Jump to content

Talk:Itanium 2

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Corporate interference with this page

[edit]

Recently, someone from an IP belonging to Texas Instruments corporation posted some uninformative opinion on the article page. Things like "death of itanium" hardly belong in an encyclopedia until such time as they come to pass as historic events. Even then, it is highly unlikely that such text, however interesting it might be one day, belongs on a page dedicated to the [Itanium 2] processor and its variants. This stuff should go somewhere else, but for the time being I've just zapped it.


Intel to spotlight new Itanium: 'Poulson' (28 Feb 05) -- just trying to keep up with developments; ok to delete this when the news gets too old <>< tbc

Moved from article

[edit]

This unformatted unincorporated comment from somebody moved from article. Gene Nygaard 01:54, 4 May 2005 (UTC)[reply]

The above links to aceshardware.com are inacurat, they used data from SPEC(standard performance evaluation corperation) and manipulated it to support the mythe that you should think about performance per Ghz. yea it would be true that an itinium 2 would be faster than a single core POWER5, but IBM doesnt make single core POWER5's there all dual-cores. Intel cant compet on design so they have create this confusing bs. the itanium2's die is to large to go dual-core at 90nm, it's twice the size and uses twice the wattage allready compared to a POWER5 using 130nm. so at 65nm they'll try dual-core. but at 65nm IBM can put 4-6 cores of the POWER5 on a single die and stil be smaller than a dual-itanium2 and use less wattage.

these are more realistic.

SPECfp2000

highest score achieved by this processor POWER5 SPECfp 2796 and by this one itanium 2 SPECfp 2675 and this one opteron SPECfp 2045

SPECint2000

highest score achieved by this processor opteron SPECint 1792 and by this one itanium 2 SPECint 1590 and this one POWER5 SPECint 1542

How is "Merced" pronounced?

[edit]

I have tried looking up the California town, too, for guidance on pronunciation, but no joy. Nonky 22:08, 10 January 2006 (UTC)nonky[reply]

It's pronounced like "mer-said". There is also a WAV file on m-w.com -- Bovineone 02:15, 11 January 2006 (UTC)[reply]

Thank you so much! Nonky 18:34, 11 January 2006 (UTC)nonky[reply]

Moved comparison with Cell to here

[edit]

I deleted the following from the page:

The latest research in the field of HPC showed that Itanium 2, along few others among highest-class processors, only concedes to Cell, when compared by performance. Cell is capable to perform any type of double precision scientific calculations 3 to 12 times faster than any desktop processor, including Itanium 2 1.4 and Opteron 248 [1]. Single precision performance (which is best fit for genetics, visualisation and scaled-precision tasks) of Cell dominates over competitors by order.


I deleted it for several reasons. The main reason was that the reference is to a paper citing research in 2004 taht compared a theoretical Cell implementation wiht real Opteron and Itanium processors of the time. The second reason is that even if it is valid to put processor comparisons in hte individual processor article, it is certainly not reasonable to put them in the lead paragraph. The third reason is that the paragraph is poorly formated and points to disambiguation pages rather than to the correct articles. The fourth reason is that Cell is not yet available in large quantity for real head-to-head comparison with current-generation Itaniums.

I do not like the Itanium. It's a dog, and its gross under-performance and exceedingly late delivery had sereral major detrimental effects on the entire industry. But just because HP and expecially Intel were guilty of severe marketing hype is no reason to hype in the other direction here on Wikipedia. Afer we get real number for real Cells, perhaps we can put a comparison back in. However, I think a separate comparison article would be better.-Arch dude 00:42, 8 November 2006 (UTC)[reply]

First of all, the reasearch is not from 2004, but from 2006. Second, it concerns not only to theaoretically calculated performance results, but actual chip perforance also -- see tables where both theoretical and practical Cell results published. So I am going to publish the mention of comparison back. Also, it is very strange that You do not like this comparison to be in the lead paragraph, while Intel's own tests are there and You somehow all right with that. Even more strange -- and this is against Wiki's principles -- that You prefer independent Berkley research to be edited out, while Intel's own one to be left here.
If You want to move comparison to some lower paragraph, I do not mind. But this should be done simultaneously with moving refernces of Intel's tests. 87.237.114.24 13:37, 12 November 2006 (UTC)[reply]
The paper you refer to was published in May of 2006 by ACM, but has copyright 2005 by LBL and reflects work that was done earlier. The tables you refer to generally have two columns: "CellFSS" and "Cellpm". FFS is IBM's "full-system simulator," while "pm" is the author's "Performance Model." Neither of these is the actual cell hardware. Please describe the table with the actual hardware. Am I missing something? -Arch dude 21:24, 19 November 2006 (UTC)[reply]

In addition to the four reasons I listed above, this comparison is inapprioriate for quite a different reason. Itanium 2 is a general-purpose processor that can run standard softwrae without source-code modification. Cell is a special-purpose system with a mix of standard and special-purpose processors. Software must be refactored for Cell to get the performance gains you discuss.Speaking personally, I think we will see a cell-based supercomputer in the near future that will utterly crush the Itanium supercomputers, but this result will be reflected in the portion of the Itanium article that relates to supercomputers.-Arch dude 21:24, 19 November 2006 (UTC)[reply]

Your reference paper compares a Cell to a mono-processor Itanium of the same clock rate. But Cell has nine processors on one die. If we allow multi-processor dies (or modules) Then a theoretical Quad-processor Itanium would win over a Cell based on this paper for 64-bit floatiing-point applications. My personal opinion is that a large class of HPC problems can be re-factored to take advantage of Cell's massive internal interconnect bandwidth, but this will show up on the "Top 500" list. -Arch dude 21:24, 19 November 2006 (UTC)[reply]

Reads Like an Advert

[edit]

"Competitive performance", "...doubles the performance..." and other passages throughout this article... it all reads like an advert. I'm about to do some rewriting, but I don't know enough about Itanium processors to go through it thoroughly, and I'm probably going to leave the advert tag on it. Also, I'm adding some more links to articles, as this article lacks them. Sloverlord 02:59, 26 November 2006 (UTC)[reply]

Thanks! Yep, it still reads like an advert. I've been adding facts occasionally to counter this, but it's too depressing, and I'm not an expert either. I did adjust your "parallel" links a bit. There are many levels of parallelism in computing, and you picked the wrong one in this narrow case, although I think the original text helped confuse you. The level we need here is instruction-level parallism. IA-64 uses EPIC, while most others use a superscalar approach. -Arch dude 17:12, 26 November 2006 (UTC)[reply]

Request for comment: consolidated article

[edit]

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.

The form of the citations still needs work, and we need still more citations. We are still missing a lot of history from the timeline,(mostly 1994-1997) and we are missing a paragraph in the "History" section abou tthe period from 2002 to present.However, this is material that is not present in the current articles either.

Thanks.-Arch dude 23:07, 19 March 2007 (UTC)[reply]

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 18:27, 30 March 2007 (UTC)[reply]