Talk:AMD K9

Why are K8 updates listed on this page? Any reason they are here and not on the AMD_K8 page? --Yamla 18:38, 2005 Feb 25 (UTC)\

Well, to be honest, I (the writer of the article) don't know to much about it.

But since there was no article about the K9, I tried to do some research, and write an article about it, hoping someone else would improve it.

The reason it's not on the AMD_K8 page is, that's a good article, I don't want to change it too much. But in general terms, you're right. If you don't like the article, I invite you (and anyone!) to improve it. --sludink, Sun Mar 13 19:27:15 CET 2005--

Folks, AMD announced quite a while ago that they are no longer referring to CPU generations by K numbers. Furthermore, they said that there is no K9 (dual core K8's are just that) and that their next major core would have been K10.the1physicist 30 June 2005 03:38 (UTC)

What the1physicist said. Should this article be deleted, or heavily revised and just basically say "K9 was cancelled, shoo!"? SVI 01:23, 30 July 2005 (UTC)

I'm going to heavily edit this page. I'm basically aiming for the "K9 was renamed, shoo!" approach.the1physicist 02:00, 30 July 2005 (UTC)

Ki9 sounds too like canine ? And K10 is the kitten. Much better for top-level processor ?

I think the page should stay, because while K9 was an internal project, AMD's single design team spent much time on it. While admittedly not released as a product, the fact the K8 core was NOT developed by the primary design team at AMD due to K9 commitments, has now started to hurt AMD, as Intel catches up with the Conroe platform. The K9 is one of those interesting 'what ifs.' Had AMD developed K8L instead of K9, just how bad would things be for Intel in 2006? K9 probably cost AMD the best chance they will ever get to draw level with Intel. Its interesting, and I would argue significant, for that reason alone. Timharwoodx 12:25, 11 June 2006 (UTC)

What are you people talking about??? There is not a single main design team at AMD. They have the original x86 design team in Austin, TX (they did the K5, K7, K9, and Greyhound (aka Family 10h)). They have another design team in Sunnyvale, CA (corporate HQ) that was the Nexgen design team - these folks did the K6 and K8 (and working on the next major microarchiture that comes after Family 10h. They also have another team in Boston, MA (mostly ex-API and ex-Sun folks).

What was K9 was canceled do to a change in direction of what would be needed to compete with Intel post 4GHz NetBurst failure. The reusable parts of the design were put into Family 10h, but the same folks working on the K9. One could argue the Austin, TX folks are the failure... K5 was a disaster and lead to AMD buying Netgen to save their ass with the K6 - and these same folks saved AMD with the K8. —Preceding unsigned comment added by Dr unix (talk • contribs) 01:07, 11 March 2008 (UTC)

K9 was a long pipeline dual-quad issue design intended to scale. Now some of you guys are pretty confused as to what went on at AMD. K5 was the last processor that AMD designed 'originally' themselves. K6 was the NexGen NX6x86 with only minor reworkings (backside L2 cache removed, Socket 7 compatibility added). K7 was developed by the DEC Alpha design team headed by Dirk Meyer after AMD hired pretty much the entire team. Meyer is now Chief Architect at AMD. This team then went on to design K9, the original long-pipleine dual quad issue as Greyhound. It was an internal AMD project from 2001 to sometime late 2002. It was about 75-90% completed when AMD realised it wasn't going to deliver. It was a similar 'speed-demon' architecture to Netburst and would have ran into the exact same leakage current problems that caused Willamette to use over 100W at 2.0GHz and Prescott to peak at 150W. Can you imagine dual-coring THOSE? AMD had the dual core patents, Meyer brought them with him as 'on-die SMP', they are also the reason why Intel used MCMs for their "dual core" chips to begin with, they had yet to agree terms with AMD. At the end of 2002, after the refinement team had gone through the whole Thunderbird/Mustang/Palomino (Tbird was a rushed release, it was meant to be Palomino with prefetch/SSE to begin with and Mustang was an Orion/Neptune with 1MB cache integrated. Mustang was renamed Thunderbird and cache reduced to 256k, Thunderbird was renamed Palomino and Mustang as a 1MB cache processor was cancelled) crap, it was decided to shrink K7 to 130nm, double the cache, and leave it at that while they worked on a large-scale refinement of the K7 core, which was K8. AMD knew that it was memory latency and cache speed which most held K7 back. AMD introduced K8's IMC for the former and doubled the L2 cache to 128 bit for the latter. Barcelona integrates the surviving technology from K9 which really isn't much. L2 pre-decode in ECC bits from K8 was also a Greyhound technology and L1 tracing was abandoned entirely. The above is taken from internal AMD documents, AMD's own presentations and developer briefs and informal chats with AMD employees in the 2002-2003 timeframe. Wayne Hardman (talk) 04:45, 14 August 2008 (UTC)

I think this this page AMD K9 should be kept. It should not be merged into other pages. K9 was an interesting microarchitecture, distinct from other microarchitectures (i.e. K9 was not in the K7-K8 family). It may be possible to add more useful information to this web page, e.g. by reference to patents such as those by Mitch Alsup, Jay Pickett, and Mike Filippo. —Preceding unsigned comment added by 134.134.139.71 (talk) 18:52, 16 January 2009 (UTC)

There is something called a "Greyhound core" - AMD has a hardware simulation software package named SIMnow also available to the general public link: http://developer.amd.com/tools/simnow/Pages/default.aspx

The concept of a greyhound core is a single physical x86 computational core - In SimNow you can add Greyhound cores together with some other logic and simulate a dual, tri or quad core equivalent of a 'normal' phenom cpu.

So in that respect it kind of makes sense that a K9 was a dog racing at the project track ;-) 85.81.121.107 (talk) 18:31, 16 January 2012 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 3 one external links on AMD K9. 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 https://web.archive.org/20070906163444/http://www.theinquirer.net:80/default.aspx?article=27421 to http://www.theinquirer.net/default.aspx?article=27421
 * Added archive https://web.archive.org/20070210133934/http://www.theinquirer.net:80/default.aspx?article=37444 to http://www.theinquirer.net/default.aspx?article=37444
 * Added archive https://web.archive.org/20070313231334/http://www.syndrome-oc.net:80/articles.php?article=94&lang=en to http://www.syndrome-oc.net/articles.php?article=94&lang=en

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers.—cyberbot II  Talk to my owner :Online 07:30, 10 January 2016 (UTC)