Slashdot Mirror
Intel's Itanium 2: Succeed or Fail?
An anonymous reader writes "'Intel's most powerful processor ever has the ability to take on IBM, sink Sun, make or break HP, and crush or revive AMD,' says Fortune's David Kirkpatrick. But the 64-bit question is what happens to the heavyweight competition if Itanium 2 succeeds or fails?"
IBM rules high-end computing, the consumer sees nothing. They probably still buy Intel because they like the jingle.
I doubt the Dell server market makes much of a difference whether it is AMD or Itanium.
I do agree with the fact that we would see a rebirth of AMD, though I don't think it's really dead.
Sun might find some breathing room for SPARC, maybe a few saving graces for poor ole Sun who has been struggling financially.
The article's last mention is that HP ends its exclusive commitment to Itanium and uses some AMD chips. This sounds like a stretch, one gamble on a processor to stain a large business relationship?
--------
Free your mind.
Yeah, if intel's new chip is a hit then the company will profit more. In other news if intell gains market shares then AMD will not have those same market shares.....i should be an econmic analyist.
As Intel now loses its backwards compatibility, they also lose their biggest advantage. Sadly, the IA64 will probably lose out to less spectacular, but IA32 compatible designs.
Alpha tried to emulate the x86 earlier and failed. Sadly.
What exactly *is* the problem Intel has with manufacturing/designing Itanic? I always liked the theory.
Cheers,
-b
It'll never be the success that intel and HP envision for it and here's why. First, it's too hot and too expensive. Secondly is doesn't have any applications. I don't mean Gnome and KDE, I mean the sort of applications that big corporations run. Thirdly it isn't backwards-compatible with any existing architectures. You can't just take your binaries over and run them, at least not at full speed. Applications will need to be ported and retested. This is not insignificant in time, effort and cost. Fourthly, most people who want 64-bit in the corporate world already have it in the form of SPARC, Power, PA RISC and Alpha. Why should they change to an unproven, immature "jam tomorrow" architecture given their working and reliable systems already in use? I'm afraid intel missed the boat by about 10 years. If they'd brought out a 64-bit RISC at the same time as SPARC, MIPS, Alpha and Power they might have stood a chance. It's a turkey, and apart from a few niches (e.g. number-crunching super computers) it's doomed to failure. I don't even need to mention how Athlon 64/Opteron will eat its lunch in the commodity sector of the market.
Stick Men
wasn't the 386 16 bit? I thought all the 386 and 486 chips were 16-bit until the release of the pentium processor in the early 90's?
I think it will either sell like crazy or not sell at all.(Probably FAIL)
Water, water everywhere so let's all have a drink-Homer Jay Simpson
Intel should invest HEAVILY in improving both gcc and linux (so that it's comparable to Solaris) for Itanium. They need decent Chipsets out that support many of the things Sun does. Otherwise it will fail. They also need DB2, Oracle, and Websphere to run on it.
Otherwise it will fail.
Wait..I have heard that before....
Neck_of_the_Woods
#/usr/local/surf/glassy/overhead
Am I alone in feeling the Really Big Question is how much the Opteron costs? They've pretty much said the Athlon64 has to wait another quarter. So then, the "desktop" just has to wait, and its success really depends on the buzz the industry gives from the introduction of the Opteron.
But I just don't see much buzz coming from the Opteron, unless they capture the hearts and imaginations of that "Workstation" market they throw right in there with the "Server" market in their roadmaps. And quite simply, to do that, they still need to keep costs really low. Slightly more expensive than the Pentium 4, but WAY below the 2nd mortgage Itanium II.
Personally, the second i find out how much the Opteron ships for, I'll make a decision on buying stock in AMD for a long term investment. If they drop the ball on this one, their new "away from chip making" strategy doesn't inspire much confidence in this investor.
I've worked on 64-bit conversion projects for applications on HP-Unix, and it tends not be as trivial as it should be. I'd compare it to converting a 16 bit Windows app to 32-bit Windows. Yes, both should be trivial, but there are enough gotchas! On a per line of code, the Windows conversion was probably more involved, but then that was because it wasn't written as well - eg. assuming an "int" is 16 bits long.
RB
Um, Windows already runs on 64-bit hardware. If programmers use the typedef'd types instead of hardcoding pointer sizes, then the port should involve little more than a recompile.
The transition to Win32 was painful enough that the newer APIs are all written so as to make the next transition seamless. There may yet be a valuable crystal waiting inside the lodestone.
The proper plural of "Itanium" is probably Itania. HTH.
Itaniums are not going to be "flying off the shelves". No one in their right minds wants them when there are much more mature and proven 64-bit architectures out there. Linux is not going to take a huge up-swing. HP wants to sell the itanium to corporations as a 64-bit Windows box to fit into corporations' "windows everywhere" strategies. If you talk to HP about 64-bit UNIX machines, they will try to "up-sell" you an itanium Windows box. I kid you not.
Itanium/2 is a 64 bit processor. So it needs 64 bit software, including the OS.
Umm, no. For example I am running 32-bit Solaris on a 64-bit UltraSPARC. And applications compiled 32-bit.
Whereas in the case of Windoze, the 32 bit stuff (and even some 16 bit stuff) is built right in to the API.
Yes, that's why it's called the Win32 API. Work is well under way on Win64, but in Microsoft's ideal world, almost no-one will write to the Win64 API - they'll target the CLR, which itself will be 64-bit native.
Then the millions of apps that people use, right now an excellent way to lock customers in, are going to turn into a lodestone around their necks.
Yes, just like when Apple moved from 68k to PPC? Nope, wasn't a problem.
I'm sure Micro$oft is pissed as hell, but Linux is going to take a huge upswing when Itaniums start flying off the shelves.
That doesn't necessarily follow either. After all, Win 3.11 didn't fully exploit the 80386 either, and it wasn't 'til the first NT that Microsoft did.
It's already happening, you just haven't noticed it yet.
Jon.
You might remember the same situation when Win32s and then Win95 were released. It took a few months for most of the apps to be rereleased in a native 32 bit format. Luckily win16 was (and is) still supported. Such will be the case with 64 bit desktops.
There have been articles in the MSDN about porting existing code to 64 bit windows API for a while.
I've perused them and there's really no major learning curve. Most stuff will just recompile as is, except for a few pointer hyjinks and some more esoteric inlined ASM.
Thing is, the same problems can apply to Linux and other OS code. Sloppy code is sloppy code no matter the social viewpoints of its authors.
I don't need no instructions to know how to rock!!!!
I think the author has it backwards. He says if the new chip fails then blah. However, the more coherent argument is that if blah happens, then Intel's new chip has failed.
But, the author doesn't seem to realize that there's more than just out and out success or failure on the spectrum. It's more likely that there will be incremental change. Intel sells X units to A, B, and C, AMD sells Y to D, E, and F, and IBM, SUN, and co. sell to whomever. And things kinda ballance out.
All this new technology that's supposed to change everything dramatically, changes things to the degree that it's touted to. My money is still on evolution rather than revolution.
In mathematics, one does not understand things, one merely gets used to them.
--VonNeumann
Is it possible to have a motherboard with two processors, a P4 and an Itanium? The core OS could run on the Itanium and non-Itanium stuff could get executed on the P4 processor(s).
I'm sure this is a stupid idea that many other posters will point out the weaknesses of, but I'm wondering why it couldn't be done.
Aw heck, I've seen plenty of those same assumptions in OSS code as well. Assuming int is 16 bits, and the such.
Not everyone has the energy to type malloc(sizeof(int)*20)
Bad coding habits are endemic in the free, Free, and proprietary worlds.
Good code will 'just work', bad code will need fixing to work natively.
I don't need no instructions to know how to rock!!!!
Plus, less power consumption could mean thousands or tens of thousands depending how many servers you have. If you're google or some other huge site with thousands of systems, the power savings means lower operation overhead.
Well, the trend broke in 2001 when people started to notice that the machines needed for this generation of software was not the fastest but the slowest machines on the market. That most users did not need a top end machine and instead could buy the slowest processor out there. During 2002, the same came true for lap tops. Now everyone is swimming in so much wasted CPU power that it is going to finally crush those that can't adapt to radically lower needs compared to what Intel and their competitors are churning out. Ask someone who runs a computer room and they will tell you that they are quickly consolidating old servers that cost $250K three years ago to a server that costs $15K and only takes up a quarter of the room.
Intel is in real danger of not surviving because it does not understand where we will be in 5 years. 5 years ago when they were in the middle of this effort they did not see our need for speed slowing dramticly and are now producing a chip that has such a limited market that it will never be profitable with all the investment that was in put in.
When you look at how a company responds to the typical S curve of development, they may make the first curve but often that screws up their timing on the second curve and they just go off the cliff.
NT Alpha is both a 32Bit version and has been EOL'd (Although it lasted longer than NT PPC).
Windows 64 is due out Real Soon now, and it's delay is likely at least half the reason the Athlon 64 has been pushed back.
Itanium 2 is going to have to make up for the pathetic performance of the first revision (Which seems to perform on par with a Via C3)
"You've got an invalid haircut" -Warren Zevon - Life'll Kill Ya
sorry but there is a Error on the frount page
Already, specialized 64-bit chips have taken over the computer game industry--Nintendo 64 and Sony's PlayStation2 both use versions of IBM's so-called Power chips.
WRONG both are MIPS cpus and the sony chip has 128 bit instuctions as well if you want to get picky
regards
John Jones
64-Bit Windows XP Has been available since first quarter 2002. Windows 2000 Advanced Server LE is a 64-bit operating system. Intel and Microsoft worked very closely when developing these 64-bit OS's specifically designed for the Itanium proc. I would imagine that .NET will launch 32 and 64 bit versions in synchrony, especially since the Opteron is due to hit shelves very soon. I'm sure that due to the source-code nature of Unix that it will (as usual) be an attractive advantage for 64-bit servers, since there is an opportunity for some extra optimization at the compilation level, but hasn't this always been the case =] Nothing has changed.
If I may quote the author:
Don't pay any attention to this guy - he gets paid by the paragraph.
-Ryan
NT4 Limited Edition and Windows 2000 64-bit have been around for a loooonnnggg time. Just because you can't buy them at Best Buy (where most analysts appear to do their market research between customers) doesn't mean it doesn't exist.
Or maybe they'll make it run in 32bit mode with a 64bit compatibility layer, in this case Microsoft will have their marketing milk^Wconvince the customers.
Trolling using another account since 2005.
It is something of a question whether this change will open up opportunities for new software. I think it will. Think shared memory -- very large memory spaces being simultaneously updated and accessed by multiple independent processes and processors performing different tasks possibly for different users. The three drivers of technology are corporate databases, games, and pornography. Huge memory spaces with multiple processors attached have many possible breathtaking applications in each of these domains. Start coding.
The article doesn't really touch on why intel is so buddy-buddy with Linux (they've helped refine GCC and other important issues).
linux will always be best on intel CPUs, because they are the most available. linux is taking over proprietary UNIX boxes by Sun, HP, and SGI.
guess what, all those UNIX boxes used to have high-performance CPUs attached to them (MIPS, PA-RISC, etc). Now they are all going the way of the dino...thanks to Linux.
the more popular Linux is in the server room, the more likely Intel will be riding its coattails. And yes I know that Linux exists for other archs, but Linux/SPARC, Linux/PPC etc are always a step behind the Intel version.
(-1, Raw and Uncut is the only way to read)
I think the success of the Itanium not only rests on its technical merits but more importantly it rests with how much mindshare they can get for the product with the business people who, more often than not, end up making technical decision in a void.
I think that Intel is aware of this. Marketing can make the product. The best engineered solution does not always win out.
Is it just me, or has there been really no mention of the Itanium on a "consumer" desktop? It sort of mirrors the "Pentium Pro" situation of numerous years back...it was pitched as a server-and-datacenter processor, but it was years before it became the Pentium II. On the other hand, AMD's Opteron/Athlon 64 has been touted as a consumer piece from the very start. The consumer and "business" processors have been developed side-by-side, and their release dates are rather close. Is AMD the smarter or the dunce here? Time will tell. I, for one, am putting off any personal computer upgrades until 6 months or so after the A64 comes out.
With Linux and Java, the actual CPU used inside a box is close to irrelevant. This was the same fact that made the DEC Alpha irrelevant: every program that ran on Alpha ran fine on Intel, with the exception of OpenVMS and Digital Unix software, which were also the only markets where Alphas sold.
Today, the OS has also become a commodity item, and niche OSes such as OpenVMS and Digital Unix are dead or nearing death. A hot expensive CPU cannot capture a market when it has to compete on a level playing field with cheap CPUs that run the same software can can be easily clustered or SMP'd to get the same performance.
The only way to break into a saturated market is to cut prices... does Itanium do this? I don't think so.
They may sell a few for the gadget hunters. But the notion of a CPU competing with IBM is so funny it's almost hilarious.
My blog
Alpha didn't have support for x86 anymore than PPC has support for 68K. They just had a good emulation system in software. Technically, "emulation" might be an understatment; they would dynamically translate the instructions, doing the work once for each block of code so programs ran faster the longer they ran.
Personally, I think that it was bad marketing and the loss of support from Microsoft (probably as a result of poor sales) that were the problem, not technical issues.
But I have a hard time envisioning the scenario in which my porn collection would require 16 billion gigabytes, and I'd want to view it all at once.
--Jimmy has fancy plans; and pants to match.
Umm, no. For example I am running 32-bit Solaris on a 64-bit UltraSPARC. And applications compiled 32-bit.
So am I, but my understanding is that such a strategy will NOT be available to Itanic users, or will come with a heavy performance penalty.
That's AMD's big selling point: Their 64-bit chips run existing 32-bit apps better than existing 32-bit chips, but Itanic doesn't.
Everyone seems to be excluding IA32 based servers in even relatively near term projections. Given the price/performance of such systems, surely they will continue to be a factor for a good while yet, especially if clustering takes off.
I remember reading an article a couple of years back where someone said that itanium's biggest problem will probably be the pentium.
There's another little-considered thing about IA-64: It's the most proprietary major CPU on the market. AFAIK, every one of the major CPUs has some form of cross-licensing or functional cloning in place, except IA-64. (Actually, I don't know about HPPA, but I'm sure there's some cross-licensing of technology through HP's IP agreements.)
It's not because of market positioning, either. It's not something that will come on as soon as IA-64 succeeds.
It's because Intel and HP set up a company specifically to hold the IP of IA-64. Intel and HP don't hold any IA-64 IP themselves, they get it from this company. That way, the IA-64 IP is not covered by any agreements of Intel or HP, either.
This is no guarantee that 100% private IP is evil. Nor is it a guarantee that it won't be licensed in the future. Nor is it a guarantee that Intel and HP won't come at each others' throats with a price war. But it's a degree of lock-in that should be a factor in any decision.
This issue isn't mentioned in either article.
The living have better things to do than to continue hating the dead.
Mr. Kirkpatrick's article draws significant business conclusions - Dell will prosper, Sun will fail- from his analysis of the relative positions of the players today. I believe that most of what he cites as fact is wrong:
but I'm not sure his conclusions are wrong.
More precisely, you can't draw his conclusions from either his "facts" or his arguments, but that doesn't invalidate the conclusions.
For one thing articles like this become self-fulling prophecies and their prevalence in management oriented publications like Fortune help explain how Sun can be both a strong company and very weak share.
He may well be right on the specific issue of Itanium's future. Technically it's a pretty good chip and the fact that it's late and under-powered won't be important in the long run -the PA-RISC, which became a significant success, was also late and under-powered.
So will the Itantic sink? In my opinion Mr. kirkpatrick's article missed most of the significant elements in today's market picture that will affect this.
For example, the right parallel could turn out to be Intel's original Pentium Pro. As Intel's first completely 32 bit chip it was, briefly, a world leader in performance but only on 32bit applications. Since most Microsoft software used the older 16bit instruction sets, its performance on the Pentium Pro was terrible. As a result AMD was able to seize significant market share with its K-586 and Intel was quickly forced to re-introduce 16bit compatiblity in the Pentium line.
Years later the Pentium Pro came back - as the xeon - and that could easily be Itanic's fate too, if management at companies like Sun and AMD get their act together and make it happen. (see my article for my comments on how this could be done).
Was this posted by an Intel PR guy? Seriously, when is a new product by any company not the most power powerful, the greatest, or the best ever? I don't think there has ever been a case where a company released a new product and said that it was substandard compared to its previous offering.
Actually, I don't mind the idea of breaking X86 compatibility - I just object to breaking it for IA-64. IA-64 was conceived in a time when it was felt that Out Of Order (OOO) execution was going to be too tough a nut to crack.
In less time than Intel and HP took to go off and crack the VLIW/EPIC problems, other design teams learned to handle OOO, and do a very good job of it. They appear to have succeeded, and have a leading-edge part - but at what cost. AFAIK, the IA-64 is the most expensive CPU ever made.
The latest-out CPU usually does seem to hold the performance crown. But IA-64 doesn't seem to hold it that solidly, and there's question about whether the latest Alpha iterations have been allowed to fully appear - for fear of embarassment.
IA-64 looks almost like a government project gone wild. It has produced results, but IMHO horribly inefficiently. Pushing a more reasonable (not necessarily more conventional) architecture might well have yielded better results.
The living have better things to do than to continue hating the dead.
and AMD can't reignite the market with faster CPUs that won't require special software and revamped OSs to take advantage of their features, what makes either company think this will work either?
I smell another marketing blitz leading to slow coroporate and consumer adoption. I can't wait until someone does a serious study of hardware and OSs still in use in like 2008? How many PII w/Win 9x will still be around? I'll bet far more than the industry will like.
Don't be so quick to predict the demise of Itanium. I would question your analysis based on past history of Intel products.
1. Heat - has been an issue since the 8087 and lower power products or improvements in heat removal technology have continuously become available. Even in current Itanium/Itanium 2 (Itanium Processor Family - IPF) products, heat is an issue but not one that is preventing IPF products from shipping. Over time you will see a significant reduction in dissipation in Deerfield/Monticito (SP?) but, in any case, solutions to the heat issue are becoming available.
2. Cost - Intel products are only expensive while customers are willing to pay high prices for them. Any time Intel has had competitive pressures, they have been able to drop the price to meet the new price point OR introduce new products that allow them to maintain their margins.
3. Nobody seems to understand that there is an IA-32 processor core built into the chips (starting with McKinley (Itanium 2)). For backwards compatibility, it's really an operating system issue more than a hardware/software emulator issue. When the operating systems are properly implemented, IPF will be able to run 32-bit IA-32 applications concurrently with 64-bit IPF applications. When Linux supports this, I think you'll see interest in Hammer wane.
4. I would disagree with your comments on the people who want 64-bit already have them. I would not disagree that there are limited projects testing out different 64-bit architectures, but I would be very surprised at there being any large server farms out there with the latest incantations of Power or Alpha and the SPARC/MIPS are probably looking for an upgrade.
5. Itanium is ideally suited for Linux. I agree with your comments with regards to Windows - but when you are upgrading to a new Linux release don't you rebuild/retest the application to make sure it still runs? In our Linux systems we have been able to port directly from IA-32 to IPF without any changes to application software.
I believe that there is a lot of opportunity in the market for a "standard" 64-bit processor and this is what IPF is designed for. IPF may not be the best or the first but they do have the track record in taking over a market and maintaining it. Nobody has made a lot of money betting against Intel and nobody has ever gotten fired for choosing their products.
There are chips out that come close. The new C3 processors (VIA) run at 1 gigahertz. They also use 15 volts of power and dissipate under 10 watts of heat. And then there's VIAs Eden, which is an embedded processor platform (yes, it will run linux) that runs up to 1 gigahertz, IIRC. And according to them, it uses up to 1.2 volts and dissipates up to 6 watts of heat. And that's less than 1/10th.
And it's not only about power consumption. A lot of people have gotten sick of machines that sound like lawnmowers, and are going to the quiet side. Quiet is the new Overclock. You now can have a 2 gigahertz machine that only puts out 20 decibels of noise at 1 foot.
"Sometimes a woman is a kind of religion, she can save your soul & set you free from all your sins" - Bad Examples
Produce a 64 bit chip which runs as fast/faseter than the fasters p4. Price it a $50 above the fastest p4.
Past jumps from 16 to 32 bit always were targeted at the end user not the server. That's why they succeeded.
The biggest problem with older 64 bit chips (e.g., alpha) is that the price was way way higher than anything remotely on a power workstation machine.
Personally, I think that itanium faces an uphill battle, and that if x86-64 wins, AMD will be sitting pretty - more than the article indicates. Compatibility is king, especially compatibility with all those windows apps that are currently out there, and if a chip vendor's servers can run the same binaries as their desktops, you have a winner because desktops are where the volume is.
Intel built its lead through what seemed like excessive compatibility during the 80x86 vs 68k wars - they now ignore compatibility at their peril. I don't think they've built up so much brandname recognition that they can suddenly ignore this.
This is what i heard about non-intel versions of windows NT. (including the alpha) If you wanted a port of windows NT you had to license the suourcec code port it yourself (the company dec or motorola) and then give it back to MS for distro or something. We have seens Motorola's track record (the slow devlopment of the PPC) is there little wonder if died.. they didnt want to keep paying MS. I thinktheonly reason DEC kept going was they actually had customers and sold a lot of their alpha NT units,and thus needed it to keep going.
I buy Intel because their chips and chipsets are rock solid stable, at least compared to other PC chips and chipsets. And for ultimate stability you can even go with an Intel motherboard. Besides stability they are also compatible with a wide range of hardware. You don't have to worry about filling up every DIMM and PCI slot, it will just work.
Maybe if the people who buy Intel today are left behind by the Itanium, they will drive a market for stable and reliable chipsets and motherboards for AMD processors.
I want to virtually whap the person who modded the original poster as a troll.
You are substantially correct that the reason why Itanium has not been widely successful is the there are few operating systems and software that take full advantage of the Itanium chip. It also doesn't help that Itanium CPU's are still exorbitantly expensive compared to Pentium 4 and Athlon XP CPU's.
Yes, Linux does support the Itanium CPU but does it support all of its functional features? I also think that the 64-bit version of Windows XP Professional doesn't fully support the Itanium CPU, too; this is unlike the 32-bit Windows XP Professional, which does support MMX, SSE, SSE2 and/or 3DNow! extensions found on today's high-end Intel and AMD CPU's.
Because to support Itanium's native registers requires knowledge of VLIW, I still think the Itanium will probably remain pretty much a niche market CPU unless someone comes out with a way to convert current IA-32 compatible programs to run in IA-64 mode really cheaply and easily.
I still think that's true. Windows on Itanium is a terrible value proposition -- almost nothing will be native for years and years to come, and x86 execution mode is way way too slow to be cost effective. I think we'll see very little Windows on Itanium.
OTOH, Itanium is virtually ideal for vendors moving from proprietary chips/UNIXen to Linux. I was still fairly skeptical about Linux' chances back then, but I'm not anymore. Linux on Itanium is going to be a smash hit and will dominate the datacenter.
Windows on servers is ... iffy. I see the possibility that AMD's x86-64 will be a hit in that market, but you'd have thought Athlon would be interesting too and it was completely ignored. Then again it's Microsoft's only real chance in the large server market so you can count on them pushing it really hard. If they succeed then expect an Itanium with a much improved x86 execution mode; I don't think Intel will go the extended-x86 route. If AMD does not succeed then Windows is going to be pigeonholed as a small server.
Regarding other chips, only POWER looks set to survive/thrive, but only in traditional IBM environments. Sun is in the middle of a financial collapse; I would be surprised if we see more than one additional generation of SPARC technology from them. Fujitsu has a nice SPARC, years ahead of Sun, but SPARC stuff is such a bad value proposition these days that it and Sun are going to die fast.
jim frost
jimf@frostbytes.com
With 64 bits making its way to PC's (no matter that intel says its heading for servers only, while AMD says they will make it possible to buy by home users), in no time, 128 bits will make its way to (real) Servers. (i.e. Sun, IBM, SGI).
errera hunamum ets
I'm not sure if the 32-bit operation in x86-64 and IA-64 work by mode switching, or by some kind of emulation. In any case I expect it's a lot different than the Win32/16 way.
Escher was the first MC and Giger invented the HR department.
The chips names went like:n tium
8086
8088
80186
80286
80386
80486
Pe
Pentium Pro
Pentium II
Pentium III
Pentium IV
I meant to add at the bottom of my names list that the "3" in 386 was just the next in series and NOT indicating 32 bit. The 386 was 32 bit and that is a useful way to remember it, but it's not "the truth."
Because you can JIT compile it and profile the code and easly(he says) optermize the code.
Intel spent a hell of a lot of money making a C compiler when they should have made a good Java VM and tried to take the SUN/IBM business away from banks running java.
thank God the internet isn't a human right.
Do you know what a tardball you look like calling it 'Itanic' ?? It's almost like you're a little kid who needs to call people names to belittle them.
Well, Mr. PhysicsGenius, you're obviously not a genius when it comes to CPU architecture, operating systems, or the Windows APIs. I hate to flame the pro-Linux crowd (mainly because they hold most of the /. mod points), but this sort of thing, both in Linux and BSD crowds, is what makes us all look stupid. Like when you see these Windows vs Linux comparisons. Advocacy is great! Ignorant advocacy, however, is detremental to the cause. I mean look at this crap. This guy's point is that Windows is bound to a 32bit architecture and will have great difficulty moving to 64bit. Yet there have already been 64bit versions of Windows, and there already ARE versions of Windows in various stages of development for IA-64 and AMD's x86-64, not to mention that there were ports to Alpha and other platforms. This is the purest ignorance. My point is this. If you love Linux, tell people you use Linux. If you love BSD, tell people you use BSD. If they ask why, say "because it is so stable" or something. But don't say "because the archtecture is tuned in such a way as to make the porting to 64bit platforms much easier than W1nd0ze. I bet you M1cro$l0th will go out of business with 64bit CPUs become the norm!!" It makes you sound like a complete tool. Especially when you say "Windoze" and "Micro$loth". May I refer you to the following Penny-Arcade cartoon:
- 07 -22&res=l
http://www.penny-arcade.com/view.php3?date=2002
The fact that a good compiler takes an inordinate amount of time to create for Itanium is bad for small businesses and free software developers. It's good for Intel, if they succeed, since it will change the game (no compatibility) and make them top dog. Hmmmm. Where have I seen this before? Oh, right, Microsoft (ie, embrace Java, extend to C#).
x86-64 will be very easy for compiler writers. My company's own compiler would take 6 weeks to port to x86-64, but an IA64 port would take person years.
We all know that there is a much higher margin in servers than desktops, but real money has ALWAYS come from VOLUME. Whoever can make a real go at the consumer desktop will end up in front. Intel dosent seem interested in a 64 bit desktop whatsoever. They have created a platform (ia64)that suffers from the exact same problems as the ones that they displaced, namely price to performance ratio. If AMD can compel consumers to move to Athlon 64, they are in for good times and happiness. Personally I would like to see IBM make a go at the desktop with power5. Good luck getting a version of windows for it though.
People who think they know everything really piss off those of us that actually do.
I don't see how a new processor from Intel, or anybody else for that matter, is going to cause "serious competition" for any vendor such as HP, IBM, or Sun. When choosing a solution, IT doesn't go for Sun because its run on a Sparc CPU. They don't choose IBM because it runs a PowerPC. I give up on why they choose HP. :)
The point is, the CPU is just 1 little part in a solution. Intel isn't going to do any damage to these vendors unless they supply the entire solution, which isn't their business! To think otherwise is pretty dumb and a bunch of PR bullshit attempting to inflate Intel's stock value.
A slip of the foot you may soon recover, but a slip of the tongue you may never get over. -Benjamin Franklin
Next question, please.
First off Big Blue is Big and Blue but definitely not #1 in High Performance Technical Computing the Alpha has been there for years. If I could make that period any larger I would.
What they are is large and have a lot of accounts in business computing i.e. IBM.
Still the stock exchange and banks seems to like Alpha running Tru64 and OpenVMS, oh by the way so does the federal government, but I digress.
The IA-64 or Itanium 2 is an EPIC (Explicitly Parallel Instruction Computing) Chip design. The goal is to move beyond the RISC performance bounds with explicit parallel instruction streams. This simple put means that most processors today are 60% idle and of that 40% a whole lot of NOOPS ( no operations, or clock ticks with no work being done) are going on. The chip has a streamlined pipeline, and massive amount of registers 128. Because the compiler does optimize code (at least the HP and Intel complier does) it can handle conditional and/or unpredictable branches in a more efficient way and creates parallel code for the chip.
The late chip now holds very impressive numbers and they will only get better with the next 2 generations already in the pipe, this is a very early chip.
Processor, Clock speed, Transistors, sys bus bandwidth, SPEC INT 2000, SPEC FP 2000
Itanium 2, 1Ghz 221M 6.4 GB/Sec, 810, 1431
Alpha 21264C, 1.25Ghz, 15M, 8GB/Sec, 928, 1327
P4, 3.066Ghz, 55M, 4.2GB/Sec, 1130, 1103
Power 4+, 1.45Ghz, 7.2GG/Sec, 935, 1295
The 4 way OLTP performance is 50% faster than a comparable system and 10% faster than Dells 8-way
HP's Rx5670 performs 50% faster than an 8-way IBM x440 on the SAP-SD benchmark
And the SPECweb99_SSL performance is 1.4 times faster than a 4 way IBM p630 @ 1 GHz.
Oh and did I mention price per performance? Why don't you check it out yourself?
In my view it is not a "will it make it" question, it already has. The question is how big will it grow.
HP as developed a lean and mean chipset to support 1-4 CPU model. These production systems cover development machines and system that support SMP technology. Later chip sets will be optimized for VERY LARGE configurations.
What will this chip be used for? Anything! It most assuredly will run Oracle databases, and High Performance Technical computing. It will be incorporated in three tier infrastructures,
Storage Area Networks, and VERY, Very large clusters 10,000 seems to be the numbers floating around for a Lustre Linux cluster but that is not a power of 2 is it.
I am not sure where all this disinformation comes from but that is just what it is and is often a marketing/Sales ploy and this is not a good forum for any of that.
IF you BUY the cheapest Hardware you'll GET the cheapest hardware and then some. Get Inspired!
> they can break that constraining 4GM limit built into Win32
Speaking of Win32, why hasn't there been much news about Win64 lately (especially this last year) ? When is Microsoft going to officially ship Win64?
My biggest issue with 64 bit computing is that it takes twice the memory than 32 bits.
... // what happens now with memory usage?
I still only need 8 bits for my gray scale images and although I could pack 8 of them in 64 bits, how will I tell my compiler that?
unsigned char* pixels;
for(int i=0;i<sze;++i){
process(pixels[i]);
}
Doesn't that just guarantee a disadvantage? I know I personally may pick up the AMD clawhammer (or whatever it's called when they release it) but I'd never touch an itanium, because there are hundreds of apps I'd still want to run that are 32-bit based.
And remember that few businesses want to incur the cost of replacing their entire software suite on their servers (you'll spend way more on that than the hardware itself if you do anything more than run an exchange server), so Opteron has a good advantage there (if it ever arrives).
U.S. DECLARES BLAH!!
I buy Intel because their chips and chipsets are rock solid stable
That's funny. I recall Intel being in the recalled motherboard of the month club recently. Between all of the problems that they were having with RDRAM and then with their SDRAM bridge chips things were getting really ugly.
Frankly, AMD's use of the Alpha's bus architecture for their dual-processor boxes makes them much more attractive. Dedicated memory bandwidth for each CPU is a nice thing. (It would be nice to see them scale up to 4 and 8 way boxes however.)
We've got a Beowulf cluster of dual-AMD boxes and the thing just cranks out the calculations.
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
Itanium is NOT ideally suitable for Linux.
The compiler requirements for Itanium are simply too high. Unless the GCC team has gotten some SERIOUS assistance from Intel, I would not expect the Itanium version of gcc to be good enough.
The real problem with Itanium is that it requires a remarkably better compiler.
A Pirate and a Puritan look the same on a balance sheet.
So I don't think Itanium has a snowball chance in
hell. The DMCA crap will cripple the box, maybe Sun will come back with a vengance.
Work is well under way on Win64, but in Microsoft's ideal world, almost no-one will write to the Win64 API - they'll target the CLR, which itself will be 64-bit native.
.Net not for writing native apps. That would be like expecting everyone to write their native applications using Java.
Umm, the CLR is for
I wouldn't like to see a CLR that was designed down to a level to allow things like DiskKeeper to work.
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
Yes, that's why it's called the Win32 API. Work is well under way on Win64, but in Microsoft's ideal world, almost no-one will write to the Win64 API - they'll target the CLR, which itself will be 64-bit native.
!? 64 bit Windows has been around for quite awhile. Windows 2000 IA64 has LONG been available, along with Windows XP IA64 for Itanium (not Operton). Windows 2003 Server IA64 will release in April with the 32-bit versions.
The
AMD's participation in the market is what makes the current generation of x68 CPU's commodities. Remember the prices you used to pay for Intel CPU's before AMD became a viable competitor?
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
I would be very surprised at there being any large server farms out there with the latest incantations of Power or Alpha
You mean incarnations?
You'd need two separate memory supplies, two separate clocks, independent busses, two graphics cards too probably...
The problem is that the two cpu's use such wildly incompatible hardware throughout the whole system as to make it impossible.
On the OS side I don't want to think how you'd deal with the different speeds and signalling challenges. Think about when Apple had their Windows PC on a board that you could add into a PowerPC box. You needed to supply it with its own memory and everything. They just captured the VGA output and rerouted it for display by the internal system.
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
you can get a 2.4GHz P4 that's so fast you can write Quake 3 in interpreted Smalltalk and it runs like lightning
:-D
Sounds almost like a challenge.
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
...and I think that they run an ATI graphics controller.
There was an article on the Inquirer about optical connections directly to a Hypertransport bus. This will be a powerful feature for NUMA, and will make your shared memory much more feasible for distributed systems.
Hear, hear. Sun can't decide what they want to do. They hate Linux, they sell Linux. This is the sort of dichotomy that will destroy them in the end. Not to mention that the only place the SPARC can compete is in massively parallel servers. They no longer have the high-end workstation market. One can get a MP Athalon or PIII (did somebody say Xeon?) that blows the doors off Sun workstations now for much less money. They refuse to let Java become a standard, instead they retain control and whine that MS has poisoned Java (which they have, but that doesn't stop anyone from downloading and installing Sun Java).
Sun, vi, Bill Joy. Dinosaurs still roam the earth!
...didn't save sparc!
itanium price/performance is still shit, btw.
Do you know what a tardball you look like calling it 'Itanic' ?? It's almost like you're a little kid who needs to call people names to belittle them.
I haven't heard anyone use the word "tardball" since I was in third grade.
OMG. You made my day.
Microsoft typically uses typedefs to ensure the right sizes of ints and such. UINT and DWORDs are 32 bits long. I think there are also INT16, INT32 and INT64 types. When porting to a new architecture, they'll just make sure the typedefs are right.
Woah, a whole 6 months.
Unfortunately, no-one had repealed Moore's law and the PPro was more-or-less a dead end.
I can't remember exactly what the problem with existing software was, but it may well have been a 16-bit/32-bit issue. Not so sure about those early K586s though, I though the first ones had some other serious problem. It could have been a very slow fpu, this is all so long ago.
Mielipiteet omiani - Opinions personal, facts suspect.
Fuck, you know PhysicsGenius is one of the most clever, infamous, funny trolls on Slashdot, right?
Oh well, it's funny to see people being genuinely trolled. (IE, clever trolls. Not goatse or blatant flames)
It's great to mention that FreeBSD fully supports ia64 as a Tier-1 platform. Now no one has to run linux any more.
I keep hearing that the heat issues can be resolved. Fair enough, but can power consumption be ignored? The power to "resolve" heat issues and the power to run the chip are not imaginary line items. Most individuals almost never factor the cost of power into cost analysis while most businesses (that avoid the big 11) tend to factor electricity costs into the bottom line.
The UltraSPARC-III never whipped any opposition in the 64-bit arena. Alpha has always been the performance leader over Sparc. Check out spec.org or any other benchmark you would care to run.
In his Fortune articles referenced above, Kirkpatrick seems to erroneously think that 64 bit computing is about the width of the data path. The most important attribute of a computer architecture is the amount of addressable memory. A 32-bit machine has 32-bit pointers, which allow a single process to potentially address at most 2^32 locations. In most operating systems, the OS reserves some address space for itself, which means that a user process can only touch 2^30 or 2^31 bytes (ie., 1-2GB at most). In contrast, a 64-bit machine like the Alpha has 64-bit pointers, which allow a single process to potentially address up to 2^64 locations. (In practice, the number of addressable locations is less due to motherboard constraints, eg., 2^40 locations).
Alpha failed due to bad marketing alone. The version of NT that Microsoft supported for Alpha only supported a 32 bit address space! So running NT on Alpha provided no benefits versus running NT on ia32. So NT support for Alpha was never very real, and certainly didn't convert any customers from Intel to Alpha.
The most bleeding edge of computers may seem unnecessary now, but it is not too long ago that certain people were pronouncing that 640k of memory would be enough for everyone (and sounding quite reasonable at the time). Keep going back and you'll find other great gems (e.g. the entire computing needs of the U.K. will be provided for by three computer!).
If a 3ghz P4 looks overpowered bear in mind that Sony's goal for the PS3 is rumored to be 1 teraflop.
Way back when, processors were 16-bit. That means that they could only reference 64 kilobytes at a time (that's 2^16 or 65536 bytes or 64 kilobytes). If someone needed more memory they would have to use what is called the "segmented model". What that means is they would have to point the processor at another 64 kilobytes of memory. This was not only inefficient, but also dangerous.
:( (that's 2^64 or 281474976710656 bytes or 256 gigabytes or 0.000244140625 petabytes). But, who cares, right? We can just worry about it ~10 years down the road.
:)
All this is why Intel then introduced their 80286 processer. This processor was 32-bit, which means it can address 4 gagbytes of memory (that's 2^32 or 4294967296 bytes or 4 gigabytes). What this processor also introduced (in a somewhat crude fashion) is what is called "protected mode". This forbids programmers from messing with segments, which therefore improved security between programs running simultaneously (multitasking). This wasn't a problem though, as nobody ever thought we would need more than 4 gigabytes of RAM.
Today, it's a different story. 4 gigabytes of RAM is becoming more and more common. That's what the 64-bit processor is going to solve. 64-bit processors allow 16 petabytes or memory to be addressed at one time (that's 2^64 or 18446744073709551616 bytes or 16 petabytes or 16777216 gigabytes or 16 petabytes)! With the talk of processor speeds no longer increasing exponentially and such, I seriously doubt we'll need more than that anytime soon.
Oh, and by the way, AMD is cheap, they're going to (for the sake of making more money) release their Opteron processer as a 48-bit processor. It can do 64-bit math, it has 64 bit registers, but it's only 48-bit. This means we're only going to be able to address 0.000244140625 petabytes
-iamthecoolest, xeon at #osdev (irc.freenode.net), http://www.xeon.terminal-n.com/
P.S. I'm new!
> Linux on Itanium is going to be a smash hit and will dominate the datacenter.
.
>SPARC stuff is such a bad value proposition these days that it and Sun are going to die fast.
Value depends on what you can do with a system. I can find thousands of third party (non-Open Source Software) applications that run on Sparc systems. If LinTanic is going to be a success, it will need those applications. Where are they?
I suspect that as they start to show up that Managers will find LinTanic systems still cost real money, and its time to do the math again. Itanium is targeted at building enterprise systems. Buying enterprise systems implies paying real money. Running critical applications on them implies needing real support. Real support implies paying real money, at least if the Linux vendor wants to stay in business. Linux isn't getting as many cost breaks these days when it comes to running commercial applications. When you add it all up, I suspect that the cost differenct will come close to a wash and that Linux may not be much, if any, cheaper than Sun or other tradition Unix vendors for high-end datacenter applications
You can see this with Red Hat now. Software and support for the LinTel systems I plan to buy this year will cost me as much or more than my Suns.
No manager that wants to keep her job will risk million dollar / hour down time costs against the savings of getting an answer from the alt.datacenter.elite.linux mailing list. If linux has a future in high end systems in the datacenter, then real $$$ support, hardware compatibility lists, extensive testing and certification, and profitable (from you) Linux companies are what are in its future. Either that, or it will be a cold day in hell before debian is used to run Etna's nightly batch processing jobs.
(Ya ya, I know, 31i73 X86 cluster, 3Ghz cpu but clogged bus, LINUX ROXOR!, kick-butt, free licenses, MySQL + ruby, whatever. Get the all facts and do the math or stay out of the datacenter.)
What... happened to Itanium 1?
No sig for you!!
CPU power. Software. Labor. There is one prevailing trend in IT - commodity solutions. The high-margin players are toast.
From what I heard the ia-32 layer is not really a layer but a few instruction sets that ia-32 code can use in a emulator. Its as fast as a pentiun-100.
Compiler support really sucks and will probably never truly be optimized because somethings can not be done on compile time. Gcc sucks on anything non intel except for maybe the powerpc which took years to get otpimized thanks to Apples help. Rememeber hackers design it so its optimized to run on their platform ( lintel).
AMD hammer is much better chip and maybe the only option for 64bit Windows because Itanium2 systems are difficult to find.
Infact the Itanium2 is rumoured actually to be the third and not second iteration of the ia-64. Why? Because the first one performed so bad intel and HP decided not to bring it out into market because it would damage the hype.
The itanium is extremely overclocked with a whopping 1 pound heat sink just so it can benchmark on an equal footing with pentiumIV's and Athlons and older Alpha's. This is why the fans are huge and chip is humongous. The put everything in the chip in order to make it fly but its really a brick.
http://saveie6.com/
Why do you think a lot of intel related optimisation stuff (such as SSE/SSE2 optimisation) improved drastically relatively recently?
The real problem here is that the chips, while designed for SMP, aren't designed for AMP. I'm sure it would be possible to design such a board (and chipset), but it would be hard, and tricky.
I was rather surprised... several times in the article the author more-or-less predicts the death of Sun, and all without any explanation as to why. Sun is a large, well-established company with a mature 64-bit processor architecture and a sizable share in the 64-bit server market. Until I hear some good arguments for Sun's impending doom, I will continue to expect Sun to be a real player in the 64-bit competition that will be ensuing shortly.
The new architecture in Itanium is a great opportunity for the open source community to innovate not just immitate.
Check it out.
i was responding to your assertion that massive registers and streamlined pipelines make a chip inherently fast (in this case, your assertion was for ia64).
yet sparc had both of those features and sparc wasn't fast at all.
*that's* the relevance.
Legend has it that 64-bit Windows NT 4.0 for Alpha was about to go into beta in 1998 when Compaq pulled the plug.
Too bad Microsoft has stalinist webmasters, or you'd find quite a bit of late-90s PR about Win64 being developed with Alpha as the primary target.
(And the big reason that Alpha NT failed had nothing to with 32-bit -- believe me as a olde time NT admin crying for CPU back in the day -- DEC and Compaq just clusterfucked the marketing.)
On NT, 16-bit code was and is run by a VM that trapped 16-bit calls to the 32-bit API. On a older box, it was noticibly slower.
I expect the 32/64 bit thing to work the same way. It's even called WOW (Windows-On-Windows) still.
Win 9x really doesn't count because it was natively a 16-bit OS running on a DOS extender that itself was extended to handle 32-bit calls.
The itanium is extremely overclocked with a whopping 1 pound heat sink just so it can benchmark on an equal footing with pentiumIV's and Athlons and older Alpha's.
Linpack for 1Ghz itanium2 vs 1.25Ghz EV68 Alphas
EV68 is the most recent released alpha, or were you talking itanium1?
This is why the fans are huge and chip is humongous.
Eh.. most servers have large fans, I really didn't see much difference between the fans I service on alpha es45's and itanium2 rx5670's except for the fact that the rx5670 has twice as many for redundancy.
The put everything in the chip in order to make it fly but its really a brick
You mean you think it's smarter to put your cache off chip on a 200mhz front side bus instead of having an l3 cache running on the 400mhz internal processor bus speed?
Please enlighten me as to how "everything" was put in the chip, or how l1-3 cache constitutes everthing. From what I understand, the internal badwidth was boosted by a faster processor bus and a wider pipe, the line size of the caches were doubled, it has additional issue ports, additional execution units, and the core frequency hasn't really gone up that much from 800mhz - 1ghz.
AMD hammer is much better chip and maybe the only option for 64bit Windows because Itanium2 systems are difficult to find.
And here I heard that sales were more than double the expectations.... But then that's just a nasty rumor and we don't pay any attention to them.
"Fail
Error: Divide by zero
they'll target the CLR, which itself will be 64-bit native
Good luck with that CLR device drive, fucktard.
IBM has now all but withdrawn support for Linux on itanic: http://www.infoworld.com/article/03/02/10/HNitaniu m_1.html
Stick Men
"A fractal is by definition a set for which the Hausdorff Besicovitch
dimension strictly exceeds the topological dimension."
-- Mandelbrot, "The Fractal Geometry of Nature"
- this post brought to you by the Automated Last Post Generator...