Talk:Itanium/Archive 1

Misc
I don't understand the second paragraph. At first I thought it simply referred to the processor as Merced and I was going to change this to "the processor" becasue the name merced had not yet been mentioned. However, I am not sure if it is correct to say that the processor was originally intended as an architecture. Would it be more like the project was orinally intended as a architecture.

I know nothing about this but refering to "Merced" before the reader has been introdeced to the word is not good. Borb 11:23, 24 May 2005 (UTC)

'it can slowly run x86 code in a firmware emulation mode' is more correct than 'it can slowly run x86 code in hardware', as far as I know. Do any experts know better?


 * Actually, both the Itanium and the Itanium II can run user x86 code in hardware. It uses the existing IPF (IA-64) datapaths to do this. - Steve roseundy 02:49, 27 Jul 2004 (UTC)

"the Itanic" -- presumably in analogy with the RMS Titanic -- could someone confirm? -- Tarquin

The Itanic joke has been going around for more than two years now. It's very well-known in the hardware enthusiast community. It started on a message board post by a man from New Mexico in 1999 and has picked up steam since.

Here's a copy of the post:

From: Kraig Finstad (kfinstad@unm.edu) Subject: Re: Itanium Newsgroups: comp.sys.mac.advocacy View: Complete Thread (35 articles) | Original Format Date: 1999/10/04

In article <7tb3od$klj@news.or.intel.com>, "Kevin M. Taggart"  wrote:

> >All the proc names that Intel has come up with are retarded. Itanium is, >perhaps, the worst. They should have called it "Titanic" ... >                                                           . >                                                             . >                                                              . >                                                              . >                                                              . >                                                              . >                                                           <=\\\=> >--KT

But they probably shelled out big bucks to have that first "T" removed from titanium. In this case, they'd have the Itanic.

-Kraig Finstad John

This article is largely redundant with IA-64 - is there actually a difference anyone outside Intel marketing cares about? Should one be folded into the other? - David Gerard 11:15, Jan 25, 2004 (UTC)


 * Unless anyone strenuously objects (and is willing to work to differentiate the two articles effectively), I'll be merging this with IA-64 in a few days


 * Hmm strictly speaking, IA-64 is the instruction set, whereas Itanium is the physical processor. The IA-64 article should specalise on just the instruction set, while the Itanium article can be about everything else such as the politics, development and processor specifications. Intel may in time put the IA-64 instruction set in processors sans Itanium, like how IA-32 eventually went in things other than the original 80386/486 core. - 203.109.254.57


 * So you're volunteering to disambiguate? :-D - David Gerard 11:32, Apr 3, 2004 (UTC)


 * I've just put the detailed stuff on EPIC into IA-64 - David Gerard 09:12, 13 Jun 2004 (UTC)


 * User:203.109.254.57 has several clear points, but I see your point as well David. However I do think a technical crowd would respect and appreciate the distinctions between ISA and implementation.  Likewise a general crowd would appreciate having an article that focuses on what they care about (mostly implementation).  The Itanium article could focus on implementation, market application, performance, and future roadmap.  The IA-64 article can focus on the ISA, software challenges/developments, and perhaps emulation developments.  I will look at working on further disambig - Vector4F 04:40, 7 Jul 2004 (UTC)


 * It is of course a marketing fiction that "Itanium" and "IA-64" are different things. But yeah, that division makes sense. The detailed stuff on EPIC should probably go in its own article, assuming there are examples beyond IA-64. And so on - David Gerard 09:26, 7 Jul 2004 (UTC)


 * Well, to be pedantic about it, the official Intel name for what is termed "IA-64" here is "Itanium Processor Family" (or IPF). So IA-64 and Itanium are the same thing. I agree however, that separating implementation (Itanium, Itanium II, Madison,.. etc.) from architecture (IPF) is a good thing. - Steve roseundy 02:49, 27 Jul 2004 (UTC)


 * I've just written the Itanium 2 page. I'm going to make this page talk about Merced only -- architecture stuff now all seems to be on the IA-64 page - Matthew Wilcox 09 November 2004

End of Itanium?
I've removed the link regarding Microsoft and Itanium for three reasons:
 * It is unclear is Microsoft ever intended to support anything else than it's server versions of Windows on Itanium.
 * This is more related to IA-64 and Itanium 2. Itanium processors are no longer on the market (see Intel's website)
 * Traditionally Microsoft has little market share in the high-end server market (the market the Itanium was targeted at)

Moreover, if someone decides it should be added back. Please cite the real source http://www.microsoft.com/windowsserversystem/windowsserver/bulletins/longhorn/itanium_bulletin.mspx and not just some Inquirer article (The Inquirer in not know to be one of most reliable news sources, although it's very good at rumors).

-- Koffieyahoo 14:59, 6 September 2005 (UTC)

I have restored the suppressed news and discussion. Microsoft has been steadily cutting its support for Itanium. The analysis of the architecture appears in Itanium and not Itanium 2. The article now cites the industry announcement.--Carl Hewitt 15:57, 6 September 2005 (UTC)

Okay,

I’m getting really, really tired of you reverting stuff which I removed under, what I believe are valid arguments. If you do so, you should provide solid sustainable counter arguments. Unfortunately, you don’t even try to do that. Three examples:


 * Your C# addition to the continuations article: you add it back by arguing that I should quote the C# spec on saying that it’s not a continuation. This is complete non-sense: 99.99 percent of specs of programming languages don’t say what a certain construct is not. They don’t do this, as it bloats the spec and as implementers of the spec are not interested in what a certain construct doesn’t do.
 * Your Microsoft addition to the Itanium and x86 page: You add it back under the cover of saying that Microsoft is big player. But, as I already mentioned Microsoft isn’t a big player on the high-end server market, which is the market the IA-64 architecture is aimed at. Moreover, it can also be argued that Microsoft is doing a very clever thing: They’re targeting their version of Windows for high-end servers at some the core application areas. Hence, it might just be the case that they want to gain experience in these areas before they start supporting other applications on these high-end servers.
 * You added back the "Prospects for the x86" section to the x86 article, but up until now you haven't come up with any arguments on why this doesn't violate the wiki policy of not being a crystal ball. Moreover, you completely ignore any application areas for which the x86 architetcure is completely unsuitable (do I hear high-end server market?).

Anyway, I beg you to provide solid arguments before you add something back of which you might suspect that the person who removed it had very good reasons for doing so.

-- Koffieyahoo 07:55, 7 September 2005 (UTC)

I can appreciate your tiredness and frustration. The Wikipedia process is hard work!


 * The net result of the process in the article on continuation was that (as you suggested) the example of delegates in C# was removed.
 * Attempting to suppress information from the Wikipedia article Itanium based on Microsoft conspiracy theories is misguided.
 * Attempting to suppress analysis in the Wikipedia article x86 on the false charge that it is "crystal ball" is similarly misguided.

--Carl Hewitt 16:25, 7 September 2005 (UTC)

I have removed the ''Microsoft has decided not to make further releases of its consumer operating systems for Itanium, concentrating only on x86-64 (AMD64, EM64T). '' because as stated here Microsoft will continue to develop the Windows Server versions for Itanium ISA. And if the removed text meant "consumer" versions of Windows, well, it's off topic and wrong too because further consumer Windows releases will not be x86-64 only. Lvn 23:22, 4 March 2006 (UTC)

Repeated deletion/whitewashing of Criticism section
Why are some Wikipedians so bent on removing the criticism section? Even after references were added, some seem determined to make sure the article only contains positive notes about the processor. &mdash;Joseph/N328KF (Talk) 05:44, 5 December 2005 (UTC)

Copyright violation
As if the edit war by User_talk:65.102.172.33 on the Opteron and Itanium pages wasn't bad enough, a simple google search found that his recent contribution of "info on architecture" was copied from this Anandtech article. The stupid idiot didn't even put in proper images, just left them using Example.jpg. Imroy 07:18, 5 December 2005 (UTC)
 * There was no need to use the Template:Copyvio tag. You only use that when all text is suspect. In this case, just revert to the last legal version. &mdash;Joseph/N328KF (Talk) 12:43, 5 December 2005 (UTC)
 * I didn't add the copyvio template, that was Peyna. Imroy 04:05, 6 December 2005 (UTC)
 * Yes, that was me. Didn't realize there was a non-copyvio version available. Peyna 04:09, 6 December 2005 (UTC)

Response to Copyright violation
Copernicus, I was in the middle of the edit when you so eagerly reverted it resulting in merge error. I fully intended to add proper citation, credit, and linkage as well as correct the example image.


 * I don't care how much citation or credit you add, that's still a copyright violation. I suggest you read the copyright section of the contributing FAQ. Unless you can get the original author to relicense the work to be compatable with the GNU FDL, or the public domain, then it cannot be added to Wikipedia. Simple as that. Imroy 07:32, 5 December 2005 (UTC)

Fine, I will write an original description of the architecture.
 * Note that this is an article about the Itanium processor. Just as there is another page for Itanium 2. There's already a good and detailed right-up of the architecture at IA-64. Imroy 07:45, 5 December 2005 (UTC)


 * Good point. The Itanium clearly failed as a product, due to technical reasons I've now put back in the text. Any arguments about the future of IA-64 should be taken to the Itanium2 page. Timharwoodx 14:15, 26 December 2005 (UTC)

HP marketing
I've seen examples of people who ADMIT to working for HP editing WIKI articles i.e. NVIDIA TNT2 recently. I think there has been a clear trend of removing vaid content, that suggested Itanium might not dominate the entire planet and make multi billion dollar profits for HP/Intel. I would hope this  product is now so obviously dead, and being laughed into its grave by guys like Linus Torvalids in public, we can leave perfectly valid technical observations on the chip intact. I restored some excellent technical observations made several months ago, for which I can see NO VALID reason for removal. They are spot on. It was a massive screw up, due to too many bloated egos at HP and Intel paid far too much money. 85 engineers designed the VIA C7. What a joke that makes of multi billion dollar Itanic project.
 * The sales persons at HP (2001/2002) kept saying that customers should buy HP Pentium based servers, because "Itanium products are difficult to obtain".

XP Porting
''Microsoft has ported Windows XP, Windows 2000, and Windows Server 2003 to Itanium. Microsoft server applications include SQL Server, Operations Manager, CRM Server, IIS, Visual Studio, and the .NET Framework. The decision was made in recent years to not include support for client applications or client operating systems (such as Windows XP) for the Itanium, because the market demand is too small to justify the porting and support costs.'' I wasn't aware MS ported XP to Itanium. Was some or all of it ported, but never shipped?
 * MSTCrow 15:45, 8 April 2006 (UTC)

Itanic derisive nickname
Why should a derisive nickname promulgated by a satirical website be mentioned on the official website of the Itanium project? I think we should remove that paragraph completely. -Սահակ
 * Because it's widely used. There has been a campaign by Itanium fans to get rid of any negative information about this processor. It needs to stop. There are 157,000 Google references to the nickname. &mdash;Joseph/N328KF (Talk) 04:45, 9 April 2006 (UTC)
 * What do you mean 'official'? --ajdlinux 10:10, 13 April 2006 (UTC)
 * That's a good catch. I guess that word didn't make sense to my mind in this context, so it just ignored it. ;) &mdash;Joseph/N328KF (Talk) 10:42, 13 April 2006 (UTC)

I did a google search and found only 532 unique results for Itanic. Type "Itanic" in google search box, and then scroll through the pages until your reach 5xx results. After that it says: In order to show you the most relevant results, we have omitted some entries very similar to the 532 already displayed.

500 mentions is not "A widely used term" —The preceding unsigned comment was added by 24.59.115.230 (talk • contribs).


 * Type "Itanium" in the Google search box, and then scroll through the pages until you reach 6xx results. It will also then issue the "we have omitted..." message.  Type "Computer" and you get the same result after 8xx results.  I don't believe that means there are fewer than 700 unique hits for Itanium or fewer than 900 unique hits for computer. --NapoliRoma 19:02, 15 July 2006 (UTC)

Most of the results come from register.co.uk, wikipedia, or sun.com who invented that term. —The preceding unsigned comment was added by 24.59.115.230 (talk • contribs).


 * A Google search for Itanic today returns 102,000 hits. A Google search for Itanic, excluding those three sites ( itanic -site:theregister.co.uk -site:sun.com -site:wikipedia ) returns 91,300 hits. --NapoliRoma 19:02, 15 July 2006 (UTC)

NapoliRoma, I think you are right, but then I don't know how would you measure objectively the popularity of a certain term. I don't see a reason to put that information into the article, because it is a subjective term. It might be very popular among TheRegister readers or Intel bashers, but not very popular among people who rely on Itaniums for everyday computer simulations. —The preceding unsigned comment was added by 24.59.115.230 (talk • contribs).
 * I don't know how you definitively measure it either, but it appears that many feel it's a part of the Itanium story worth mentioning. It's clear that it has a life far beyond El Reg; in fact, I just saw it used in my local paper earlier this week in their story about the Montecito launch, which seems to contradict your "now rarely used" edit of today.
 * Personally, I think many reading the current "now rarely used" edit are likely to think "huh; didn't I just see it used the other day?" and view that description as an attempt at whitewash. --NapoliRoma 00:17, 16 July 2006 (UTC)

I've only heard the term "Itanic" used once or twice while the Itanium was popular and that was at the Register as mentioned above. It seems it was more their "ha-ha" buzzword than anyone elses and it really doesn't seem relevant to the article or add any necessary or interesting information.--Urbanriot 17:19, 19 July 2006 (UTC)


 * I still see it used on occasion. In any case, as NapoliRoma said, it is now part of the story. Think of it as being like the name "Spruce Goose" for the Hughes H-4 Hercules. &mdash;Joseph/N328KF (Talk) 17:38, 19 July 2006 (UTC)

Later Note: User added some information about the origin of the term on 2006-09-19. I've just edited the section into more encyclopedic style. Feel free to improve my wording! Cheers, CWC (talk) 09:02, 29 September 2006 (UTC)

Itanium development
HP and Intel started working together in 1989. See http://news.com.com/HP+discontinues+its+Itanium+workstations/2100-1006_3-5381398.html

unnecessary sentence?
"Importantly, it was expected that AMD would be unable to clone it."

Firstly this is a speculation.

If its not a speculation but a fact, it should be written in present tense: "Importantly, it is expected that AMD will be unable to clone it."

and lastly I don't think that this sentence is important. This is an article about Itanium, not a gossip website discussing intellectual abilities of AMD.


 * Then, remove it. —BorgHunter ubx (talk) 21:52, 25 April 2006 (UTC)
 * AMD didn't want to clone it. Their 64-bit server strategy all along was to develop x86-64, which came to fruition as the Opteron. —The preceding unsigned comment was added by 74.245.91.25 (talk) 15:18, 12 February 2007 (UTC).

No need to compare the popularity of Xeon with Itanium's performance.
This paragraph:

"The original Itanium was succeeded by the Itanium 2 which delivered on the performance promises for the architecture with leadership benchmark results across a wide range of workloads. However, recent data from the TOP500 supercomputer list suggests that Intel's Xeon architecture is more popular among the scientific computing community, with 263 Xeon systems and only 37 Itanium 2 systems as of June 2006[10]"

compares Itanium's performance with Xeon's popularity. Which is like comparing apples to oranges. Xeon is clearly more popular than Itanium. But Itanium is still faster than Xeon. So I am removing the last sentence which compares these two processors.

Entire Itanium family, or just the first processor?
Is this page supposed to be about the entire Itanium family, or just the first (Merced) processor in that family? Guy Harris 00:51, 29 July 2006 (UTC)

Intergraph v. Intel
Is Intergraph v. Intel worth mentioning in the articles? Alex 17:02, 8 August 2006 (UTC)

Itanium vs Xeon
How do these processor lines compare? Now than Xeon is 64 bit, is Itanium being phased out?

--68.103.154.140 21:12, 4 September 2006 (UTC)

Itanium based supercomputers
I found one Itanium 1 based supercomputers at NCSA with 320 processors. The rank of the system is #53 in 2001. Is there any other faster system based on Itanium 1?

Difficulty in obtaining the CPU
In 2000/2001, there was a shortage of Itanium 1 CPU. I plan to order the evaluation board, but Intel put the priority on government contracts for Supercomputers. Buying the HP machines based on Itanium 1 was almost IMPOSSIBLE for several years (because of CPU supply problems).

Itanium Based Solutions
This section, added in September, was very weak. Most information was wrong and referred to x86-64 rather than IA64, for example all the 3d animation ports described as porting to IA64 were actually porting to x86-64. I updated this section (mostly by deleting stuff and leaving a historical note to an abandoned port). Also the Streaming Media section is pretty content free and probably not really worth keeping around. Need some more sections in here describing real applications, because the two categories in there right now are really niche markets for Itanium. Maybe just blow away the whole section for now until someone can compile a representative list?

Splitting this Article
Unless someone objects, I will move this article to Itanium processor and then add a replacement Itanium article to discuss the Itanium family as a whole. When this article started, it was about the whole family because there was no Itanium 2. Based on the recent edit history (mine and especially others) there is considerable confusion, and a lot of the market and application stuff relates only to the Itanium 2. After the split, the Itanium processor article will become purely historical.

The name change is important because, as User:NapoliRoma pointed out to me in a side discussion, most people who type "Itanium" into the search box are likey to want info on the latest processor.

I feel that the new Itanium article should not be integrated into the IA-64 article. The IA-64 article should focus on architecture, and we have no particular reason to believe that all future implementations of the architecture will be called Itanium. The Itanium article can be used for market, price, comparisons, applications, etc. Unless someone else either objects or undertakes this split, I will do the split on 2006-10-22. -Arch dude 17:28, 15 October 2006 (UTC)

Speed benchmarks missing
How about adding some specmark specs so people can understand how powerful a processor is.

This should be done for all of the processors.


 * This link:


 * [Ideas SpecINT2006 Benchmarks]


 * Shows that for Single Small/Medium/Large+Scientific installations,
 * Opteron/Xeon is tops in the SpecINT2006 benchmark,
 * while Itanium 2 dominates the Entiprise server market, but:


 * IBMs benchmarks in TPS, show the Power6-Based p560 at the top of the 16 core benchmarks.


 * [IBMs "Benchmarks"]

As for Benchmarking all the processors,


 * Tomshardware had a great roundup of 65 processors at:


 * [65 processor marathon of benchmarking]


 * and of course, the wikipedia article on benchmarking is a bit dated,

but covers almost all of the relevant issues:


 * [Benchmark (Computing)]


 * But it neglets to mention 'The Chang Factor' where a chinese computer vendor slowed down the real time clock to increase benchmark performance.


 * There are design factors that impact performance:


 * [The Opteron onslaught on The Good Ship Itanic]

"a quad-CPU Opteron 848 has 25.6 GB/s memory bandwidth, compared to just 6.4 GB/s in a quad-Itanium2"


 * Finally, just so your brain can go tilt: The IA64 Itanium processor uses speculative branch predition more than any other processor. Profiling the benchmark code, and improving the compiler can and will have very dramatic effects on the benchmarks, and will show up in application performance. IBM has multiple decades in doing this. —Preceding unsigned comment added by Artoftransformation (talk • contribs) 19:11, 29 September 2007 (UTC)

I recall a while back they ran Sun Solaris under emulation on a IA-64 box and it out performed any available system running it natively. I will have to look up the article but it would be worth adding to the CPUs history as it was a very impressive achievement. —Preceding unsigned comment added by 81.159.120.166 (talk) 13:34, 21 May 2008 (UTC)

from Talk:Itanium when this was at Talk:Itanium (original)
I arbitrarily created Itanium (original afer trying very hard to fix the mess left to us by Intel. Intel first named the Itanium, and then named the Itanium 2. Some referenced to "Itanium" refer to the original, some to the product line, some to the newest Itanium 2, and some to the architecture. Upon sober reflection, I decided that this disambiguation article is the least bad way to fix the problem.

Recall that last year we tried to normalize this by moving most of the generic stuff to "IA-64." This did not solve the problem.

I moved the old "Itanium"artcle to "Itanium (original)" This preserved all of the old comments, etc.

If there is a better way to do this, please discuss it here. -Arch dude 06:14, 1 January 2007 (UTC)


 * I would suggest doing what happened with AMD64/EM64T/x86-64/x64 : much agonizing and carping, not everyone happy, but eventually all melded into one page.


 * After a cursory look at the three pages IA-64, Itanium and Itanium 2, I don't see enough original content to justify three separate pages. I would suggest merging all three into one.  Sections could be: history, generic architecture, Itanium specific stuff, Itanium 2 specific stuff, market analysis, competitive.  Or not, but that would seem to fit in the space of a normal article without too much problem.


 * As part of the process, I highly recommend getting into a heated debate on the article's primary name; always good for killing time over lunch. Worked for the x86-64 page.  (It'd make sense to me to call it "IA-64", but I'd hate to spoil the fun by suggesting it this early in the game...)--NapoliRoma 21:05, 12 February 2007 (UTC)


 * But x86-64 is an architecture. There are also separate articles for each processor (Pentium, xeon, etc). And of course we all know (because HP, Intel, ISA, and IDC have told us many times) that there will be many more great implementations of the architecture as Itanium sweeps all other processors into the dustbin of history. But seriously: I would prefer to make "IA-64" a pure architecture article, convert Itanium 2 into "Itanium processors", and merge the current Itanium article into it and convert it into a pure technical implementation article, and then use the "Itanium" article for all the non-technical stuff: hype, history, market share, competitive analysis, etc. So we have three article but two of them are completely non-controversial and purely fact-based. After moving the bloat out of the "Itanium processors" article, we can also merge the Montecito article into it.-Arch dude 23:51, 12 February 2007 (UTC)


 * The cool thing about Wikipedia is that it's dynamic. If at some point, our Itanium overlords pause from their whipping ("we just like whipping!") to allow we, their fleshy minions, to write of the glorious 200 year history of IA-64 and how it came to dominate the human race, we could perhaps create an entire IApedia to hold it.


 * Today, all the Itanium content can (IMHO) fit in one page. If it starts to overflow, then consider splitting it off.  Today, having separate articles appears to cause confusion, as no one understands the divisions.  I don't think it's necessary to architect for tomorrow's potential IApalooza until there's a content glut to justify it.--NapoliRoma 01:38, 13 February 2007 (UTC)


 * OK,go for it. Create the mega-merged article and we can discuss it. I must admit that my strongest reason to separate into three articles was to try to isolate the technical from the controversial, but I will support you if you do the merge. Don't forget to merge Montecito. I'm not clear on how I can help, but if you want to split the work up, I will try to help. We probably need an administrator to unscrew the names after our community reviews your merged article. -Arch dude 02:22, 13 February 2007 (UTC)

Article misses some important historical perspective
I think this article is missing some important historical perspective on the Itanium. It is important to remember that Itanium and IA-64 were not conceived of as Intel's effort to split beyond desktop and server lines. Itanium was supposed to be leading the way as a next generation architecture which would gradually replace x86 altogether (backward compatibility was there only to allow an interim bridge). At that time Intel (with Microsoft) really did control the market. And they truly believed that they could convince the entire market to follow their migration path simply because they said so. They believed their new architecture would both improve performance and set back puny AMD's efforts to catch up. They also believed that developing a brand new architecture would be only slightly more challenging than developing a next generation x86 architecture (i.e. at the time they never dreamed that Itanium would fall so radically behind schedule). Their errors on all of these assumptions gave AMD a huge window of opportunity and forced Intel to make some rapid roadmap changes (which they actually did pretty well to their credit). Among other things they quickly stopped talking about Itanium as a lead vehicle for their overall roadmap. The whole episode, though, was a very good example of how success can go to a company's head and cause it to stumble.

I can't find an authoritative source on this (much of this is based on my own experience in the business and seeing earlier marketing efforts). Anybody know of such a source? --Mcorazao 22:14, 5 March 2007 (UTC)


 * There was nothing "rapid" about Itanium. The repositioning you speak of was too little, too late. This nightmare started in 1989 at HP, and Intel became involved in 1994 with a goal to release product in 1996 or 1997. Crippled product was finally released in 2001, with real product (Itanium 2) in 2002. By about 1997, the original premise of IA-64 (i.e., EPIC) had been overtaken by events. In 1989, EPIC looked to be the only way to break the one-instruction-per-clock barrier. However, super-scalar, pipelining, and related techniques turned out to be reasonably easy, and EPIC turned out to be hard. Intel (probably because of contractual commitments to HP) refused to acknowledge this until 2003 when Opteron and the market rubbed their noses in it. Even then, Intel's marketeers and upper management STILL tried to treat x86 as a second-rate technology suitable only for toy systems: the very words they used to describe x86-64 (e.g., IA-32 with 64-bit extensions") show this. It was not until the Core 2 Xeons in 2006 that Intel finally caught back up with AMD, and Intel even now will not acknowledge that Core 2 is a better choice than Itanium for big enterprise-class machines. I think we need to restructure the entire article set to accurately and objectively address this. See prior discussion. -Arch dude 23:32, 5 March 2007 (UTC)


 * Just about the entire computer press. When Intel and HP launched Itanium in 2001, the companies positioned the 64-bit chip as a replacement for RISC processors in large computer systems, and perhaps eventually for x86 chips on smaller servers and even desktops.


 * http://www.networkcomputing.com/channels/storageandservers/showArticle.jhtml?articleId=190900353


 * Google searches using terms like "itanium replace x86" which I used to get the above, well reveal such sources. Perhaps you should learn how to use google one day. Its free, and it works really well. Timharwoodx (talk) 18:32, 17 August 2008 (UTC)

Request for comment: consolidated article
I have attempted to write a consolidated article, capturing all the important stuff from Itanium, Itanium 2, and IA-64, with the intent of replacing the three with one article named Itanium.

The proposed article is in my workspace. Please review and comment. (Comments on the proposed article's talk page, please.)

In addition to simplification and consolidation, I have attempted to add citations wherever I could, and I have removed some material that I could not find citations for. The article is basically a complete rewrite, but almost everything in it is from the original articles, with two major exceptions:
 * The material on the architecture is considerably more detailed and focused on architecture.
 * I added a timeline, moved many statements to the timeline, and added material to it.

We are still missing a lot of history from the timeline,(mostly 1994-1997) and we are missing a paragraph in the "History" section about the period from 2002 to present. However, this is material that is not present in the current articles either.

Thanks.-Arch dude 23:06, 19 March 2007 (UTC)

UPDATE: Colleagues, the WP:MERGE guideline is that a proposed merge can be completed after a two-week discussion period if consensus is reached, or after a four-week period if no comments are received. So far, there are no comments. I intend to perform the merger on or about 16 april if there are still no comments by then, but I would really prefer to have someone review the proposed new article. -Arch dude 18:26, 30 March 2007 (UTC)


 * I am not a strong supporter of merging these three articles into one. I think that there is enough information to have separated articles and architecture is different thing than a processor. However, your article is quite good (for example, table for released processors looks much better than different sections in Itanium 2 article now) but I dislike a timeline. It feels like a collection of facts which unnecessary expands the article. All these facts can be (if not already) mentioned in appropriate sections of the article. However, if the merge is necessary I think we will need articles called Itanium (processor), Itanium 2 which can be smaller articles just about that processors and some facts about them without architecture details (in similar style like POWER and POWER5). Still when somebody will try to find information about IA-64 (architecture), it can be confused when it will be redirected to Itanium (series of microprocessors) so I would prefer at least to articles Itanium and IA-64. --Vezhlys 18:28, 7 April 2007 (UTC)
 * The most important stuff in the timeline is reflected in the chart, which is a later addition. I will attempt to consolidate the timeline facts into the history section, But the protracted development cycle is the central fact of Itanium, and we need to convey that. -Arch dude 19:13, 7 April 2007 (UTC)

Starting merge now, please stand by--DONE! -Arch dude


 * Sorry Arch dude but I do not think I can agree with your self. You could not really call the original Itanium as crippled as sunning native applications it was very impressive for its time. The real problem lay in compatibility. HP and Intel gave the original Itanium a very basic and slow hardware computability with X86. I’m this aspect they shot them self in the feet. Its quite simple really, The IA-64 was a high end item and people using them wanted high end performance. So no one is going to get a Itanium system to run X86. Yet there was very limited native IA-64 code apart from true high end applications where code is generally developed for it like in the supercomputers already talked about in the wiki. A very similar situation accrued with the old Digital Alpha architecture. Most people are not bothered about how technically good a architecture is how greater the future potential is from crossing to a modern architecture. Most people will only see the small picture and if something runs the programs they have faster they want it. So no matter how inherently poor X86 is its what we have and what we have used for many years now. In so X86 has the most money invested in it because more people are after X86. But pleased don’t be fooled in to thinking X86 is anything special as its a long way from it. As for your Intel VS AMD remarks I think its best not to start that as it can not be won. Yes AMD can make code CPUs that can be used well in some situations. But on the time scale you are talking I was using Intel derived systems using 64 separate CPUs controlled under a single OS. At the time that was not possible from AMD. —Preceding unsigned comment added by 81.159.120.166 (talk) 14:03, 21 May 2008 (UTC)

Help with futures?
This replacement article removed all of the un verifiable hype in our existing articles. I cannot find anything positive to say that is not speculative.

However, there is still hope for Itanium. In particular:
 * Montecito has fairly good SPEC numbers by comparison to Xeon and Opteron, even though Montecito is still on a 90nm process while the other two are on more modern processes. This shows that IA-64 really is a superior architecture.
 * Tukwila will be able to use commodity chipsets, thus permitting ISVs to build Itanium systems without investing huge amounts of money to develop a chipset.
 * The original theoretical advantages of EPIC are finally realizable: the inherent codespace inefficiencies are much less relevant (cache is proportionately cheap) and the simplicity of EPIC permits a much smaller processor core to do more work. With multi-core dies, this means you can have more Itanium cores in the same space.

So, the question is: how can we add this information to the article without violating WP:OR? -Arch dude 12:31, 16 April 2007 (UTC)

Unisys MCP and OS2200?
Are these environments supported on Itanium? Unisys systems have a mix of processors, and MCP and OS2200 are supported on the Xeons, for sure. Are they also supported on the Itaniums? if so, are they supported natively, or by instruction simulation? -Arch dude 12:31, 16 April 2007 (UTC)

On Gelato
The reference to Gelato is slightly incorrect. Gelato (see http://www.gelato.org ) is more than a developer community ... in fact most of its members are supercomputer sites, although it does have some developers (See the gelato membership page on http://www.gelato.org/participants/members.php for a breakdown).

Gelato is essentially an IA-64 user group, but certain members are sponsored to do particular projects ... for example UNSW to do superpage work, or the University of Waterloo to develop microC++. in addition, as compiler technology is so important to IA-64, they've been sponsoring work on gcc (see http://gcc.gelato.org/)

Note: I'm Peter Chubb, from Gelato@UNSW. I didn't want to add my biases into the main article. 65.91.54.2 04:55, 17 April 2007 (UTC)
 * Hi, Peter. I wrote that sentence based on cursory examination of the Gelato web page. It's possibly accurate, but clearly incomplete. May I reccomend that you please correct it, but in the context of this article? Then, please create a Gelato Federation article. and we can link to it. State you bias on the talk page of the new article and request a POV review. I for one will certainly try to help. Could you also comment on this new version of the Itanium article set? Thanks. -Arch dude 12:00, 17 April 2007 (UTC)
 * update - since you do not have a Wikipedia login, I have created a stub for your article. If you do not have time to fill it out, I'll add some general stuff from the web site tonight. -Arch dude 14:25, 17 April 2007 (UTC)

spin: most vendors? increasing deployments?
I'm going to pull two spin-laden not-overly-factual statements.

Article says: By 1997, most enterprise systems manufacturers (with the exception of Sun) were designing systems based on the Intels's projected IA-64 processor

Besides the two grammar errors, DEC definitely wasn't, IBM was still quite devoted to the POWER2, and SGI was still mostly bound to MIPS, this would mean that only one of the top five enterprise vendors was taking the platform seriously, that being the one that invented it in the first place. Yes, IBM and SGI eventually shipped Itanium machines, but it's unclear that they were serious about it in 1997.

Article says: Deployments have increased steadily since its introduction.

What the heck does this mean? That they still sell some of them? That initially deployed units are now being re-deployed to trash heaps? -- Akb4 04:50, 23 April 2007 (UTC)
 * The 1997 statement was unsupported, so you were correct to remove it. However, I think is was broadly accurate. I will do some research. IBM and SGI both released Itanium systems in 2001 that included their own custom chipsets: A chipset is a long lead-time design. DEC was almost gone. The "Deployments" verbiage was my attempt to balance the negative tone of tone of the article. I initially used "sales" instead of "deployments," but many of the initial systems appear to have been given away. Sales increased year to year from 2003 through 2006, to about $2.4Bn. -Arch dude 09:28, 23 April 2007 (UTC)

Change name to "Itanium Processor Family"?
I was thinking that calling this article "Itanium" makes it seem like it just talks about the original Itanium, and not the ISA and Itanium 2. I was thinking that maybe it would be better to call it "Itanium Processor Family"? What do you think? Imperator3733 20:16, 25 April 2007 (UTC)
 * I considered that. Take a look at NapaliRoma's comment in the top comment block above. He stated back in February that if we were to do a merged article, we should first engage in a heated debate about the correct name. I then created the new merged article and asked for comments for a four-week period before finally replacing the article. I guess I should have waited for the heated debate on the name first: now we must do it in the wrong order. :-) OK here is my reasoning: the word "Itanium," when used in the press, can refer to any member of the family, or to the architecture, or to some related subject. Thus, a random non-specialist reader of a general-purpose encyclopedia is likely to type in "Itanium" What should this "intelligent 12-year-old" see? the lead paragraph is intended for just such a person, and that paragraph tries to very concisely give all of the meanings (brand, processor name, and architecture name) of the word, while also putting them in context. If you wish to move this to "Itanium processor Family," then what will the lead paragraph be, and what will the lead paragraph of the "Itanium" article be? May I reccomend that you create drafts of the two new lead paragraphs? I think that we could then discuss them here, and perhaps improve them collaboratively as part of any name change. Or suggest any other method to collaborate on this. Note that this is my third attempt to tidy this set of articles up. Any comments on the quality and suggestions for improvement would be greatly appreciated, (or, just edit it if you prefer.) -Arch dude 21:37, 25 April 2007 (UTC)


 * Hey Arch dude -- sorry it's taken me so long to say so, but great job on cleaning/merging the article! (and apologies that I had no time to contribute or comment as you did so, but I did add a few items today.)
 * Not so surprisingly, I would agree that there's no particular benefit to renaming the article at this point. But if anyone feels all that strongly about it, they could certainly add a "Itanium Processor Family" redirect page.--NapoliRoma 23:11, 25 April 2007 (UTC)

Dedicated Windows Longhorn Itanium Edition
I was on the Longhorn Beta website and it would appear the Microsoft has developed an Itanium Based version of its Server. "Windows Server "Longhorn" for Itanium-based Systems This edition is designed for use with Intel Itanium 64-bit processors to provide web and applications server functionality on that platform. Other server roles and features may not be available." Thought this might be of interest to you all. 205.200.147.100 15:24, 3 May 2007 (UTC)Jon

Failed Passed "good article" nomination
This article failed good article nomination. This is how the article, as of May 14, 2007, compares against the six good article criteria:


 * 1. Well written?: Yes
 * 2. Factually accurate?: Yes
 * 3. Broad in coverage?: Yes
 * 4. Neutral point of view?: Yes
 * 5. Article stability? Yes
 * 6. Images?: Several images do not contain a fair use rationale, so this article can't be passed.

When these issues are addressed, the article can be resubmitted for consideration. If you feel that this review is in error, feel free to take it to a GA review. Thank you for your work so far. — ~ G1ggy!  Reply 00:02, 14 May 2007 (UTC)
 * after addressing the problems with hte images, I have re-nominated the article for GA status. The problems were significant the the reviewer was quite right to raise them. I added the prpper "fair use rationales" on the two logo image pages, and I removed the improper justification on hte picture of the Itanium 2, tagged it as a copyvio on its image page, and removed it from this article. If Anyone has a digital camera and access to an Itanium 2, PLEASE take a picture of it and upload it to Wikicommons: ask me for help if you need it. Thanks. -Arch dude 23:31, 14 May 2007 (UTC)
 * OK, it has passed GA, after passing the images criteria. Congratulations! ~  G1ggy!  Reply 23:45, 14 May 2007 (UTC)
 * Yea, zoooooooooooom! Right through GA. Congradulations! This article clearly is the result of hard work. -- Jreferee (Talk) 23:18, 15 June 2007 (UTC)

Bus speeds correct?
In the processors table the mckinley has a 200MHz Bus and Montecito 533MHz - this imples a jump of more than 2 in bandwidth between the two which is very misleading. I suspect the frequencies listed aren't comparable or plain wrong. The original mckinly had 128 bits * 400MHz = 6.4GB/s of system bandwidth.
 * I'm fairly sure these numbers are correct. I just did a google for itanium montecito 533 and another google for itanium mckinley 200 and did a spot-check of the hits. The numbers are supported by the articles. Note that Montecito is newer than McKinley by four years and two process generations. Also note that much of the criticism of the early Itaniums centered on the inadequate memory bus speed. Furthermore, a Montecito needs a faster bus because it has two cores. I did not add a reference for each reference in the table, as it would add too much clutter. I created the table from information that was in the earlier versions of the Itanium and Itanium 2 article when I did the merge. If you find references that show that the numbers are wrong, by all means fix the article. -Arch dude 01:04, 29 June 2007 (UTC)
 * As a cross-check, based on the List of Intel Xeon microprocessors, the Xeon in June 2002 had a FSB speed of 400MT/s, and the Xeon in June 2006 had a FSB speed of 1066 MT/s, so the two processors are on similar technology curves. (Note: nomenclature for FSB speeds is not well standardized.) -Arch dude 13:33, 29 June 2007 (UTC)
 * The original Itanium 2 (McKinley) used a 400MHz bus (200x2, or 100x4). Montecito uses a 533MHz bus (266x2 or 133x4).  Imperator3733 16:12, 11 September 2007 (UTC)
 * So in that case as the 533Mhz stands for 17GB/s, does the 400Mhz of the original Itanium 2 (McKinley) than stand for 12,8Gb/s? Massimo2007 12:18, 12 September 2007 (UTC)
 * I don't think so. I believe that the Itanium bus is 128-bit (or 16 byte).  16 bytes * 400 MHz = 6.40 GB/s and 16 bytes * 533 MHz = 8.53 GB/s.  I'm not sure, but I think that those are the correct BWs.  Imperator3733 16:12, 12 September 2007 (UTC)

If you check the Montecito (processor) on it, it will tell you if I´m correct that there's a dual 533mhz FSB (2x128bit @17GB/s) available per node (4dies/8cores) or `system level`. This actualy means that every die or cpu has access to 4,26GB/s on a `full die` node. To me that´s very hard to belief and therefore would be nice to have some confirmation on that as well.. —Preceding unsigned comment added by Massimo2007 (talk • contribs) 08:41, 11 September 2007 (UTC)

Itanium logo
Itanium2 has been renamed into Itanium and the logo has been updated. The new logo is http://www.intel.com/sites/sitewide/pix/badges/itanium/itp_62.gif Old logos should be replaced with new logos according to http://en.wikipedia.org/wiki/WP:LOGO —Preceding unsigned comment added by 71.174.127.189 (talk) 05:49, 6 November 2007 (UTC)
 * But in this case we really need to update the article to account for all of the events of the last two weeks. We should not just relpalce the lead Logo, we should move the old logo down to the correct spot in the History, change the name on the lead infobox, anc change all references to the oritinal (2001) from "Itanium' To "original Itanium." We also need to convert from "Montecito" beingthe current Itanium to "Montvale" being the current Itanium. I will do this on  hteweekend if obidy does it first. -Arch dude 14:58, 6 November 2007 (UTC)
 * Made preliminary changes including the new logo, But we still have work to do distinguish "original Itanium" from this new usage. -Arch dude 14:34, 7 November 2007 (UTC)

Hardware branchlong?
In the table of processors, there is a reference to McKinley having "hardware branchlong". This page provides no explanation of what that is, and Google was not very helpful other than to find another wiki that pulled an old version of the Itanium 2 WP page (which now redirects here) and indicates branchlong is an IA-64 instruction of some sort. So does anyone have any further information? — Aluvus t/c 02:09, 2 December 2007 (UTC)


 * When I did the merge and rewrite, I collapsed a long set of short paragraphs into this tabular format. Each of the short paragraphs had (some of) the information that ended up all but the last column. In a few cases there was some very brief additional information. I mindlessly put that additional information into the last column. "Hardware branchlong" was not available in Merced.I got the impression that the instruction was therefore synthesized, probably via a trap. The real instruction was added to the architecture starting with McKinley, and it exists in all subsequent Itaniums. I cannot recall where I saw this described, but I did check it at that  time. If I can find the reference, I will put it into the "architectural changes" section. If you find it first, may I recommend that you put it there? -Arch dude 13:29, 2 December 2007 (UTC)


 * Just an update: some research for another project turned up more information about this instruction. It is used to allow branches to make large jumps (skipping a large number of instructions), rather than making multiple short jumps to accomplish the same result.  The mnemonic is brl.  It is described on page 69 of the Itanium 2 reference manual.  I am not sure how much of this information is worth including in the article, but in any event this takes a lot of the mystery out of this instruction. — Aluvus  t/c 02:33, 27 February 2008 (UTC)

MiB and KiB v.s. MB and KB
Note. For interested authors, debate and a vote is ongoing on Talk:MOSNUM regarding a proposal that would deprecate the use of computer terms like “kibibyte” (symbol “KiB”), “mebibyte” (symbol “MiB”), and kibibit (symbol “Kib”). It would no longer be permissible to use terminology like a “a SODIMM card with a capacity of two gibibytes (2 GiB) first became available…” and instead, the terminology currently used by manufacturers of computer equipment and general-circulation computer magazines (“two gigabytes, or 2 GB”) would be used. Voting on the proposal is ongoing here. Greg L (my talk) 21:20, 27 March 2008 (UTC)

Timeline format
I reverted a substantial set of formatting edits made by an anon using 68.148.164.166. The edits were a good-faith attempt to improve the article, but I arrived at the original format after quite a lot of thought and experimentation, and I personally like the old way better. If there is a consensus that the new format is better, we can revert my revert. Please discuss here.

Basically, timelines are generally deprecated. I retained the timeline originally because I thought it was the best way to succinctly express the long and twisty history that is the true story of Itanium. However, the newer format caused a major visual expansion to the timeline, giving it undue weight. Worse, the new format, which uses additional levels of TOC headers, caused a huge expansion to the TOC, Which destroyed the continuity from the lead to the second section. The TOC expansion also gives a casual reader the impression that the TOC is the bulk of the article.

The best solution would be a timeline using the magic timeline template, but that template is very hard to use. -Arch dude (talk) 17:51, 26 April 2008 (UTC)

gigabyte (GB)
This incorrect statement that 1 GB = 2^30 bytes has been repeatedly inserted into this article, despite my attempts to remove it. Please stop. Thunderbird2 (talk) 14:42, 14 June 2008 (UTC)

To Arch dude: I see that you have re-inserted the binary definition for GB. The only use I see of gigabytes in the article is in the data transfer unit GB/s. I changed this to its decimal meaning because that is the correct convention for data rates. Are you saying that the binary gigabyte is used here for these data rate values? Thunderbird2 (talk) 10:13, 15 June 2008 (UTC)
 * Hi, T-bird. I changed it in the footnote to keep the footnote self-consistent. As you are painfully aware, the entire Gigabyte versus Gibibyte situation is a total mess. I very carefully converted this article to KiB and MiB a year ago pursuant to the then-current WP:MOSNUM, even though it looked really strange to me, and I defended this change against opposition. WP:MOSNUM has now changed. I don't particularly like the change, either (I would prefer a user preference.) I think consistency in the footnote is less bad than the ambiguity. I will add GB/s in the footnote. -Arch dude (talk) 12:06, 15 June 2008 (UTC)
 * I added GB/s to the footnote and referenced it from the first instance of GB/s in the article. -Arch dude (talk) 12:15, 15 June 2008 (UTC)


 * User preference would definitely be awesome. {{#if:|{{#if:|$$}}{{#ifeq: {{{anti}}}|yes|[[User:Headbomb|Headbomb { }}|Headbomb { }}{{#if:— The greatest sin is willful ignorance.|$— The greatest sin is willful ignorance. — ταλκ / κοντριβς/Projects of the Week$}}]]|{{#if:|$$}}{{#ifeq: {{{anti}}}|yes|$\overline{Headbomb { }}|Headbomb { }}}$ 12:31, 15 June 2008 (UTC)
 * To Arch dude: That's better, thanks. (I have removed the binary definition as it wasn't used anywhere)
 * To both: Yes, a user-preference might solve some of the problems.
 * Thunderbird2 (talk) 14:27, 15 June 2008 (UTC)

Itanium processors need an article seperate from an article about (ia-64)
There is apparently some widespread confusion between x86-64 and ia64 ... Based on this very article (or worse, based on the Core 2 article) one might not realize that IA-64 (which redirects to this article about the "itanium" processor) is an instruction set, and is used on core 2, centrino, and xeon as well.


 * intel's list of ia-64 based processors someone else will need to fix this as I'm pressed for time and must log out very soon.


 * Arg, this sort of misinformation makes wikipedia really unreliable for software porting usage ((end of line)) Kuzetsa (talk) 22:57, 16 June 2008 (UTC)


 * Seems my cookies acted up again... Decided to re-sign my own post using my proper login. Kuzetsa (talk) 23:04, 16 June 2008 (UTC)


 * You should better get your data sorted, you are mixing IA64 (Itanium) with x86-64 (AMD 64) which Intel labeled as Intel 64 (previously EM64T).

Please read the article carefully: it is correct. The IA64 instruction set implemented on Itanium is radically different from the x86 and x86-64 instruction sets. Itanium an Itanium 2 are the only processors in the whole wide world that implement IA64. By contrast, x86 was first implemented on the 80386 and has been implemented on about eight successive generatin of Intel priocessors, on five generations of AMD processors, and on processors from VIA, Cyris and Transmeta. Starting in 2003, AMD implemented a 64-bit superset of x86 that they called AMD64. Intel implemented a vert slightly different superset called EMT64. The industry uses the generic term x86-64 to refer to both: neither of then has anythigh whatsoever to do with the IA64 instructins set, which was developed in 1994. Please carefully read the Itanium article and the x86-64 article. Xeons and Core 2 implement x86-64. They do not implement IA64. -Arch dude (talk) 03:54, 17 June 2008 (UTC)

Should MOSNUM continue to deprecate IEC prefixes?
A discussion has been started at WP:MOSNUM concerning the continued deprecation of IEC prefixes. Please comment at the MOSNUM talk page. Thunderbird2 (talk) 19:19, 5 July 2008 (UTC)

Reasons for adding POV tag
This page used to include extremely intelligent and factually correct technical criticism of the Itanium. For no reason that has been explained, all criticism has been removed. I am simply asking someone to explain why descriptions of the the technical shortcoming of the Itanium can not be allowed? Surely the utter failure in the marketplace of the Itanium needs some kind of explanation? To give only one side of the story is self evidently POV. Timharwoodx (talk) 20:24, 17 August 2008 (UTC)

Early Technical Challenges

The first versionof the Itanium processor, code named Merced, shipped in June 2001. Manufactured in a 180 nm process, it was offered at speeds of 733 and 800 MHz, with a choice of 2 MiB or 4 MiB off-die L3 cache.

In IA-64 mode, it was the fastest floating point processor available in the marketplace. However, on integer calculations it performed only slightly better than an equivalently clocked x86 design, and when running legacy x86 code, performance was only about 1/8th that of a similarly clocked x86 processor. Software emulation would have been faster as demonstrated in the Itanium 2 processor, for which legacy x86 code runs at the level of a similarly clocked x86 processor.

The main structural design flaw with the initial Itanium processor was the high latency of its level three cache. Intel's engineers had evidently been hoping that the amount of bandwidth available would offset this, but the latency was so high that it actually slowed the cache, to the point where it was not significantly faster than the main memory interface. With the faster first and second-level caches set relatively small (32KiB and 96KiB respectively), this further increased the load on the main system bus.

Compounding the performance impact of the lack of available cache bandwidth was the fact IA-64 code has a larger footprint than x86 code. So the number of instructions that could be contained in the cache was in fact even smaller than the sizes alone would suggest, all of which might have been mitigated had Itanium been designed around a fast processor bus. However, at a mere 266MHz it was only equal to consumer Athlons of the period, and a full 33% slower than first generation Pentium 4s. Again, this was worse than it would appear, due to the fact that Itaniums were designed to be used in systems with several processors. Itanium clock speeds were also disappointing, relative to the GHz speeds being delivered by the x86 architectures of the period.

Overall, it is generally believed that the technical specifications indicate an original 1998–99 target launch date. But the repeated and lengthy project delays effectively meant the processor was out of date before it had even begun shipping. Hence, the Itanium was not a competitive product when launched, although it would have been two years earlier.

Despite these intial shortcomings, Intel and HP continued their development work. Merced was succeeded by the Itanium 2 (code-named McKinely), which delivered competitive and in many cases leading performance across a range of industry benchmarks, as well as many technical and commercial workloads. Since that time, the Itanium processor has continued to deliver competitive performance.


 * The article is currently in substantially the same form it was in after I consolidated three articles and substantially re-wrote the result. That was more than a year ago. The original articles, taken together, looked like Intel/HP press releases, and I was disgusted by the extreme bias in favor of the Itanium. The rewrite was intended to remove the extreme pro-Itanium POV. It is possible that I did not go far enough because I was overcompensating for my feelings about this steaming pile of dung. One of the things I did during the rewrite was to attempt to find reliable sources for all assertions. That's why the article now has more than 50 references instead of almost no references as before. I removed statements for which I could not find references. Some of the removals were in response to a (failed) Featured Article review.  If you would like to re-integrate some or all of the material you feel was improperly removed, please feel free to do so, but please provide references. I would be more than happy to help. I did cut down on the material in the "original Itanium" article, but that was because it is a relatively small part of the whole Itanium story as a whole. I hope that the sales projection chart and the timeline convey the true nature of the Itanium tragi-comedy. -Arch dude (talk) 21:20, 17 August 2008 (UTC)


 * (After some brief research) The material you copied above was present in the "Itanium" article as of this version, in October 2006, just before I started messing with the articles. Note that there were no references. After working on the articles off and on, I finally completed my merge in May of 2007. I just did a cursory check, and I appears that all of the material appears in some form in the merged article, but not all in one place, because I re-organized the article as part of the emerge/rewrite. As far as I can tell, the only material that did not survice is material for which I could not find a specfic reference. Over the course of many edits, I made the material more concise. This was not an attempt to soften the criticism: In fact, I believe the "Original Itanium" section is fairly damning and conveys the intent of the non-architectural part of the older material. -Arch dude (talk) 23:09, 17 August 2008 (UTC)

I'm removing the POV tag, since we do not appear to have a "dispute" to "resolve." If anyone wishes to add referenced statements from the above text, please feel free to do so. or discuss here first. -Arch dude (talk) 00:25, 19 August 2008 (UTC)

Floating-point execution in Itanium
The "Instruction execution" subsection under "Architecture", which I assume is about Itanium, describes it as having two FMAC units and "miscellaneous" units. My understanding is that it has two 82-bit FMAC units and two SIMD FMAC units that each operate on two single-precision floating-point numbers. Does this mean that the "miscellaneous" units are the SIMD FMACs? If they are, then they should be described as such instead of a vague mention with scare quotes. Rilak (talk) 08:16, 19 April 2009 (UTC)
 * Welcome to Wikipedis, the encyclopedia that anyone can edit. "Anyone," in this case, means you. Please feel free to read the Intel architecture document that is cited as a reference, and then update the article. The section as it stands today is mostly my work and is a major improvement over what it looked like before I started, but I'm not an expert. I merely spent about half an hour reading the refreence, and then tried to condense the relevant stuff into this article. You are probably at least as qualified to correct it as I am. If you have a conflict of interest or prefer not to edit the article, please suggest wording here, and someone will come along and fix the article. -Arch dude (talk) 15:07, 19 April 2009 (UTC)
 * Thanks for the welcome, but I have been editing for quite a some time (check my contribs). The reason I am not editing was because I don't believe I have sufficient understanding of the Itanium's microarchitecture. I'll get some research done, and fix the article. Rilak (talk) 07:59, 22 April 2009 (UTC)
 * Done. Rilak (talk) 08:23, 23 April 2009 (UTC)

Binary Prefixes for K(i)B etc.
I edited the page to give consistent use of KiB etc. when I saw it. Someone later told me that there have been edit wars on the subject, and that the consensus keeps changing. So let me justify my edit: the use of footnotes to explain what KB means is just lame prose. The term KiB means that, and it can be hyperlinked so those unfamiliar with it can just click it right there! I made sure at least one instance was linked in each section.

Elements of the CPU architecture such as the cache are technical items, so readers should not have any trouble with it if they understood the paragraph anyway.

If there is still concern over using terms introduced to solve a specific problem, perhaps we need a template that will expand out to the reigning approach. If avoiding the new terms, it would need to use the suggested parentetical comment of spelling out the meaning in-line.

Also note that MHz and MiB are in close proximity, so one article-wide definition of what "mega" means will not work cleanly.

Długosz (talk) 22:43, 31 August 2009 (UTC)
 * Lets us not argue it here. Instead, Please argue the on the talk page of WP:MOSNUM. I think we can agree that the Itanium article should adhere to to WP:MOSNUM, even if we do not agree with the consensus there. There have been 16 separate extensive (heated) discussions there, the first in 2005 and the latest in 2008. I personally do not care either way, but it is extremely irritating to have to re-edit this article to keep up with changes in the MOSNUM. I agree that we should template this so we can let the fanatics change all references in a single place. this entire issue is WP:LAME to the extreme. The current consensus(?) is that the overwhelming majority of references use kB/MB/GB to meand 210/220/230, and that the overwhelming majority of readers understand these terms but do not understand KiB/MiB/GiB. I propose that we add templates, , 🇬🇮, etc, that use two variables: #USE_IEC and #EXPAND_IEC. The two variables are not set by default. If #USE_IEC is set, we use KiB, etc. If not set, we use kB, etc. If #EXPAND_IEC is set, we add the parenthetical decimal expansion (1024) or whatever. If not set, we do not. For now, we will set (or not) these variables in another template, , which goes at the head of the article, but a user may also set these in their CSS. The varous templates can be made to also add a footnote. -Arch dude (talk) 01:10, 1 September 2009 (UTC)