Talk:Opteron

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

Unless I'm mistaken. Itanium has a complete 64-bit instruction set, whereas Opteron implements AMD64 instruction set, which is just an extension of x86 32-bit instruction set by a few 64-bit instruction and larger virtual addresses. So for it is not surprising that Opteron runs ``legacy`` code at full speed, since it is near-native. It is like an additional real mode... --MarSch 15:48, 3 November 2005 (UTC)[reply]

The reason the Itanium runs 32-bit code slowly has nothing to do with the instruction set at all in fact. It is the architecture's approach which is completely different, namely that the Itanium runs code using EPIC instead of out-of-order. This means that the IA-64 is not made to analyse the code's parallelistic features speedily since this is done at compiletime in EPIC. See IA-64. 03:39, 16 November 2005 (UTC)
It's also worth noting that Intel made the 80386 by extending the 16-bit instruction set of the 80286 and previous processors. AMD basically repeated the same trick in making the Opteron/Athlon64 - extend registers to 64-bits, add more instructions to work on the extended registers, and add a new execution mode. So, MarSch, it could also be said that it is no wonder the 386 runs 'legacy' (8086/286) code at full speed since it too is near-native. Imroy 10:56, 18 November 2005 (UTC)[reply]

some linkes in the article (to internal other ones, like the 8 processor, etc, dont work...)

Cell comparison[edit]

What is the point of the comparison with one specific microprocessor? And why is this comparison placed at the beginning of the article (in the introduction)? Theses are fan-boy statements from people who don't have a clue that CPU performance is a bit more than peak raw computational power. This alone is no metric for performance in real-world applications.

I bet I can find code which runs slower on Cell than on Opteron. But this is not the point. The statement that Cell is faster than Opteron is overly generalized and should therefore be deleted or at least be moved away from the intrduction. --85.216.19.4 00:42, 30 July 2006 (UTC)[reply]

Agreed, I removed the whole statement with the cleanup tag since I didn't think the "competitors" listing added any value beyond what is already mentioned in the article. -- intgr 01:08, 30 July 2006 (UTC)[reply]
My problem with the paragraph was mention of the UltraSPARC T1 as a competitor. The T1 is really only good for highly-multithreaded workloads with little floating-point work e.g web, database, email, etc servers. I would consider the other processors, even the Cell, to be better all-round performers. The Intel Xeon is mentioned in the opening paragraph as a competitor, perhaps the POWER5 and/or PowerPC 970/G5 could be added? Imroy 07:15, 30 July 2006 (UTC)[reply]

The same claim was added again on 2006-07-31. I personally think that the floating-point performance comparison to the Cell processor has absolutely no relevance to the leading section of this article since the Cell is architecturally different. If that statement was removed, then I don't think the rest ("Opteron is capable to perform any type of double precision scientific calculations on the par with Itanium") adds anything to it either. I've reverted the whole paragraph again. Ideas? -- intgr 10:39, 31 July 2006 (UTC)[reply]

=Cache coherency[edit]

When an Opteron accesses ANY memory (local or remote), it must snoop all other CPUs in the system. That is a fundamental aspect of cache coherency, and you can see this if you look at Fred Weber's presentations at Hot Chips or MPF. All memory accesses require bandwidth on HT to communicate with other processors. —The preceding unsigned comment was added by Dkanter (talk • contribs) 11:35, 10 February 2007 (UTC).[reply]

List of Opteron Supercomputers[edit]

On PowerPC page, they mention IBM and Virginia Tech supercomputers based on PowerPC. I had created a section about top 10 supercomputers in the world based on Opteron CPU. —Preceding unsigned comment added by 210.210.145.7 (talk • contribs)

[edit]

AMD gives permission to the public to use their image as long it's not manipulated in any way and is the same one given out to the public. You can see the guidelines here [1] ViriiK 08:06, 13 October 2006 (UTC)[reply]

But it's only usable under the Fair Use regulations in the Wikipedia as it's not available under a free license and free from restrictions. --Denniss 08:09, 14 October 2006 (UTC)[reply]

Socket 939 and AM2 support[edit]

You guys do realize that there are no motherboards ever produced for socket 939 (and possibly AM2)that support Opteron? I thought this was interesting, but I don't see it listed in here. Now when I say support I mean the companies don't advertise support for the CPU's, but they will run just fine on a system. I have an Opteron 165 on my 939 system and it out performs my cousin's X2 3800+, but mine isn't advertised as supported by ASUS A8N5X motherboards. —The preceding unsigned comment was added by Severen (talk • contribs) 14:04, 19 March 2007 (UTC).[reply]

There are definitely mobos which support the 939 and AM2 Opteron. Perhaps you're getting confused. AMD has refused to allow mobo vendors to officially support the Opterons which are intended for workstations & servers in their consumer oriented mobos. Therefore even ones like the DFI NF4 don't officially support the Opteron even if it is/was obviously an important part of the DFI NF4 market due to the popularity of both amongst the enthuasiast crowd. However workstation & server mobos like Tyan ones do support the Opteron (but not the A64s of course) Nil Einne 10:30, 28 April 2007 (UTC)[reply]

no 939 Opteron boards?[edit]

Are you sure? I've been using this Tyan board (link below), and this is what Sun uses (used?) in their Ultra 20 workstation:

http://tyan.com/product_board_detail.aspx?pid=85

MattSF 02:57, 28 March 2007 (UTC)[reply]

121x HE[edit]

My hoster offers me Opteron™ 1210 HE, Opteron™ 1212 HE, Opteron™ 1214 HE, Opteron™ 1218 HE. What is HE? Can this be added to the article? Tobias Conradi (Talk) 17:52, 24 April 2007 (UTC)[reply]

x86-64 vs AMD64[edit]

This article says "AMD64 (formerly x86-64)" where the athlon page has "AMD64 (later renamed x86-64)". Which is right?

Both are wrong, it has always been and still is AMD64. It was probably M$ trying to name it x86-64 but not AMD. --Denniss 06:57, 10 June 2007 (UTC)[reply]
I have a stack of books titled "AMD x86-64 Architecture Programmer's Guide", from AMD, sitting in my closet. They are from 2002. That was the original name as used by AMD. — Aluvus t/c 07:09, 10 June 2007 (UTC)[reply]

floating point performance?[edit]

could some knowledgable person add a few words about the performace of the floating point and vector processing units of these processors? comparison with athlon would be welcome. compariosn with xeon and ppc g5 might be helpful as well. —The preceding unsigned comment was added by 68.80.214.27 (talk) 19:43, August 23, 2007 (UTC)

opteron vs athlon[edit]

so amd has these two product lines opteron and athlon each aimed at different market segments, but marketing aside, how are they similar? how are they different? what makes an opteron an opteron? there are a few words about this, but i'd welcome a bit more. —The preceding unsigned comment was added by 68.80.214.27 (talk) 19:45, August 23, 2007 (UTC)

Enterprise servers only > 8 sockets?[edit]

Someone must have a strange idea about what an "Enterprise" computing environment is. In the article, it states:

The Opteron CPU directly supports up to an 8-way configuration, which can be found in mid-level servers. Enterprise-level servers use additional (and expensive) routing chips to support more than 8 CPUs per box.

I don't know any "Enterprise" environments that run ANY servers > 8-way SMP. And, I've seen a lot of very, very large data centers. (The whole point of x86 servers was always that they are inexpensive, and you can run a bunch of them. But I digress.) That's not to say that they don't exist and that maybe a few out there have one or two 16-way servers, but it's absolutely not common. Enterprise environments generally run 4-way systems for high end servers..

The only time I've seen Opteron systems over 8 sockets is for custom built systems for super-computing purposes, or for special systems like the 16-way Fabric7 server that is intended for high end engineering tasks (ie- processor design and modeling.)

I think this should be changed.

But hey, if an 8-way, Dual-core Opteron system is considered Mid-range these days, let's leave it the way it is =) —Preceding unsigned comment added by Cbreaker (talk • contribs) 06:56, 31 August 2008 (UTC)[reply]

Workstation?[edit]

Is Opteron a workstation CPU like Xeon? —Preceding unsigned comment added by 68.173.208.135 (talk) 21:09, 31 March 2009 (UTC)[reply]

Intel Xeon vs Opteron[edit]

The article needs to be updated as it talks about Nehalem as an upcoming technology. It seems to be comparing to a Xeon with Core architecture and no integrated memory controller. The new Xeons have an integrated DDR3 controller. —Preceding unsigned comment added by Hill-Beast (talk • contribs) 07:09, 3 June 2009 (UTC)[reply]

Part numbers[edit]

Is the section on part numbers really necessary? This is an article in an encyclopedia, not a data sheet. Rilak (talk) 00:51, 12 July 2009 (UTC)[reply]

It seems that it is mostly redundant to List of AMD Opteron microprocessors. — Aluvus t/c 05:41, 12 July 2009 (UTC)[reply]

Roadrunner on the Opteron-based supercomputer list[edit]

The "Supercomputers" section has a list of "Supercomputers based on Opteron mentioned in the top 10 fastest supercomputers in the world." I've included The #1 ranked IBM Roadrunner with a description of the configuration as a Opteron/Cell system, but it keeps getting removed. Stop it. The computer has six thousand Opterons in it. It deserves some mention even if they "are mainly used to control I/O." According to the IBM Roadrunner article, "As a supercomputer, the Roadrunner is considered an Opteron cluster with Cell accelerators, as each node consists of a Cell attached to an Opteron core and the Opterons to each other." The computer would not have the performance it does, and thus its rank on the list, if it wasn't for the Opterons. --216.254.228.94 (talk) 20:11, 1 August 2009 (UTC)[reply]

Dual Socket AM2 Opterons[edit]

The article states that socket AM2 Opterons are only available in single CPU arrangements. I know of at least one server line that uses dual Socket AM2 Opterons, namely the IBM System x3655 (i.e., 7985-xxU). These can have up to two Socket AM2 Opteron processors. Kleinhhl (talk) 17:16, 16 December 2011 (UTC)[reply]

IBm page talk about 22xx Opteron, those are Socket-F CPU for dual-processor systems [2]. --Denniss (talk) 19:22, 16 December 2011 (UTC)[reply]
Could it be that IBM made its own dual socket AM2 system boards? Or that they are AM2+? The IBM xSeries Reference (xRef) specifically states that certain x3655s have two Socket AM2 sockets on board: For the x3655 (7985-xxU) - 2-way SMP / up to two Opteron Processors (same speed/cache size required) System board contains two AM2 sockets, sixteen memory DIMM slots, integrated SAS & SATA, dual ethernet, graphics, one PCI-E x4 slot, two PCI-E x8 slots and one HyperTransport (HTx) slot via riser card (on page 285). IBM xSeries xRef (Withdrawn) Kleinhhl (talk) 22:11, 18 December 2011 (UTC)[reply]
They would have a hard time trying to fit Socket F Opteron 22xx and 23xx into an AM2 socket - maybe with a hammer? I don't know why they wrote AM2 socket. AM2 does not even support 8 memory slots per CPU. --Denniss (talk) 01:07, 19 December 2011 (UTC)[reply]
Okay, you're right. A quick look at the AMD website for the 22xx series of Opterons confirms that their all Socket F. I'm not sure why the IBM document would state Socket AM2 either. It's definitely a typo or mistake in the IBM xRef. Kleinhhl (talk) 01:46, 19 December 2011 (UTC)[reply]

[edit]

The new logo corresponding to the Bulldozer processors (Valencia/Interlagos) should be added to the article as the main image, and in the models chart. It's this one. Digital infinity (talk) 18:12, 20 February 2012 (UTC)[reply]

Cray Titan[edit]

Someone (more knowledgeable than me) should add Titan, the supercomputer currently ranked #1 according to TOP500, and remove Jaguar, from which the Titan was upgraded to, from the list of supercomputers. SubTenebra (talk) 12:35, 17 November 2012 (UTC)[reply]

External links modified[edit]

Hello fellow Wikipedians,

I have just modified one external link on Opteron. 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:

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 18 January 2022).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 12:22, 21 December 2017 (UTC)[reply]

Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Opteron&oldid=1204445401"