Talk:Yonah (microprocessor)

Yonah and Jonah
Currently, the article reads "...codenamed Yonah (Hebrew transliteration for Jonah)..." Actually, Jonah is the English transliteration for Yonah. Yonah is a name mentioned in Hebrew in the Torah, a language thousands of years older than English.

Faulty Core Duo sold as Core Solo
I read somewhere (can't remember the source) that Core Duo with faults in one core might be sold as Core Solos (seems a sensible idea in terms of wastage). If anyone knows about this, perhaps it could be added to the article. CharlesC 12:43, 12 February 2006 (UTC)


 * It's been a fairly commonly known policy of Intel's to take chips with slight imperfections in certain areas, block off that area, and sell it as a 'lower-end' model. This doesn't mean that EVERY one of the lower-end models is this way.  For example, the original Celeron processor was just a Pentium II without the external cache chips.  Later, the Celeron became Intel's first chip with an on-die L2 cache.  Even later yet, when the Pentium III got an onboard L2 cache, larger than the Celeron's; it was pretty much known that if Intel found a fault in part of the L2 cache of a Pentium III, they just blocked off that part of the cache, and called it a Celeron.
 * In those cases, I actually find this explanation unlikely. At the time, L2 cache was still a small physical area of the processor.  It was more likely to find a flaw in the main processor core, making the processor unusable, no matter what.  So claiming that they would take the time/effort to screen the cores, and then set aside the 'bad' ones for further testing and relabelling as a low-end processor seems silly.  It seems much more likely that they just 'lopped off' the extra L2 cache during main manufacturing to produce a Celeron.
 * In the case of designs, where the Celeron's cache is SIGNIFICANTLY smaller than the Pentium 4's, this approach seems more plausible. But, I still think it unlikely.  Why waste time/effort re-testing a chip and re-validating it when it's only going to end up as a low cost chip, anyway?
 * As for the Core Solo and Core Duo? Well.....  We run into the problem that the Core Duo is manufacturered as a single coherent unit.  It's not the like the current design of Pentium D, which you could theoretically lop in half and have two fully functional Pentium 4's.  The Core Duo is a single 'processor' with two cores.  They can't be split apart.  While in this design, we have the best possibility yet for such a move, I still really doubt that even if Intel is doing this, that any significant number of Core Solos are really 'failed Core Duos'.  (Intel has an amazing fab system, the lowest rate of bad chips of any major company.)
 * What it all boils down to is that for this to be the case, Intel would have to spend extra effort re-testing a chip, and then using extra post-manufacture measures to 'downgrade' it. This seems like a lot of effort for a chip that will end up being the 'low end' chip.  If the end result was still a high end chip, it would make more sense (like taking a high-end Xeon processor, and, upon failing stability validation, declaring it a Pentium Extreme Edition chip at lower speeds.) Ehurtley 07:54, 23 March 2006 (UTC)

This is absolutely true and is mirrored on several websites. In addition, all it takes is close inspection of a Core Solo chip to reveal that both execution cores are still present. In response to Ehurtley, the market for Core Solo is indeed small enough that it can be supported by the relatively low number of Core Duo manufacturing failures. It makes sense for Intel because they can still make some income off what would normally be a total loss. - Aanhorn

Labels

 * Intel Core, Core Duo, and Core Solo are umbrella labels and not particular architectures like Banias, Dothan, Yonah, Northwood, Prescott, Conroe, Merom, Woodcrest, etc. This article should be changed to reflect that fact.
 * I think Intel Core is just for the mobile processor line. I haven't seen anything associating it with the Conroe desktop line or the Woodcrest server line yet. Frankchn 09:29, 2 January 2006 (UTC)
 * Agreed. All official information only refers to the Core brand being used for Yonah and its derivatives. The only places I've heard otherwise are articles speculating what the changeover from Pentium M to Core might mean for Intel's future desktop brands. Jgp 02:59, 9 January 2006 (UTC)

Merging Core Solo and Core Duo

 * Personally, I support merging the Core Solo and Core Duo articles into this one. Not only are they redundant, but those two articles need massive cleanup. Both of those articles are filled with grammatical errors and horrible wording. There are run-on sentences, sentence fragments, misspellings, and some of the worst phrasing I've seen in my life (e.g. "the target for the mirgration of"). Even if there was no separate Intel Core article, I'd support deleting those two articles just to save Wikipedia's integrity. Jgp 02:59, 9 January 2006 (UTC)


 * I think that the articles should be together. It is one family of processors. Hohohob 06:34, 11 January 2006 (UTC)


 * I also agree that the articles should be merged. It's even the same die/design. Hachu 22:34, 11 January 2006 (UTC)


 * Agree, merge them. Ehurtley 19:56, 12 January 2006 (UTC)


 * I agree, merge them, especially since the Core Solo is just one Core Duo CPU core and there are otherwise no differences in features or architecture of the chips. Pgk1 15:59, 15 January 2006 (UTC)


 * Agree, merge them. Guidofd 18:45, 15 January 2006 (UTC)


 * Agree, merge. PoorLeno 19:43, 15 January 2006 (UTC)


 * I agree. Mushroom 20:22, 15 January 2006 (UTC)


 * I see no need for separate "Intel Core", "Core Solo", and "Core Duo" pages, at least for now. I'm not sure what Intel's plans are for the "Core" name once they come out with the Intel Next Generation Microarchitecture chips - if they just call them "Intel Core XXX" with nothing in the name to distinguish them from the Yonah processors, then this page would be somewhat like the Celeron page, as Intel's used that name for chips with different microarchitectures (e.g., P6, NetBurst, and Pentium M Celerons).  I guess I'd vote "merge", and worry about the NGMA chips when they come out. Guy Harris 23:20, 29 January 2006 (UTC)


 * Done. Feel free to purge, fix, or otherwise run through the food^H^H^H^Hword processor any of the stuff I've picked up from the Core Solo and Core Duo pages (I did, at least, avoid picking up "the target for the mirgration of").  If any NGMA chips come out as "Intel Core", they can be added into their own sections after the "Yonah" and "Sossaman" sections. Guy Harris 09:59, 30 January 2006 (UTC)

Dual core?
Are these real dual core CPUs like Athlon X2 or just 2 dies packaged into one like Pentium D? Personally I believe is the latter since both versions look exactly the same, and Intel could have easily disabled a core in core dual that have artifacts and brand it as core solo, thus increase the yield. This is something you cannot do with an Athlon X2 or other real dual core CPUs. -- antilived T 02:27, 10 January 2006 (UTC)


 * Dual core CPUs are by definition two cores in one package. Whether they are on individual dies or on the same die, it is not difficult to design a die to enable disabling a core by hardware (fuses) or software (flags). Afterall, this feature is important for being able to power save by switching off a core. AMD may have chosen to not market a "X1" but it's definetely possible to do it. Hachu 22:33, 11 January 2006 (UTC)


 * It is a "real" dual core CPU instead of a MCM (see die images floating around the web). 61.14.88.87 08:49, 14 January 2006 (UTC)

Die Size
This is important information, someone please add it.the1physicist 02:25, 12 January 2006 (UTC)


 * Die size on the Core Duo is 90.3 mm^2 (151.6M transistors). I would think that the Core Solo would be similar as they would likely be Core Duos with one faulty core. Pgk1 16:12, 15 January 2006 (UTC)

USB Power Consumption Bug
from everyone's favorite site I just thought everyone might want to know.the1physicist 02:27, 30 January 2006 (UTC)
 * Thiss seems to have moved to http://tomshardware.co.uk/2006/01/28/toms_hardware_uncovers_power_drain_issue/ Plugwash 02:59, 30 November 2006 (UTC)

32-bitness
"the Core Duo is a 32-bit processor, though this means little in terms of actual performance"

I don't think that this necessarily particularly accurate, and probably shouldn't be here. x86_64 reveals 8 extra general purpose registers, and can make a very significant performance difference under certain circumstances. Some applications, in particular databases, also benefit greatly from 64bit arithmetic ops. Rsynnott 19:10, 30 January 2006 (UTC)


 * I removed everything past "the Core Duo is a 32-bit processor"; a broad claim that it "means little in terms of actual performance" is questionable, even if for most of the apps that would be run on Core Duo machines the performance improvement would be small, and the bit about addressing extra memory is also a bit broad (you don't need 64-bit virtual addresses to handle more physical memory, although you do need it to handle the additional memory in a single process without mapping stuff into and out of the address space). Guy Harris 22:50, 30 January 2006 (UTC)

Binary unit prefixes
I recently made some edits to change some unit prefixes from decimal-based multipliers (e.g. M, G) to binary-based multipliers (Mi, Gi).

This is in line with WP:MOSNUM.

The values in question were memory sizes. Memory hardware nearly always exists in sizes which are a power of two, or a power of two with a few other prime factors.

In particular:
 * CPU L2 cache will be far more likely to be 2 MiB (2 &times; 220 bytes) than 2 MB (2 &times; 106 bytes), and
 * addressable memory space will be 4 GiB (4 &times; 230 bytes, i.e. 232 bytes, a result of having 32 bit address values), and certainly not 4 GB (4 &times; 109 bytes).

My edits were made with a proper edit summary, referencing WP:MOSNUM.

Some of those changes were reverted, presumably well-intentioned, but without any edit summary.

These reversions would therefore appear to be unjustified. Accordingly I have reverted them speedily.

I have also placed comments into the wiki markup to alert any other potential editors to this issue.

Please discuss here, and obtain consensus, before making any further changes on this issue.

Thanks, Duckbill 16:39, 22 February 2006 (UTC)


 * According to WP:MOSNUM, "The use of the new binary prefix standards in the Wikipedia is not required". Furthermore, such prefixes are rarely used and only cause confusion for most people. No one I have interacted with uses such prefixes (I am a Computer Science student at a major CS-centric university; no one, not even the professors, use the IEC prefixes). I very strongly believe that common usage should trump pedantry, especially when the pedantic solution involves the use of terminology that very few people have heard of and even less people actually use. In fact, I'll say that throwing in new units that most people haven't heard of creates more confusion than the concept that memory should be based off powers of 2 instead of 10. Jgp 01:04, 24 February 2006 (UTC)


 * They may not be required, but I believe that they are preferable. Why preferable? They are more accurate, and I do not believe they cause confusion. The binary multipliers are deliberately made to be easy to understand by using the same letter as the decimal multipliers (so 'M' and 'Mi' have roughly similar values). The first use in any article is usually linked to an article explaining the meaning. I believe your case based on the usage of in a university may be flawed. You talk about the usage within a computer-science-centric university, especially by professors. That does not support your case but rather counts against it. In a very computer-science-centric context, the audience will likely include a high proportion of people who are able to disambiguate the specialist (mis-)usage of the decimal multipliers. However, the audience of Wikipedia is comprised of a wider cross-section of society. People outside the computer science community will not necessarily have the specialist skills needed to disambiguate the units. In essence, referring to 2,097,152 bytes as "2 MB" is jargon, i.e. subject-specific language, and Wikipedia should not use jargon unnecessarily. In this case it is wholly unnecessary, as there is a more accurate, widely-accepted term which is just as concise. Duckbill 13:15, 24 February 2006 (UTC)


 * I do not exclusively converse with those who are computer geeks. I also talk to many people who are barely computer literate, and literally everyone I've ever talked to (geeks and laypeople alike) says "gigabyte", "megabyte", "kilobyte", and the like. Furthermore, every source I've seen, including all manufacturer websites do not use the IEC prefixes. Everyone, including people who barely use computers, refers to memory in terms of gigabytes and/or megabytes. Most people have never heard of the IEC prefixes, and their usage will only confuse people. Real people do not use the IEC prefixes, and the use of IEC prefixes is jargon, and a particularly egregious form of it. Futhermore, I'll even go as far as to say that their use is intellectually dishonest. It's a made-up "standard" that isn't in common use, and using it as if it were common use is highly deceptive. [note: since I broke up your paragraphs, I copied over your signature, since they were originally part of the same edit--I hope you don't mind the change] Jgp 00:13, 28 February 2006 (UTC)


 * So, having them not required should mean that new contributions should not need to use the binary multipliers. However if other editors come along later and refine the content to use the proper binary multipliers, that updated content should be considered better than the original content. That being the case, those edits should not be reverted, especially in a way which would appear arbitrary (i.e. with no edit summary). Does that make any sense? Duckbill 13:15, 24 February 2006 (UTC)


 * Actually I don't need to argue my case at all. WP:MOSNUM says (my emphasis) "The use of the new binary prefix standards in the Wikipedia is not required, but is recommended for use in all articles where binary capacities are used." and also "If a contributor changes an article's usage from kilo- etc. to kibi- etc. where the units are in fact binary, that change should be accepted." Duckbill 13:35, 24 February 2006 (UTC)
 * Way to duck responsibility for your actions. For the record, "should" != "must". Jgp 00:13, 28 February 2006 (UTC)
 * Well then, let's get it changed to "must". This is an encyclopaedia, not a tabloid newspaper. 202.147.107.2 10:25, 11 March 2006 (UTC)


 * And there's now more jargon being added to this article. I have very serious issues with made-up jargon that almost no one uses being edited in to replace terms that almost everyone uses. This is intellectually dishonest, User:Duckbill. You're using made-up, highly obscure terms as if they were standard and editing them in to replace perfectly good, commonly-used terms. Please stop. Jgp 01:24, 9 March 2006 (UTC)


 * Au contraire. I replaced jargon with proper terminology. I can't see anything special about this article which makes it an exception from WP:MOSNUM on this issue. The proper thing to do in Wikipedia is to follow standard Wikipedia practice. If you don't like it, you are free to try to change it. Duckbill 22:48, 9 March 2006 (UTC)


 * WP:MOSNUM is objectively wrong then. Let's see--how many years were computers in common use before the IEC pulled those units out of thin air? They were introduced in 1999, at the tail end of the first dot-com boom, long after computers and basic terminology such as "Megabytes" and "Gigabytes" became commonly used by the general public. I talk to many people, both computer-literate and not, and none of them use that -bi- garbage. When I help people with their computers (including people who aren't computer-literate), I ask them how much RAM they have, and they reply with something along the lines of "It says I have 512 Megabytes", "I've got a gig", etc. I open up a newspaper and look at the ads: when memory is mentioned, units such as "MB" and "GB" are used. I bring up the website of any major electronics store (including stores that cater to common people and not to geeks), and they all refer to memory in MB, GB, etc. The binary unit prefixes are the very definition of jargon: it's something that's never used when communicating to the general public (and when the general public must be communicated to, alternative, more commonly-used terms are used instead), it's completely unknown except to computer geeks and professionals, and only a very small subset of those geeks and professionals actually use them. Jgp 00:15, 10 March 2006 (UTC)


 * The older terms, on the other hand, are in common use by everyone except a small group of geeks and pedants. Furthermore, the common person knows that words can have different meanings in different contexts. Here's an example: "I was just looking at my hand". Normally, this means that I was gazing at the appendage at the end of my arm. But if I'm playing cards, it means that I'm looking at the cards I was dealt. The same word, or even the same phrase, can mean something entirely different depending on the situation, and any intelligent adult can understand that. Jgp 00:15, 10 March 2006 (UTC)


 * Anyway, this page isn't the place. I'll cheerfully refine multipliers anywhere in Wikipedia, not just on this page. And I don't suppose I'm the only person following Wikipedia practice. If you're really not happy with the guidelines, get them changed, and I'll follow that. Duckbill 22:58, 9 March 2006 (UTC)


 * As you have requested me to "please stop", as a courtesy I should assure that I have read and considered your request. Wikipedia guidelines support my standpoint, and as far as I can tell there is nothing special about this page which means that the guidelines do not apply in this case. I see you are now in progress with debating the Wikipedia guidelines on this point, which is great &mdash; if what you're saying is right, it should come out in the discussion, and we may see a guideline change. If that happens, I'll follow that. But until that happens, I plan to continue following the current guidelines. Duckbill 16:06, 12 March 2006 (UTC)


 * I've been dealing with computers daily for 20 years. I've been in the 'computer industry' for 15.  I have never, not once, heard spoken out loud, any of the 'binary' prefixes.  Not once.  Heck, even the arguments I've had over it are all in email, blog, wiki, IM, etc.  Try going into a computer store (a real one, not Best Buy or CompUSA,) and asking for a "five hundred twelve Kibibyte memory module," or a "400 Gibibyte hard drive".  You'll get a funny look.  While more 'tecnically accurate' than using standard SI prefixes, the term just isn't catching on.  And I doubt it ever will.  Until it does, the decades old, (but, yes, technically incorrect,) SI prefix use is the more 'clear'. If someone adds it, I won't change it back, but I do think it's silly expend effort adding it.  I don't recall what the page has right now; but everyone just leave it there.  Unfortunately, again, this page isn't the place to debate it.  Ehurtley 07:36, 22 March 2006 (UTC)


 * Didn't this whole redifine MB as 10^6 rather than 2^20 start out as some sort of marketing ploy? I remember that the CD-R vendors started doing this because the decimal KB, MB, GB etc. is smaller than the Binary KB, MB, GB etc. and thus they could rack up the purported capacity of the CD-Rs and later products that they sell.


 * No. SI defined these prefixes to be powers of ten well over fifty years ago.  They were originally used in computing contexts as a rough approximation (210 is roughly equal to 103).  The problem became that higher valued SI prefixes came to be used to express higher powers of two, and the approximation error becomes more pronounced.  The IEC prefixes were created as a response to this problem.  They provide prefixes with unambiguous definition to use in lieu of ambiguously using SI exponents of ten prefixes. -- uberpenguin


 * Just checked with Britanica (The standard of English encyclopedias) and its articles appear to closely follow the traditional standards for the use of the words. For storage- binary kilobytes etc. are used for data transmition- decimal is used. The first time that decimal kilobytes etc. were used for starage was it appears to be a marketing stunt to exagerate sizes thinking they could use SI prefixes as defence against deceptive marketing lawsuits. This has not worked however as there have been lawsuits by users noting that the OS vendors use binary kilo- etc in their products, and the courts seem to agree. As long as operating systems makers continue to use Kilo- Mega- etc. for displaying storage statistics the encyclopedias should maintain that standard. It seems as though the Kilo- etc found in bytes are in fact NOT SI prefixes though they appear they use the same greek roots.


 * Replaced MiB with proper MB prefix. --129.116.46.127


 * Please stop vandalising the article, 129.116.46.127. MiB is the correct and unambiguous suffix.  --Yamla 00:47, 29 July 2006 (UTC)


 * You aren't the first to take issue with this, but allowing the usage of IEC prefixes to prevent ambiguity is and has been a Wikipedia policy for some time. This page is not the appropriate place to make any arguments you may have.  Please take a look at WP:MOSNUM. -- uberpenguin

Names
What is the source/significance of the names, Yonah, Merom, Sossaman? —wwoods 20:52, 4 March 2006 (UTC)
 * Intel's codenames are traditionally named after random geographic areas, especially those close to where the chip was designed. The Pentium and Core Solo/Duo were designed in Israel, so the names tend to be of Israeli origin (interestingly, Yonah seems to be one of the only exceptions to the "random places" rule: it's named after a Biblical figure, not a place). Many of Intel's mainline processors use Oregon and northern California as a source. I've heard it mentioned in some comments on a Slashdot article, from a former Intel engineer, that it's common practice to make a list of random names from Mapquest and send them to someone from Intel legal for approval (I can't find the article right now--if I find it later, I'll link to it on this page). This is because Intel wants to avoid trademark conflicts--it's much harder to trademark a geographical region. Jgp 21:03, 4 March 2006 (UTC)

I have worked for Intel, in a position that dealt with working names and final product names. I have my doubts that "Yonah" was named for a Biblical figure because Intel's naming policy is that geographical names will be used for working names of products. IOW, the approvers of names have to be able to find the proposed name on a map. That's not to say that an Israeli team didn't find a geographical feature with a name they liked, but Mount Yonah in Georgia or maybe a geographic feature in Israel named Yonah (the town of Kefar Yonah?) was submitted to the names approval department.

Yonah = Jonah - man thrown overboard by his shipmantes and eaten by large fish. Not an auspicious name at all. ;-)

New WP:MOSNUM "policy" re lengths
I went to have a look at WP:MOSNUM. It appears that WP:MOSNUM has recently received a significant large edit from a new user without the normal discussion taking place first. The edit has been reverted by more experienced hands. Discussion is in progress. We might be best off allowing new guidelines to settle for a while before rushing out making changes to pages such as this one. Duckbill 17:47, 10 March 2006 (UTC)

Application & device drive in Windows x64
I have improved the wording. Anyone who has ever used Windows x64 should know the problems are not as major as some people appear to have suggest. General application support is not really a problem. There are few applications compiled for x64 but most 32 bit apps applications work without problem on Windows x64. Applications which require the use of device drivers or more direct access to the operating system like anti-virus software and optimisers can be a problem but many of these now have x64 versions. Games are also rarely a problem. I've only had 2 kinds of problems vis-a-vis application compatibility. Ones related to stupid game installers which fail because they try to force an install of DirectX and when this fails they simply quit. The other ones are with programs that create boot disks since many of these use Dos programs. Nil Einne 12:55, 1 April 2006 (UTC)

Device driver support is a somewhat different issue since they need to be written for x64. Given we are talking primarily about modern systems, primarily laptops, we can assume motherboard, sound cards, network card and video card drivers are not a problem. Generic devices like USB key also are not a problem. Digital cameras, printers and scanners can be a problem. Also some specialist devices like mobile phone data cables and I believe some fancy mice and joysticks don't have specialist drivers which provide more customisation or accuracy. Nil Einne 12:55, 1 April 2006 (UTC)

the lack of on-die memory controller
I think that "the lack of on-die memory controller" should be moved from shortcoming to advantages. It makes the processor more compatible with memory modules.
 * No, it makes it easier to change memory standards. However, the performance penalty is often seen as more important.the1physicist 04:55, 6 May 2006 (UTC)

Disadvantages?
Can someone please explain why lack of support for EM64T/AMD64/x86-64 instructions is a "disadvantage?" I was unaware that slight increases in big integer performance and extra native addressing space were such marked advantages... I propose this line be removed from disadvantages. -- uberpenguin


 * There are obviously applications for which the extra native addressing space is a marked advantage. I suspect most of those applications aren't used on the sort of machines that would have Yonahs in them, however, with the possible exception of Photoshop, and I don't know whether there's a 64-bit version of Photoshop.


 * The article does note that this isn't an important disadvantage (and Core 2 will be 64-bit, so the "future-proofness" is irrelevant, as Yonah is the last of the Pentium M architecture line, and the future is Core 2 and the Intel Core Microarchitecture), and that it's more of an issue with the server derivative, Sossaman. Guy Harris 22:11, 10 June 2006 (UTC)


 * My statement was marred by a touch of sarcasm. It simply struck me as silly to broadly label the support or non-support of 64-bit instructions in a contemporary general-purpose desktop microprocessor an "advantage" or "disadvantage."  This especially being the case when the number of common PC applications that significantly benefit from higher big int performance can be counted on one hand.  I'm just asking if there are any reasonable objections to either removing or rewording this particular statement? -- uberpenguin


 * I agree; the 64-bit architecture is more hype than anything else. After all, how many users really go over the 32-bit limit of 4GiB of addressable memory?  Few people have a use for PAE mode.  What really goes in to using over 4GiB of memory?  Database servers.  How many people are running database servers?  Few.  I don't see it as an "advantage" or "disadvantage" for the common end user. —Preceding unsigned comment added by Jdstroy (talk • contribs)


 * I agree that the average user is not going to touch a 64-bit operating system in years and thus lack of its support is probably insignificant. However I'd like to point out that the AMD64 architecture also has other improvements in front of x86, beyond >4GiB addressing. You can call it hype, but someone has to push the technology forward. -- intgr 01:37, 12 August 2006 (UTC)


 * I've heared rumours that the version of unrealed for UT2K7 will require a 64 bit OS, given unrealeds already high memory requirements (for large maps it grinds swap like hell on a machine with a gig of ram) it wouldn't surprise me if this was indeed the case. Also the old guideline was 2.5 times as much swap as ram so if you have a box with a gig of ram and wan't to have that much ram/swap and have it all availible to one process then you will need to go 64 bit (or possiblly 4G/4G i386 linux but i belive thats rare). Plugwash 23:37, 1 December 2006 (UTC)


 * Not inasmuch as IA-32 is concerned... As was alluded to before, PAE has been around for awhile, and that will get you up to 64 GiB virtual memory. -- mattb


 * IIRC both windows and linux use a flat memory model for apps and for simplicity divide a 32 bit space into an OS region and an app reason (i belive there are patches that change this for linux but i don't think they are in mainline). IIRC windows is limited to 2 gigs for any one app and linux is limited to 3 gigs for any one app. So for those who use all thier memory for one app the limitations of IA-32 as practically used are already looming. Plugwash 15:14, 2 December 2006 (UTC)


 * It's actually 4 GiB per process, and this limitation can be dodged with various means if the process really needs to map more than that. -- mattb

Large Deletions Without Explanation?
Were the sections removed by 58.69.61.111 appropriate? Should the edit be reverted? Frankie


 * Yes and yes. Jgp has already reverted the vandalism. Guy Harris 18:17, 30 June 2006 (UTC)

"Dual-core-ness"
Hi,

I was thinking about this one for a while...

When Intel markets these "dual core" chips, such as Core Duo and Pentium D, are they really "dual core," much like two physical processors, and able to work concurrently, or are they similar to the the "HyperThreading" technology, where two logical processor cores are allocated for a single core, and then a secondary hardware thread is executed while the primary thread is stalled (in cache miss, branch misprediction, or data dependency)?

Thoughts?

Jdstroy 06:08, 11 August 2006 (UTC)


 * They are really "dual core" - either two dies in a single package or a die with two complete processors on it. Guy Harris 17:01, 11 August 2006 (UTC)

"Pentium Dual Core"
The "Pentium Dual Core" T2060 from the Core processor list is not mentioned anywhere in here. It is just a budget version of the Core Duo, right? Shouldn't it be part of this article, despite its name? And has Intel given any official reason for resurrecting the Pentium brand name for its budget version of the Core Duo? 205.157.110.11 04:07, 14 February 2007 (UTC)

Reply: The Pentium Dual Core is a chip shrouded in mystery as Intel is not releasing much information. Some claim it is a dual core version of the Pentium M, some say it is a cheap Core Duo with less cache. Until Intel comes out with more information,I firmly beleive we should leave Pentium Dual Core as a seperate article with perhaps a reference in the Core 2 Duo article. My reasons are as follows: - Pentium Dual core architecture may be based on any type of chip from Pentium M to Pentium D or even Core Duo. Until firm evidence of what family of chip it is related to is produced, we should keep it separate. - The chip has many technical difference to the Core Duo line of chips and is aimed at a different market. - Intel uses a different brand name for the Pentium Dual Core. If Intel does not want to call it Core Duo, there must be a sizable difference between the Pentium Dual Core and the Core Duo, worthy of creating a whole new brand. Hotdog111 10:25, 22 February 2007 (UTC)

"Pentium Dual Core" is the non EM64T version of the Pentium D. The only one available, T2060, is NOT A CORE PROCESSOR and still uses the Netburst Architecture that Pentium 4 processors use. It's the same construction as a Pentium D: two cedar mill dies flipped in on each other on a LGA 775 socket. Source: http://www.intel.com/products/processor_number/proc_info_table.pdf 71.103.162.89 20:37, 15 March 2007 (UTC)


 * The source doesn't give the socket information; is there another source for that? (If so, that belongs on the Intel Pentium Dual Core page.) Guy Harris 21:16, 15 March 2007 (UTC)


 * Pentium Dual Core is based on Core Duo, not Core 2 Duo neither Netburst, the CPUID is 06ECh (family 6h, P6 and derivates with model 0Eh, Yonah)http://download.intel.com/design/mobile/SPECUPDT/31651501.pdf, about the merge, no, there are some Celeron M based on Core Duo but they aren't in this article too. EduardoS 17:19, 31 March 2007 (UTC)


 * You might want to update the Intel Pentium Dual Core page with that information. Guy Harris 19:21, 31 March 2007 (UTC)

Reason for creation: Intel's reason for creating the Pentium Dual Core branding was to get a cheap processor into developing country markets. The Pentium name was resurrected for this task because of the familiarity that the Pentium name holds. It was intended to be kept away from the US but PC manufacturers started wanting it for a low end option for their customers, in most cases using it to replace the Celeron M. Source: Conversations with several Intel employees regarding the creation of it.

I do not believe it should be merged because of the differences between the chips. I do however believe that there should be a link to it through the Intel Core page.


 * I agree. There is text currently duplicated between the two articles.  This redundancy is a maintenance problem (for example, I had to make the same change both places just now).  I think that the majority of the text could reasonably be either place, but not both.  DHR 17:49, 27 April 2007 (UTC)

It sure would be nice to know more precisely what features these chips have. VT (I would guess not)? EM64T (highly unlikely)? According to the sparse datasheet from Intel it does have: Execute Disable Bit, Enhanced Intel SpeedStep, SSE2, SSE3. DHR 17:49, 27 April 2007 (UTC)

http://translate.google.com/translate?hl=en&sl=zh-CN&u=http://cn.tech.yahoo.com/070109/472/2oh88.html&sa=X&oi=translate&resnum=1&ct=result&prev=/search%3Fq%3DT2060%2Bcpu-z%26hl%3Den%26sa%3DG That link is an English translation of a Chinese site that has English screenshots of the program CPU-Z being run on a Pentium Dual-Core T2060. Perhaps this will help show some of it's features.

Disambiguation is confusing
The disambiguation at the top of the page is confusing. "Mobile processor" seems to imply something to be used in a cell phone, palmtop computer, or similar device. Basically I read that it was directing me to Intel Core (CPU architecture), itself a redirect, which then proceeded to tell me (indirectly of course) that I was reading the wrong article (I should have been reading this article, apparently). (I was looking for the article on the processor used by Apple.) —16:16, 7 December 2007 (UTC)

Splitting out yonah, making this page an overview for core solo/duo/2/i3/i5/i7/i9
Intel now refers to all Core 2, Core i5 and Core i7 CPUs as 'Intel Core' processors, and the pages for Xeon and Celeron (not Pentium though) are structured to give an overview over all product lines that are have mostly identical (Celeron M, Celeron D, Celeron Dual-Core) names but are based on different technologies.

I propose moving this (minus the introduction) page to Yonah (microprocessor) to preserve its history, and to start a new overview page for the brands, structured something like


 * Pentium-M microarchitecture based
 * Core Solo
 * Core Duo
 * Core microarchitecture based
 * Core 2 Solo
 * Core 2 Duo
 * Core 2 Quad
 * Core 2 Extreme
 * Nehalem microarchitecture based
 * Core i3
 * Core i5
 * Core i7
 * Core i9

Much of the information in such an article would of course be duplicated from the current Core 2/Core i5/Core i7 articles, so those could eventually be merged into this page. Arndbergmann (talk) 08:10, 25 September 2009 (UTC)

Dubious
Article currently claims that the Core Duo was the world's first sub-25W dual core processor, and was launched in Jan 06. These claims seem doubtful to me, as I cannot find any source for this statement, and I can easily find earlier examples of multicore processors that appear to consume less than 25W (often significantly so), e.g. Asynchronous array of simple processors which appears to have used <1W for a 36-core chip. Also, while I can find no contemporary description including the power dissipation figure, BlueGene/L appears to have been built in 2004 based on dual-core nodes dissipating 17W each, AFAICT.

The author may have meant that it was the first x86-compatible dual core with this power consumption, but that is not what it says, and even this I can find no source for confirmation. 212.159.69.4 (talk) 13:54, 26 November 2011 (UTC)


 * This paper describes a BG/L installation with 16,384 dual-core nodes as dissipating 400KW. This means the max per node is 400000W/16384 = about 24.4W.  This also includes memory, ethernet switches, and I believe cooling, so the node power is clearly less than 25W.  At this point, I'd say this is enough to remove the claim from the article, so I'm going to do it now. 212.159.69.4 (talk) 15:01, 26 November 2011 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 1 one external link on Yonah (microprocessor). Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive http://web.archive.org/web/20060206203721/http://www.intel.com:80/products/processor/coresolo/ to http://www.intel.com/products/processor/coresolo/

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.—cyberbot II  Talk to my owner :Online 18:27, 29 February 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 1 one external link on Yonah (microprocessor). Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20070809234956/http://balusc.xs4all.nl/srv/har-cpu-int-c1.php to http://balusc.xs4all.nl/srv/har-cpu-int-c1.php

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.— InternetArchiveBot  (Report bug) 19:39, 20 July 2016 (UTC)