Talk:Athlon 64

It's a alot better that the x2 has it's own page since the tech is different and people who are looking for information actually get a good digested form of what the x2 is all about. JPN --68.22.204.205 23:29, 9 August 2005 (UTC)

What is with the weird redirect for Athlon 64 FX?

Click on Athlon 64 FX. I don't think it should be redirected to the basic article about the AMD Athlon. If you guys are re-directing that term, I think it should point directly to this article. Just my opinion. --Anonymous Cow 20:35, 15 Feb 2005 (UTC)

I've checked, it appears that it was a article before for about a year and it was redirected recently (Feb 14). I reverted the redirect. The Athlon 64 FX should have a seperate article because of the differences between the two proccessor types despite the similar names. Norman Rogers\^talk 23:03, 15 Feb 2005 (UTC)

Some changes

I have made some changes.

For an article to have a subheading "Athlon 64" even though the article has a title "Athlon 64" seems a bit redundant.

I removed references to the Athlon 64 architecture (Athlon 64 is a product that implements the AMD64 architecture. There is no Athlon 64 architecture), and information abou the AMD64 architecture that is not specific to the Athlon 64 processor. This fits the pattern of the rest of wikipedia, where articles about processor products don't go into details about the processor's architecture, but instead just point out that they implement a certain architecture, and let the AMD64 article handle that information.

Samrolken 01:42, 18 Sep 2004 (UTC)

Differences and distinctions, please

Can someone differentiate between "ClawHammer", "Newcastle" and "Winchester"? Thanks.

DanielVonEhren 23:22, 20 Feb 2005 (UTC)

The article should say something about the different revisions of the Athlon 64. Clawhammer and Newcastle are basically the same processor, but with only 512 kB Cache in the latter case. Both can either be revision C0 or CG. CG has an improved memory controller and is more energy efficient. Then came revision D0 aka "Winchester" (90 nm with an even more improved memory controller and less power consumption). --Echoray 17:03, 21 Feb 2005 (UTC)

Any chance the person who "should say something" could be you?  :-)

DanielVonEhren 19:35, 21 Feb 2005 (UTC)

Merging Athlon 64 FX and Athlon 64 pages?

It seems that Athlon 64 page contains info also about FX model, and in the future maybe X2 dual cores also. Currently there is at least short article about FX cores, so should those be merged? --Thv 10:18, 2005 May 4 (UTC)

Both the FX and the X2 should have their own article - they are different than the standard Ahlon 64. --Denniss 12:26, 2005 May 4 (UTC)

A question: how is it that the Athlon 64 FX gets its own page when ALL K7-based Athlons (from Pluto to Barton) share the same page? The modern Socket 939 Athlon 64 FX is pretty much just a San Diego with unlocked multipliers, but there are many differences between the Athlon Classic and Athlon XP. SVI 22:47, 26 July 2005 (UTC)

I've now added merge request from Athlon 64 FX and Athlon 64 X2. Currently this page contains some info about X2 and FX, and also List of AMD Athlon 64 microprocessors contains also FX and X2 models. Thv 07:18, July 27, 2005 (UTC)

PS. List of AMD microprocessors separates X2 from other Athlon 64 models as K9 series, thus at least FX should be merged, but maybe X2 can have it's own article? Thv 07:22, July 27, 2005 (UTC)

X2 should definately not be merged, FX could be however, as it is exactly the same as the normal A64, just the higher end model. Bluemoose 08:35, 27 July 2005 (UTC)

How do we merge the rival Intel processor(s)? We separate the desktop from the server processor, but do not separate the performance advanced version from the normal consumer version (nor the dual core). I'm an AMD guy atw, but whatever we do to Intel we need to do it to AMD... unless there is a big difference or special situation. Is the X2 a special situation? Dual core is new but nothing special. FX is a high end consumer edition, but eh... Overall I'd like Athlon 64 and Pentium 4 go one way or the other, and that means I want Athlon 64 combined and merged. --x1987x 17:12, July 29, 2005 (UTC)

X2 is just two A64 cores on one die. They are different processors, but I am unsure as to whether they are different enough to justify separation. Besides, article separation is not done on a technological basis; Pentium II and Pentium III are separate articles, even though early (Katmai) P3s are the exact same as late (Deschutes) P2s with just SSE and the infamous PSN added. More significantly, the Celeron article includes Celerons from Covington to Dothan to Prescott; there are HUGE technological differences between them all. Processor articles are separate on a name basis, not a technological basis-- and even if they were on a tech basis, surely two of the same core is a smaller difference than a complete arch shift (as with, say, Cu128 Celerons and Willamette Celerons). For these reasons, X2 should definitely be merged.

Also, List of AMD microprocessors is wrong, and has since been corrected. AMD themselves classify the X2 processors as K8, as does sandpile. K9 was originally meant to be dual-core, but AMD revised their plans. K9 is cancelled, or at the very least not in official existence yet.

The 64 FX is easy. It's almost identical to a regular 64 AND is also an "Athlon 64". Definitely merge. SVI 01:12, 30 July 2005 (UTC)

I vote for merging both articles into this one. The FX should be merged without a doubt. As for the X2, AMD renamed their K9s to dual core K8s a LONG time ago. Each core in an X2 is really just a regular K8, so a merge is definitely in order. Somebody just do it already.the1physicist 02:32, 30 July 2005 (UTC)

An athlon x2 is not "just" 2 cores on one die (well, ok it nearly is) intels dual core is just 2 cores on one die, However as long as no info is lost and it is still clear I won't cry if they are merged. Bluemoose 22:31, 31 July 2005 (UTC)

And a Venice isn't "just" a Winchester with SSE3, but the other differences are so minor that they barely deserve mention. Same applies here. Besides, as I said, it's done by nomenclature, not technological advances, so even a huge difference among processors with the same name would not merit article splittage. SVI 23:05, 31 July 2005 (UTC)

Well, if no one else is going to do it, I'll merge, scared newbie or no. All relevant info on the A64FX except the SH core (130nm Opterons, used for the early 940 FXes) was already in this article, so I just copied over SledgeHammer and redirected Athlon 64 FX to this article. Are there any objections to merging Athlon 64 X2 with this article? SVI 18:06, 8 August 2005 (UTC)

Please update the Athlon 64 article with all Athlon64 FX info found in the old article !! And do update the cores with the FX model numbers as well. Do not merge the Athlon64 X2 - if ou are doing this then please don't forget to merge the Pentium D and EE into the Pentium 4 as well !! The X2 is more then just using two dies on one core as Intel tried on these two CPU. --Denniss 22:24, August 8, 2005 (UTC)

Already did, the 64FX is practically identical. And as I've said before, articles are not separated on a technical basis, they are separated on NOMENCLATURE. The "Pentium D" is not the Pentium 4D (that's Gallatin, if sandpile is to be trusted), it is a "Pentium D"-- its different name merits a different article. (The same goes for the "Pentium Extreme Edition", although not the "Pentium 4 Extreme Edition".) The Athlon 64 X2, on the other hand, is clearly an Athlon 64 variant. The Celeron article includes processors from Covington to Northwood, surely a much bigger difference than just two cores (and yes, that's pretty much what it is-- I realize that AMD has a special interconnect, but that doesn't make the tech gap any larger). The Pentium III article includes IIIE, IIIEB, and late Tualatin variants, because they're ALL PENTIUM IIIs. Are there any objections that I haven't addressed? SVI 06:27, 9 August 2005 (UTC)

I again vote to merge the X2 article. There is no reason to keep them separate. Whoever did the FX merge, good job.the1physicist 01:25, 10 August 2005 (UTC)

In my opinion no matter the simmilarity between Athlon 64 and the FX and X2 series, they all deserve a separate page, it would be much better. I totaly disagree with the1physicist it so wrong to be merged like this. Apocalypse_Bg

Merge of Athlon 64 X2 to Athlon 64

There have been many objections against this merge, and I understand that X2 can have enough content to deserve own article. However, my main point is that currently intro chapter of Athlon 64 says that X2 is one of three variants of Athlon 64. So if the intro chapter contains clear difference between X2 and other processors called "Athlon 64 anything", then I won't object removing merge request. --Thv 07:49, August 10, 2005 (UTC)

Confusing memory terminology

"940: Opteron and old Athlon 64 FX, 128-bit memory interface - Requires registered DDR memory, but can use ECC"

This wording is misleading, and possibly inaccurate: Registered DDR and ECC are far from mutually exclusive--they are usually paired—Trevor Caira 12:52, 9 August 2005 (UTC)

You're right, that's very misleading. Removed the ECC bit, as it was kind of pointless anyway. SVI 15:33, 9 August 2005 (UTC)

It should be merged

The Athlon 64 X2 bears very little difference from its single core brethren, other than a second core being linked through the integrated crossbar. At its heart, an Athlon 64 X2 is in reality two Venice or San Diego cores fabricated next to each other and linked by that crossbar. As a result, the data of this page should be merged with the main Athlon 64 page, which currently lacks a lot of information about the X2 processor.

I think they should not be merged as 1) Athlon 64 will become too long and complicated; 2) if a person searched for "athlon x2" they would be disappointed to be redirected to the athlon64 article; 3) the athlon x2 article is clearly long enough to sustain itself, it's hardly a stub, and it duplicates very little material. Martin - The non-blue non-moose 18:51, 16 August 2005 (UTC)

Athlon 64 is not nearly as long or complicated as other articles; I doubt anyone would be disappointed to find a redirect to the relevant article; this has nothing to do with duplication. Again, I point to the Celery article as an example: it's long, there are no separate articles for the (significant) core differences, and even if you split all the cores into separate articles there'd be no duplication.

The issue here is that the Athlon 64 article should list all the Athlon 64s. The Celeron article lists all the Celerons regardless of differences and length, Pentium 4 has everything from the original suffix-less Willamettes to the A/B/C Northies and A/E Prescotts, Pentium MMX redirects to Pentium. It's, again, a question of nomenclature. SVI 10:27, 18 August 2005 (UTC)

I agree, it should be merged. The article has room for expansion, and it's really a waste of space to have an article for every variation of a processorAmren ^(talk) 04:49, 23 August 2005 (UTC)

The FX may be a variation but the X2 is a little bit different from every single core Athlon 64. --Denniss 12:20, August 23, 2005 (UTC)

Should there be a page for the K8 microarchitecture?

Are the micro-architectures of Opteron, Athlon 64, Turion 64, etc. sufficiently similar that there should be an "AMD K8" page for the K8 micro-architecture, with the pages for particular K8-micro-architecture chips referring to it (as is already done for Intel P6, NetBurst, and Intel Next Generation Microarchitecture)? (If so, an AMD K7 page might also make sense.) Guy Harris 01:24, 9 January 2006 (UTC)

I've wondered about that (more in the context of the lack of a K7 article, but K8 too). I do think such a thing would make sense; processor articles seem to be devoted mainly to describing the history and forms of said processor rather than in-depth technical details, and most people looking for technical details would probably look up the architecture rather than the brand name of one of the processors based on said architecture. Plus, as you said, there's quite a precedent with P6 et al.

In short: I think it makes perfect sense, and I'd support such a thing, but I am too lazy and uneducated to make something like that myself. SVI 05:35, 9 January 2006 (UTC)

Yes, there should be a K8 page, and it would be that of the Opteron, Athlong 64, X2, ... 132.205.45.110 21:56, 23 January 2006 (UTC)

• AMD K8

And now we see why a "demerge" template was suggested on Template_talk:Split; the "split" template speaks of a disambiguation page, but none would be needed here. As there isn't a "demerge" template, "split" will have to do. Guy Harris 22:48, 23 January 2006 (UTC)

Athlon 64 FX-60 released!

As it was, the article referred to the Athlon 64 FX-60 as an unreleased CPU. I changed this to reflect its recent release date (today!) and placed emphasis on the dual-core nature of this CPU as opposed to the other models in the FX line. I also defined the analogy between it and the Athlon 64 FX-53, both of which saw/will see two versions for two different sockets (940 then 939 for FX-53 and 939 then AM2 for FX-60). Oh and it's fine and dandy that the Athlon 64 FX and Athlon 64 articles are merged now, but I think the FX series should at LEAST get its own section! Another user pointed out that modern FXes are just Socket 939 San Diego chips with a high clock rate... but let's not forget the architectural basis of the FX line! The Athlon 64 FX-51 was very similar to the Opteron when they both were released originally (even note they were both on Socket 940). Because this gives it a clear distinction between it and the Athlon 64 line, and now the fact that the FX line has both single and dual core items, I think it is worthy of its own section in this article. What do you think? pinky 06:38, 11 January 2006 (UTC)

The following discussion is an archived debate of the proposal. Please do not modify it. Subsequent comments should be made in a new section on the talk page. No further edits should be made to this section.

The result of the debate was don't move. —Nightstallion (?) 10:19, 24 January 2006 (UTC)

Requested move

Athlon 64 → AMD Athlon 64 – More specific. Manual move failed.

Voting

Add *Support or *Oppose followed by an optional one-sentence explanation, then sign your vote with ~~~~

• Oppose - I don't see a need for more specificity (nobody else makes Athlons, as far as I know, and the Intel processor pages don't have "Intel"), and the other pages only acquired the extra "AMD" recently. Guy Harris 12:01, 19 January 2006 (UTC)
• Oppose - It is more intuitive this way, and other articles follow similar naming guidelines. For example, Pentium 4 and PlayStation. - Hahnchen 23:50, 21 January 2006 (UTC) (Moved from WP:RM --Lox (t,c) 16:06, 22 January 2006 (UTC))

Discussion

Add any additional comments

I'm not sure why "more specific" is necessary - I don't know of any Athlons from anybody other than AMD.

On the other hand, most if not all of the other AMD chip pages, including the Athlon page, have "AMD" in the name, so perhaps the Athlon 64 page should be renamed for consistency. (The Intel processor pages don't have "Intel", but to be consistent with them you'd have to take the "AMD" off of all the other pages, which is probably A Bit Much.) Guy Harris 11:53, 19 January 2006 (UTC)

The other pages only do because he just moved them, they need to be moved back. Martin 11:56, 19 January 2006 (UTC)

OK, then I see no reason for the move of this page - and see no reason for his move of the other pages, either, and have changed my vote to Oppose. Thanks for the info. Guy Harris 12:02, 19 January 2006 (UTC)

The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made in a new section on this talk page. No further edits should be made to this section.

Further comments on moving

• Oppose - I don't think that this is necessary. Athlon 64 is not less or more specific than AMD Athlon 64. I will agree that if such a change is made, then more products should be changed too (e.g. Pentium 4 as Hahnchen stated). --kostas213 17:07, 26 January 2006 (UTC)
• Oppose - "article naming should give priority to what the majority of English speakers would most easily recognize, with a reasonable minimum of ambiguity, while at the same time making linking to those articles easy and second nature." (from WP:NAME). The usual reference to these processors would simply be "Athlon 64", not "AMD Athlon 64", there is no ambiguity in that name, and certainly it makes more easier and more likely linking. --Stormie 13:23, 30 January 2006 (UTC)

K8's sucessor, K8L

I just created a page for the K8L, would it be good to mention it? Pueywei 14:03, 31 March 2006 (UTC)

Manchester core / single core

There should be a statement under the vencie cores, that some are infact manchesters with a cut HTT bridge between the cores. These prossessors did not make the grade as a X2 but make great single cores. —The preceding unsigned comment was added by Jthull (talk • contribs) .

Thank you for your suggestion regarding Athlon 64! When you feel an article needs improvement, please feel free to make whatever changes you feel are needed. Wikipedia is a wiki, so anyone can edit any article by simply following the Edit this page link at the top. You don't even need to log in! (Although there are some reasons why you might like to…) The Wikipedia community encourages you to be bold. Don't worry too much about making honest mistakes—they're likely to be found and corrected quickly. If you're not sure how editing works, check out how to edit a page, or use the sandbox to try out your editing skills. New contributors are always welcome. —BorgHunter ^ubx (talk) 23:57, 19 May 2006 (UTC)

About the DRAM speed formula

The mathematical expression for calculating the DRAM speed, displayed on the main page is wrong!

The inaccuracy is in the celling function where the "DRAM divider" and "CPU multiplier" parameters must be multiplied NOT divided.

A typical situation to illustrate:

• CPU speed: 2500MHz;

• CPU multiplier: x9;

• DRAM divider: for DDR333 mode == 6/5 ratio;

1. Wrong formula -

 (CPU_speed)/(Multiplier/Divider) = 2500/(9/(6/5)) = 2500/(9/1.2) =
  = 2500/8 = 312MHz for the DRAM interface;

2. Correct formula -

 (CPU_speed)/(Multiplier*Divider) = 2500/(9*6/5) = 2500/(9*1.2) =
  = 2500/11 = 227MHz for the DRAM interface;

First, there's no A64 with 2500 MHz, either 2400 or 2600 MHz. Multiplier at 2400 MHz is 12 (12x200). I'm not that expert in A64 DRAM interface but I know PC-2700 (using DDR-333 chips, 166 MHz) will run slightly underclocked with standard 200 MHz system clock. Memory divider is CPU frequency divided by memory frequency and rounded up to the next multiplier. Example with PC-2700: 2400/166 = 14.457 = 15. 2400/15 = 160 MHz memory clock. --Denniss 18:53, 4 July 2006 (UTC)

There is 2500MHz or any other frequency in the feasible clock range for the given CPU architecture. Haven't you heard about overclocking? The example above was just a random picked number for the core speed. Whatever, your case is indicative that the DRAM divider can't take halfs but only integer values, that's why memory clock got a little bias because of the divider value rounded up from actual 14.4 to 15. Well it could be in the opposite direction, rouded down to 14, but this will result in non-standart and slightly overclocked memory (~171MHz in this case)... well not that 160MHz is stantard, but it's safer. Fellix 11:39, 10 July 2006 (UTC)

The formula is fine, you're just interpreting DDR333 as 6/5, when its actually 5/6. 6/5 is 1.2, and using that would be meant to INCREASE the RAM speed, not decrease it; yet DDR333 is meant to produce lower speeds than the normal DDR400. With that in mind, divide works.

DRAM clock = CPU clock / ( ceil(CPU multiplier / DRAM divider) ) = 2500 / ( ceil(9 / (5/6) ) ) =
= 2500 / ceil(10.8) = 2500/11 = 227.

Using a multiplier worked in that case, but switching it back around and using divide produces the same result. To fully prove it to you, take this example:

Multiplier =10.5, LDT = 266, Divider = DDR333 (5/6). 2793 / ceil( 10.5 / (5/6) ) = 215.

2793 / ceil( 10.5 / (5/6) ) = 215.

With half multipliers, your version of the formula doesnt stand up. If you have an A64 PC handy, give it a try.

On a final note, please mark your comments with ~~~~. That will put your name etc after your message. AthlonBoy 20:43, 6 July 2006 (UTC)

Well you'r right for the DDR ratio flipping, but that's simple mathematics - when you divide by fraction number that or either way you have to change the positions of the numerator and denominator and thus "my" formulation comes in to place, so both cases are correct by my mean. Now I wonder how is the case with the new AM2 DDR2 platform? And yes, I have a K8 setup and thus I've verified much enough cases in practice ;). Fellix 11:57, 10 July 2006 (UTC)

Yeah, AM2 calculates its DRAM speed in the same way. Thats why the speed ends up being about 780MHz instead of 800MHz on some chips. You're right about both formulas being valid depending on how you take the DRAM divider, too. The trouble is that DDR333 is 5/6, due to the original meaning for that divider.

I cant remember exactly what motherboard or chipset had the divider feature first, but it was in the Pentium 4 Northwood era if I remember correctly. You had various options offered as 6:5, 5:6, 4:3, etc etc. The board would take the ratio and just multiply the FSB by it, to get the DRAM frequency, without any of this formula nonsense (you probably know all this, but bear with me). Some manufacturers labeled them as DDR266, DDR333, DDR400, DDR466, and the rest, depending on what the resultant speed was. Multiplying 200MHz by 5/6 resulted in a speed of 166.67MHz, or DDR333, and from that day on we've been using 5/6 as DDR333, even now the FSB has taken a long walk of a short pier. In keeping with that, you need to use the formula already on the page. --AthlonBoy 15:22, 10 July 2006 (UTC)

Main 939 benefit

The newly filled-in History section (by the way, nice work BorgHunter) states that the main benefit of socket 939 was its 1GHz hypertransport. Surely the dual channel memory controller did more? It has been shown that even a Hypertransport speed of 600MHz is fine, by lowering the HT multiplier, so a small 200MHz bump hardly made the difference. The near-doubling of memory bandwith did, however. I edited the section but it has been reverted because the citation is incorrect, but the citation was a Toms Hardware review of the first 939 parts, and it clearly states that 754 has a single channel memory controller.

Just thought it would be best to put something in the talk page rather than start a revert war.AthlonBoy 21:05, 6 July 2006 (UTC)

Good catch, that last edit was made in a rush so I wouldn't be late to work. :-P I've added the dual channel mention and included a citation. —BorgHunter (talk) 04:19, 7 July 2006 (UTC)

List of models

Should that list be on the main article, or should we make a subarticle List of Athlon 64 models? I don't much like having it there, but I want some input before I go about moving it. —BorgHunter (talk) 05:00, 8 July 2006 (UTC)

It might get some objections when (and not if, but when) you nominate this for FA. Also, 47 references is insufficient. There should always be at least 200 references in an article, including a reference to WP:EL for the heading "external links". :-) — Deckiller 05:03, 8 July 2006 (UTC)

There's already a List of AMD Athlon 64 microprocessors page, so there's no need for a "List of Athlon 64 models" page. One could argue that the list on the Athlon 64 can shrink (or disappear), with the details going on the "List of AMD Athlon 64 microprocessors" page. (The same probably applies to other x86 processor pages, whether Intel or AMD; there are already "List of" pages for several of them.) Guy Harris 07:44, 8 July 2006 (UTC)

"Hardware-set permission levels"

I've been all over the Net trying to get some more information on this "Hardware-set permission levels" business, and unless it's part of Presidio or that virtualisation thing (Pacifica?) can't find anything! Where did this bit of info come from and which 64s is this feature in?

(Unless that was just a reference to the NX-bit)

It sounds like you're reading about Data execution prevention. DigitalEnthusiast 21:09, 20 December 2006 (UTC)

Mobile Athlon 64

What's this shit about "Mobile Athlon 64" leading to Turion 64? They're two separate processors. The Turion 64 is the low energy consumption one, and the Athlon 64 Mobile was, for a time, the high-performance one that was essentially the desktop processor with SpeedStep added. I know this because I've got a shitty Athlon 64 Mobile in the computer I'm typing this with.