, yeah, where's the trash can. You want one of these... its all about memory. Itanium 2's running stuff in solid state memory with 3 caches between it and the memory totally obliterates everything else out there short of maybe Power5's, which are actually 4 processors on a chip and will cost you a small south american country. You do have to flash upgrade your EFI bios with the lastest to enable EPIC, and the machines work best with lots of RAM because they execute code in chunks. The issue of running 32bit code has never mattered to me, as the entire OS was recompiled for 64bit anyway, and if you're using linux everything is in the distro you need as a 64bit native app ready to go. I run Centos Linux which is free, but have also used Redhat ia64, Suse Enterprise ia64, and the latest 64 bit Windows on them. The 32bit enabler was installed by default but I never used it, except under Windows. Most people gave the Itaniums a bad rap, because the EFI didn't enable EPIC by default, and all that capability just was left bridaled and brimming there under the surface. So be sure to flash your EFI bios to enable EPIC. Also, I always get a better video card by default for my server, most servers ship with the bare minimum for video. I won't even touch the Opterons or Xeons, they are kludges by comparison. Once you figure out the little tricks on how to make the ia64 chips sing you can do some serious number crunching. Wikipedia still lists the Itanium 2 as Intels top dog processor.
Having half our mail servers run on Sun (SPARC) and half on HP-UX (PA-RISC) meant that a problem with any one of the components could only take out half the complex, and we hoped we'd fix that problem before a different problem took out the other half.
It worked, but with most of our vendors jumping on the Itanium vaporware wagon, and the remaining non-jumper screwing us on service we got very nervous about being able to continue.
Turns out we were lucky. The scaleability and reliability of Linux on several processor types has made our tradition of two OS vendors for any major subsystem moot, and we can still split architectures between i86, Opteron, PowerPC, and even re-use the PA-RISC and SPARC gear that we have not lease returned or sold off.
not my apps, standard Oracle diskless TPS benchmark is slower. I should be fair and say that the floating point and linear algebra benchmarks are smoking on Itanium2, but that's just not what usual business apps do, it's all bcd/integer, memory shuffle, and such which show no benefit
Hmm... Didn't Transmeta go with VLIW too? Maybe Intel should just buy them and implement their on-the-fly instruction translation software (code morphing) for the Itanium?
But it's probably too late to save it now, anyway...
Cray uses Opterons. Such as in this product.
If I understand correctly, you're referring to the off-chip L3 cache (which is considerably slower than the on-chip Itanium one)
The Raven
Everything you said, plus:
The overall effect was a loss of credibility, and that a number of hardware companies were either weakened or went under. So don't be surprised if the Itanic gets some derision.
The BIOS replacement they use is not functional. It's very difficult to set up disks for use, and if you lose the disk that the BIOS data is kept on, you're screwed.
Your EFI configuration lives in flash memory you stupid fuck. You are certainly brainless if you don't know how to set up an EFI partition (if you even need one that is: you don't have to have one) - Linux (Debian, Ubuntu, RedHat, SuSE) and HP-UX will all set these up automatically, don't know about windows or VMS.
The HP systems are riddled with hardware problems
I've been using an rx2600 for about 3 years now and it has been in absolutely perfect working order? We got an rx4640 a few months ago and it too has been just fine. Sure, two systems is hardly a representative sample, but would you care to give an example of what you mean by "hardware problem"?
I think you're just making up total and utter bullshit. You've never actually used one of these systems, have you. (Easy test: what's the logo next to the flashing green led on the baseboard of all HP Itanium systems? This is easy to see (even in the dark) and you don't need to open the system (just look through the grille at the back))
I've created an article from this comment on http://www.zbalai.com/_ideas/index.html I've quoted Jhan, please check, if the quote is OK.
In my opinion many of the comments explain why the Itanium is not widely adopted, but they mostly do not explain the admitted community attitude about the Itanium.
Specifically, yes the processor has a chicken-and-egg problem-- if it were popular, software would be ported and the cost would come down (or at least this is conceivable). So one would think our attitude would be,
Instead of the current,The post that (IMHO) maybe does explain the situation was this:
So maybe problem is that people want to feel good that their 3.4ghz processor is THE FASTEST thing out there, and the Itanium is there in the background denying this claim.
Some repeated themes that (though true) do not explain the attitude are:
-
Integer versus Floating point performance.
- Compatibility
This is the chicken-egg issue.
- heat/size
I don't know - could this be a checken/egg issue too? In other words,
if the chip were more popular, would intel spend more resources
trying to address this, or is it intrinsically large and hot?
The most frustrating part of the discussion to me was the conflicting statements about whether the architecture is potentially a good one. I don't have the background to judge, though I'm more inclined to give it a chance than some.The problem here is that, when purchasing a computer for personal use, floating-point (games) is in most cases the only issue. You are't upgrading to a 3.4Ghz processor to run Office faster.
Certainly running xyz.com's website and database are a different story (integer performance needed), but this by itself wouldn't prevent anyone from craving the itanium for personal use, any more that it would distract us from wanting a PS3.