Talk:Torrenza

Major Update and Disclosure of the Appearance of a Conflict of Interest

• The page has been updated to clarify the definition and scope of the initiative, as well as its relation to other organizations, programs, projects, and technology.
• These updates were made by an employee of AMD [1], who happens to be the technology leader for the Torrenza program.
• This contributor is an internationally known authority in high performance computing and has a strong track record of objectivity and employer-independence as creator and owner of the STREAM Benchmark [2]
• Because of the potential for a lack of objectivity in this area, independent review and clarification are critical and will be greatly appreciated.
• Although the entry is now longer than the previous version, it seems best to keep it short to minimize the temptation to advertise. The content is currently focused on the definition and scope of the project -- related technical content should probably be included in pages describing the specific technologies, rather than a page describing AMD's business + marketing + technology program.

John D. McCalpin, Ph.D. 16:58, 10 February 2007 (UTC)

Extention & Cleanup Tags

• Since the article is marked as a stub I removed cleanup and extention tags as they are pointless.
• Re. the cleanup and request for expansion banners:

Not a lot is known about the subject, as most information is from press releases (hence the stub branding). Once more information is known, it will be added.

Torrenza is not a kind of HTX.

Yes, the first goal of Torrenza is to plug a co-processor in a HTX slot and then communicate with CPU using HyperTransport. However, Torrenza is also a technology that incorporate co-processors into one single CPU package. Please see the slides from speakers at the AMD Technology Analyst Day [3]. --202.71.240.18 06:43, 25 July 2006 (UTC)

Page out of date

First, according to Tom Yager at InfoWorld, Torrenza is code name for the Next Gen platform (which you may already have assumed from reading the article). However, Barcelona, which is to be released mid-2007, is said to be the first chip from this intiative. [4]. It's been said (possibly wrongly) that Barcelona comes under K8L, but this was the Turion 64. [5]

So assuming this is all true, the article needs to be updated, along with the inaccurate article currently called K8L. In terms of the first chip to come from the Torrenza intiative, Barcelona, there's a lot of current information. According to reports, it's to be 40% than Cloverton, and will be released mid-2007 [6]. Further, it's supposed to have 128 bit wide SSE, be 80% faster in floating point over Opteron, offer new VM and power management techniques. Dedicated L2 cache per core, and a L3 cache [7]. So, again, this is the biggest revamp since 2003. The first L3, and 65nm process for AMD [8]. Nja247 (talk • contribs) 13:02, 11 February 2007 (UTC)

AGAIN, K8L is not equivalent to Torrenza... It's the one last time that Bacelona is not a chip from the TORRENZA initiative!, "Bacelona" is a chip under K8L or K10 (K8 naming from Intel or the Inquirer's Charlie; and K10 naming according to the Inquirer article number 37444, link above.), it features 2 MiB L3 cache, and is a Quad-core CPU. Torrenza on the other hand, is a coprocessor initiative, allowing HTX add-in cards as well as the ultimate goal towards "accclerated computing" era, that is coprocessors integrate into one CPU package, or even on one piece of silicon. I cannot see "Bacelona" has coprocessors integrated onto the die, though it is a server CPU under the Opteron brand... And that the first CPU to be under this initiative is yet to be announced, but there is a project named "Fusion", it's similar to that but the coprocessors are replaced with GPUs. And while K8L/K10 introduced a modular approach for easier design of CPU, as well as paving way for the Torrenza initiative as well as "Fusion", but I am yet to see that the coprocesors are now being integrated into "Bacelona" CPUs. --202.71.240.18 07:00, 13 February 2007 (UTC)

Page is Not out of date

Tom Yager's article at Infoworld uses the label Torrenza incorrectly. The revisions I made to the article on 10 February 2007 provide an accurate and (hopefully) clear explanation of what 'Torrenza' means. This can be easily verified by reviewing AMD's public web sites and presentations linked to from the main article. John D. McCalpin, Ph.D. 16:58, 17 February 2007 (UTC)

Well it does seem out of date now. Five years later, the web sites are defunct. So I used archives and changed to past tense at least. W Nowicki (talk) 18:56, 29 May 2011 (UTC)

Related projects: POWER7

Quote: "It is rumoured that the future IBM POWER7 processors are socket compatible with Opteron processors"

What does this have to do with anything related to Torrenza? It's very out of place there and it looks like something that'd belong to POWER7's article or possibly Opteron's, but its relation to Torrenza seems non-existent. --M.A. (talk) 09:57, 15 May 2008 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified one external link on Torrenza. 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/20070927060315/http://www.instat.com/promos/07/dl/IN0703889WHT_DahenUd9.pdf to http://www.instat.com/promos/07/dl/IN0703889WHT_DahenUd9.pdf

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) 09:16, 30 December 2016 (UTC)

To anyone who edits this page, I have the following information. The Torrenza program was started because of my work (see https://docs.google.com/presentation/d/1azrWvRj-xehnWt2vkA_6rA4Jo-a3wE0uGgbE5liU_dI/edit?usp=sharing). A few weeks after the Opteron came out I figured out that I could put an FPGA in one of the Opteron sockets. I got in touch with Peter Robinson in the Boston HPC center and said I wanted to do this. I then raised money, moved to San Jose and built a small board that put an FPGA into an Opteron socket. Two years later we got in touch with Peter again (who came out to visit us) and he kept telling me "you're not the only smart guy to wants to do this" Third time he said it I said "was it a company called Virtual Computer Corporation from Reseda?". He scrambled to his laptop opened his email and confirmed the letter was from me. He said "I guess I'll stop saying that". Then someone internal to AMD passed a letter around to the top executives with the title "Should we help these socket stealers?" The answer was yes and they created the Torrenza program. The program went on for many years. In the meantime Intel came to me and asked if I could do the same for Intel parts. I said "I need your top secret information on your bus". They gave me the data sheets for their front side bus. I concluded that this was indeed possible but they would have to bin out parts that could make the timing so they went directly to Xilinx. Intel then came up with a similar program because customers kept asking how Intel was going to respond to the FPGA in the socket technology that I had come up with (and patented). In the end AMD was going to shut down the program but I convinced Jay Owen that they shouldn't officially cancel the program because they had Intel spending lots of money on it. Steve Casselman 10/4/2020. — Preceding unsigned comment added by 50.45.170.57 (talk) 09:30, 4 October 2020 (UTC)