Talk:Power Architecture
This is the talk page for discussing improvements to the Power Architecture redirect. This is not a forum for general discussion of the article's subject. |
Article policies
|
Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL |
![]() | This redirect does not require a rating on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
Aggressive editing
[edit]I feel that the editing that 68.15.20.63 has made are somewhat inappropriate and a tad agressive, especially the language and the attitude in the edit summaries. I invite him to discuss his editing here if he's got larger issues with the article. He obviously knows his stuff, and it would be great if he could contribute in a civilized manner but for now all I really want is to know the reasons and sources behind his edits since they contain information tha's not readily available. I want to edit his contributions, since I think they are part wrong, part off topic and part just bad writing, but I fear that I will get unfairly bashed for doing so. I've started a Power.org page that's more appropriate for infromation regarding how that organization is run and I hope that he'll contribute to it. If my wording comes across as biased due to me being a Mac user, I'm so very sorry.. That's certainly not my intention, and I applaud any correction in the right direction. It's really more the opposite since I feel that several PowerPC articles are influenced too much by Apple, like PowerPC G4 and PowerPC G3. Please: let's cooperate! -- Henriok 18:06, 2 November 2006 (UTC)
ok, i'll respond.
- Great. I really appreciate it! -- Henriok 15:19, 3 November 2006 (UTC)
Paragraph 1. begins with "Power Architecture is a broad term to describe similar instruction sets for RISC m...." 90% inaccurate.
- Care to elaborate? What exactly is inaccurate? How did you calculate that it was 90% inaccurate? Do you have any references to contradict what it says in the article? -- Henriok 15:19, 3 November 2006 (UTC)
Paragraph 2. begins with "The "Power Architecture" term sho....." 99.9999% inaccurate.
- Care to elaborate? What exactly is inaccurate? How did you calculate that it was 99.9999% inaccurate? It's so extraordinarily wrong that I think that IBM and Freescale would be surprised by our calculation, their marketing have reflected excatly what the paragraph says since a couple of years. Do you have any references to contradict what it says in the article? It would also be interessting to know what's 0.0001% accurate. I might learn something. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 3. begins with "Power Architecture began its life at IBM i...." 45% inaccurate.
- Care to elaborate? What exactly is inaccurate? How did you calculate that it was 45% inaccurate? -- Henriok 15:19, 3 November 2006 (UTC)
Paragraph 4. begins with "in 1992, Apple, Motorola and IBM formed the AIM alliance to develop a mass market version of the POWER pro...." 65% inaccurate.
- Care to elaborate? Did they not form the alliance in 1992? Did they not specify the PowerPC specification? Did they not want to develop a mass market version of the POWER processor? Was the first implementation not the PowerPC 601? Was it not based on the RSC? Did it not end up in PowerMacs and RS/6000? I'm finding it very hard to believe that it is 65% inaccurate. Care to break it down for me? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 5. begins with "IBM expanded their POWER Architecture for the....." 10% inaccurate.
- Ooooh.. only 10% inaccurate. I'm honored! How can I make it 0% inaccurate? You are so very precise in your calculations that it must be an easy task to point out exactly what's wrong, not just blurt out that something's wrong. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 6. begins with "In the early 1990s IBM sought to repl...." 75% inaccurate.
- Amazing.. but I'm glad that some things are right. Care to elaborate what's wrong with it? It's 75% wrong so it can't be that hard. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 7. begins with "The AIM Alliance kept developing PowerPC in 1995 through 1997 and released the second generation PowerPC process..... 30% inaccurate.
- I'm glad that this paragraph was relatively accurate. It is one of the longest. Care to elaborate on what's wrong with it? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 8. begins with "The last effort of the AIM Alliance w..." 20% inaccurate.
- Again.. 20%.. I'm stating to belive that this is a good sign. It's better than 99.9999% inaccurate. but it's still one finth wrong. Care to help me get it right? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 9. begins with "In 1998 came POWER3 which unified the PowerPC and POWER2 ar......." 40% inaccurate.
- 40%. Pretty good, but almost half is wrong. It's a short paragraph and I really had hoped that my lies and my FUD could go unnoticed. Care to point out what it is about it to make it 40% inaccurate? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 10. begins with "2000 saw the last implementation of the Amaz....." 20% inaccurate.
- Again.. pretty close, but no cigarr. I realy don't know what's wrong with it. I'm not trying to delude the public, I'm trying to write truthfully. Why is this 20% inaccurate? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 11. begins with "In 2001 IBM introduced the POWER4 which unified an....." 70% inaccurate.
- This is amazing. Just one sentance, but it's 70% inaccurate. Did they not introduce POWER4 in 2001? Did it not include the instruction sets of POWER3 and Amazon? Wasn't the new instruction set calles PowerPC ISA 2.00? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 12. begins with "In 2002 Apple desperately need a ne...." 65% inaccurate.
- OK.. Perhaps, Apple wasn't desperate.. it might be considered biased, and not NPOV. I don't think it was inaccurate though. And the other part of the paragraph is pretty accurate I think. Care to elaborate on what's wrong? -- Henriok 15:19, 3 November 2006 (UTC)
- WOW! The 13th paragraph was the first that was absolutely 100% accurate! Amazing. It's 100% amazing that that paragraph was 100% accurate. Iäm personally not 100% satisfied with that sentance though.. Oh well. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 14. begins with "Culturecom lice...." 2% inaccurate.
- Wow! This is really amazing. I almost got this one exactly right! 2% is almost negligible, but I aim high. I want it to be 100% accurate. What can be done to make it just perfect? just 2 measly percent left! Care to help me get it absolutely right? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 15. begins with "POWER5 from IBM, introduced in 2004, is an evo........" 65% inaccurate.
- Oh damn.. now I got hit again. Doh! This paragraph is more inaccurate than it is accurate. Really now? What the h*ll is inaccurate in this paragraph? Can you please help me? Feels like my writing is some kind of russian roulette. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 16. begins with "AMCC buys IBM's 130nm fab, and lic......" 4% inaccurate.
- Aah.. a paragraph you yorself wrote, and it's still inaccurate. And.. I really can't find anywhere, not in your extensive reference collection you provided, that AMCC bought IBM's 130 nm fab. And by they way.. Your edit broke the grammar in the sentance. I'm beginning to believe that you are just making the figures up. You can't actually be serious that this paragraph is only 4% inaccurate when paragraph 11 is 70% inaccurate. How did you calculate these numbers anyway? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 17. begins with "Power.org is founded in 2004 by IBM....." 2% inaccurate.
- I was about to ignore this paragraph since it got such high grades, but then I readu it again an concluded that it was quite off topic. The date format is wrong, and the last sentance, though it is correct (by my standards), just don't make any sense. And the erference you gave doesn't back up your claims. -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 18. begins with "In 2005 Apple announced that it's droppi......" why is this even here... 5% inaccurate.
- No, actually. Paragraph 18 begins with: "involved in developing products revolving around the.." and it's what's left of your edit. I'd say that this paragraph is 100% inaccurate since it just doesn't make any sense any more.
- You might mean paragraph 19.. While I can't find what's inaccurate about it since Apple DID announce that they were dropping PowerPC in 2005, and begun doing so in 2006. I might agree with you that it is off topic. However.. Since Apple was in the AIM alliance and they did contribute to the adoption and specification of a lot that's Power Architecture, and still sits on patents.. I think it's this paragraph is very relevant. For all I know, Apple is still holding PowerPC licenses and are still able to sell licenses and technology to whomever they please. This might be wrong though. It would be interessting in knowing for sure. -- Henriok 15:19, 3 November 2006 (UTC)
Paragraph 19. begins with "2005 also saw the specifications of the Cell processor,[3] jointly developed by IBM, Sony and Toshiba over a four ye...." 7% inaccurate.
- I'm content by it being only 7% inaccurate, but I'm still curious of that I did wrong. Care to help me? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 20. begins with "P. A. Semi licenses PowerPC techno....." 20% inaccurate.
- Even though I don't know what's wrong with it, I DO know that the date is right, and that they are going to build PWRFicient processors based on the PowerPC spec, using the license that they bought from IBM. So.. what's left? -- Henriok 15:19, 3 November 2006 (UTC)
- I notice that there are two paragraphs here that either went by unnoticed or are accurate. Great! However.. You leave the paragraph about Rapport without any objections, and still erase Rapport from corporations manufacturing Power Architecture processor. I find that curious. What gives? -- Henriok 15:19, 3 November 2006 (UTC)
paragraph 23. begins with "Power.org released the Power ISA version 2.03.[4] in Septem...." 10% inaccurate.
- I'll just take your word for it. eerrr.. no! Care to provide some more specific arguments about what you believe is wrong? -- Henriok 15:19, 3 November 2006 (UTC)
- I see that the last three paragraphs did pass the scrutiny. And all the other parts. I'm glad. -- Henriok 15:19, 3 November 2006 (UTC)
This list you provide doesn't help anyone, and the little help you did provide by editing the article just left it broken and polluted. I'm not trying to be stubborn here, but you aren't really helping. Those inaccuracy figures you have is hilarious, especially since you don't back them up, either with arguments or references. This is just hilarious. Care to be anything other than a troll and malcontent? I'm really trying to make a really good article here, and I really don't want to make it alone, and I really wan't to use accurate and reliable information. However.. This is not intended to be an end all, be all article about all things Power Architecture. Every scrap of information is not wanted and I think it's perfectly OK to cut some corners and condence a lot of boring nit picking to readable paragraphs. It's not about getting the whole truth, cause it's pretty large and it would ammount to books and books of information. I'm not the man to write the compelte book on Power Architecture. I'm satisfied just to write a decent article on wikipedia, a hub for further reading.. Perhaps beeing the one place on the Internet where you get a compelte overview of what's Power Architecture. That includes working on sub pages and it includes making additions to this page. Like.. I'm really cant define the technical specifications. What's the differnece between Power and MIPS, ARM and x86? Stuff like that. Theres really now place on the Internet that provides that information. Wikipedia should though but I'm not qualified. Care to help? -- Henriok 15:19, 3 November 2006 (UTC)
Aggressive Whining
[edit]- Good to get this out in the open :) I like agressive whining, as long as all parts are living by the same rules. -- Henriok 22:13, 13 November 2006 (UTC)
Who created the "POWER Architecture"? IBM..
- I agree completely! That's why I'm desperately trying to keep "POWER architecture" and "Power Architecture" apart. This article is about the latter, and how it's used today. Today, Freescale, IBM and others are using the camel case "Power Achitecture" to denote PowerPC, POWER and Cell, while the all caps "POWER" is still IBMs sole responsibility. Just adding to the cunfusion. The history of Power Architecture must include the histories of POWER, PowerPC and Cell, but I see no point to delve deeper than I aldready did. If someone want the finer details in PowerPC, POWER and Cell, there are specific articles about that. The mission here is to establish a base of the what and why of Power Architecture. There's a separate article about "POWER architecture" where you probably have a lot of issues.. keep then there though. I'll merge all appropriate sections to this.-- Henriok 22:13, 13 November 2006 (UTC)
What is PowerPC? it is a subset of POWER, or the 32-bit portions of the 64-bit POWER Architecture.
- What PowerPC is isn't really a topic of this article, there's a separate PowerPC where this discussion is more appropriate. However.. I'm not sure that PowerPC is just a subset of POWER. There's some stuff from Motorola 88000 and some things uniquely PowerPC, especially when you include Book E, Altivec and stuff Motorola added like APUs. This is more true than ever since "Power ISA 2.03".
Does AIX run on PowerPC? No.
- Well.. yes it does! It do run on PowerPC 604 and 970. They are not POWER, only PowerPC. But that's really not an issue for this article since it most certainly runs on "Power Architecture" -- Henriok 22:13, 13 November 2006 (UTC)
But what about AIX v3.??, it ran on the PowerPC 604? The 604 was the first to define the PowerPC ISA v1.{don't forget PowerPC means 32-bit}, which was incompatible with the 32-bit portion of POWER2, Now if your to consider the development cycle of POWER2 which started several years before the thought of PowerPC came to mind you might be able to understand why POWER2 was not 100% compatible with the PowerPC v1. ISA.
How many processors does IBM work on at a time, is it one or forty? Who knows. But IBM has always said even as early as just before the launch of POWER2 that they where currently working on POWER3 and POWER4. And from this we know they work on at least POWERn+1 and POWERn+2. So when IBM said that the "AMERICA Project" would be the platform that would replace there "AS/400 platform" and "System/390" it is evident in there commitment to POWER.
- All very interessting! But an non issue of this article. This should be added to IBM POWER or PowerPC. And I mean this in the most positive respect. They shoud be added to their appropriate articles. I'm really not qualified to write that since all documentation is unavailabe to me, and it was so long ago that I wasn't interessted in this stuff then. I'm not sitting with first hand information from back then. -- Henriok 22:13, 13 November 2006 (UTC)
Does OS/400 run on PowerPC? No.
- No it doesn't, but it does run on Power Architecture, which is the topic of this article. -- Henriok 22:13, 13 November 2006 (UTC)
But what about OS/400-AIX on the PowerPC 620? First we know that PowerPC is a 32-bit arch(but 620 is 64-bit}, so the 620 was PowerPC compatible. And the 64-bit aspect of 620 was classified just in the same way as the [POWER Architecture, POWER3, POWER4] because it was being refined through various projects {PowerPC 620, POWER3, Apache, RS64, RS64II, RS64III, RS64IV, POWER4}. So you have to realize that OS/400 started on CISC and has been migrating to RISC {RSC-PowerPC620-Apache-RS64-RS64II-RS64II-RS64III-RS64IV-POWER4}
- PowerPC 620 was NOT classified in the same way as POWER or RS64 class processors. 620 was jointly developed by the AIM alliance in Somerset, just as 601, 603 and 604 was, by Motorla (who manufactured some) and IBM. It was offered publicly and used by third parties, something that no POWER nor RS64 processor have been. AND.. PowerPC was a 32/64 bit architecture always. So says all documents that I've read and it's the official truth that's been communicated through the AIM-alliance since day one.
Does AIX run on POWER? yes.
Does OS/400 run on POWER? yes.{starting with POWER4}
Can any code compiled for POWER run on PowerPC? No.
Can any code compiled for PowerPC run on POWER? yes.{with the exception of POWER1}
Can any code compiled for Power run on POWER? Yes.
Can any code compiled for POWER run on Power? No.
what is the difference between PowerPC and Power? PowerPC is a 32-bit architecture and Power is a 64-bit architecture.
- This is just wrong, or as you might call it, "100% inaccurate". POWER was a 32 bit architecture (before POWER3), and PowerPC is, and have always been 64 bit even if it wasn't implemented until 620. -- Henriok 22:13, 13 November 2006 (UTC)
What is the difference between Power and POWER? It similar to how PowerPC was defined, IBM said all code written to the PowerPC ISA v1. specification would be supported in future processors, so now IBM has release the Power ISA v2. Which means all code written to the Power ISA v2. specification will be supported in future POWERn processors.
How stable is the PowerPC ISA? very, there's been little change since being define almost 13 years ago.
How stable is the Power ISA? Even though the Power ISA is fairly new its been a work in progress at IBM for over 17 years thru research applied towards the "POWER Architecture"
How stable is the POWER ISA? POWER is volatile it may break compatibility from one generation to the next, yet IBM creates the tools to maintain code transportability. POWER has maintained the constant PowerPC ISA compatibility and now that IBM has defined the Power ISA as an open standard it's assumed that the Power ISA will be a constant within the POWER ISA.
When will System/390|zOS, run on POWER? Who know, but POWER5 incorporated field-gate-programable-logic. And zOS Systems use POWER5 processors alongside CISC. So we might see POWER6 taking on more responsibilities in zArchitecture.
Does IBM guaranty that code written to POWER will compile on PowerPC or Power? No.
Does IBM guaranty that code written to PowerPC will be compatible with future processors? Yes.
- Very fascinating! And I'd VERY MUCH appreciate a sub section to this article which specifies the ins and outs of precicely this, what exacltly is Power Architecture from a technical standpoint. A description of the RISC arcitecture, and so one might separate it from similar description of ARM, Sparc, MIPS or x86 for that matter. But in a more narrative manner perhaps. -- Henriok 22:13, 13 November 2006 (UTC)
Since 'PowerPC' means "Power Performance Computing" and all POWER processor are PowerPC compatible and POWER means Performance Optimization With Enhanced RISC and PowerPC are RISC processors optimized for performance, and PowerPC was enhanced with AltaVec Then aren't all PowerPC's enhanced RISC optimized for performance computing, and the distinction between PowerPC and POWER is trivial, and even Frank Solitis said "When is PowerPC not PowerPC" and did not mention POWER, so are POWER processors just ways for IBM to trick people in to thinking that by buying a POWER processor that has a few hundred-million more transistors then PowerPC that those extra few hundred-million transistors are there just to drive up the cost, and drive up the heat dissipation, and Power does this mean 'Performance optimization with enhanced risk' and does IBM just charge more because 'POWER' is in all CAPS. Or am i to believe that ever company who has a PowerPC license can make any processor which would be compatible with POWER, so the distinction between a Power license and a PowerPC license amounts to nothing more then less typing that i have to do, right? and if there compatible then there the same, almost. But more same then different, right. I'm right, right. Because nothing matters more to me then being right, right. If you don't think so then your not confused enough, let me go on. So IBM-Motorola-apple created PowerPC, and it matters not that engineers form Motorola-apple went to IBM's research campus, and the chief Architect for Somerset was an IBM Fellow. And it so much easier to call PowerPC a creation by IBM-Motorola-apple, and will totally disregard the idea that IBM was only to help Motorola with three designs {601,603,604}, and that Motorola-IBM got together again to work on the PowerPC 700 series, and we forgot just why they got together again, or did we. So from this we got AltaVec{and for those apple-user who didn't know what POWER was in 1996, its called VMX on this side of the fence}, but that doesn't matter it gave us a supercomputer on the desktop, and forever and ever apple injected into every mention of POWER. oh, to hell with all this PowerPC/Power/POWER/RS64/RSC/CELL crap lets just call them all GeMcPOWER, ok ;)
- OMG, that's was an incoprehensible rant. Do you mind if I just ignore it? -- Henriok 22:13, 13 November 2006 (UTC)
umm, well if you keep using the wrong context your surly to burr the lines between fact and opinion.
- Yeah.. as if your last blurb didn't blurr the lines enough. Is it OK that I just touch up the article to my standards ans disregard almost everything you've said since it doesn't contradict, clarify or add anything that was already written? -- Henriok 22:13, 13 November 2006 (UTC)
What is Power Architecture?
[edit]Ok.. the debate continues. Mr 68.15.20.63 seems to have quite a dim view of this. Let's hear what IBM have to say since they ought to know. Quote from ibm.com/power:
"About Power Architecture: It’s PowerPC, POWER4, POWER5."
–– Aha! Both POWER and PowerPC is Power Architecture.. who would have known? Wait! I want so learn more. Let's follow the link! Learn more
"Power Architecture™ technology is an instruction-set architecture that spans applications from consumer electronics to supercomputers. Power Architecture encompasses PowerPC®, POWER4™ and POWER5™ processors."
––Aha! There we have it again! Power Architecture is PowerPC AND POWER. Still not convinced? What about all the other POWER processors, and Cell? What about it? Let's continue digging. Let's follow the link New to Power Architecture technology. <-- Even Power.org links to this page.[1]
What different kinds of processors are based on the Power Architecture platform?
[edit]"Power Architecture family members are stealthy chips, and they like to go by many different names. They can be hard to keep track of because they don't all use the Power or PowerPC trademark." and "..the Cell Broadband Engine™ (Cell BE) processor (which is to power the upcoming Sony PlayStation 3.."' and "The chips in the POWER line are numbered 1 through 5, with POWER6™ anticipated somewhat soonish.".
––So.. Processors are called "Power" and "PowerPC". And "Cell Broadband Engine" and even POWER1 through POWER6, anticipated soonish. Check.. I've included all those in the Power Architecture.
What machines are based on Power Architecture processors?
[edit]"While the POWER line of chips is found only in high-performance servers and workstations from IBM, PowerPC chips are found in everything from embedded devices to supercomputers." and "Contrary to popular belief, IBM mainframes (or "System z™ ") are not based around Power Architecture technology."
––Well.. I did know the latter, and that's why you won't see any mentioning of z9 or Series z anywhere on Power Architecture pages on Wikipedia. There you have it! Power Architecture is PowerPC, it is POWER and it is Cell. Just as I write it, but not as 68.15.20.63 would want you to believe. Hey.. don't take my word for it, take IBM's. Wikipedia is not a place to do original work. Go to the sources. I think they know best. What do you think, Mr 68.15.20.63? -- Henriok 22:16, 6 December 2006 (UTC)
WoW Thats really impressive its taken you 15years to understand that the "POWER Architecture" is more than just the cpu
[edit]first and foremost i'm not trying to prove anyone wrong, because that would be a wast of time{i've tried three years ago}, yet trying to enlighten you. So you will stop slandering "IBM" and the "POWER Architecture". I'll never understand how you equate 'based on' as 'equal too', continuously you try to muddy the waters of logic between POWER and PowerPC and also Power. Over and Over again the emphasis of the "AMERICA Project" is repeated, this is the platform that IBM will build there business upon, The "POWER Architecture". This is a perfect example of why the "IBM POWER" article needs to be in chronological order, so people can see for themselves how the "POWER architecture" has evolved/funded/enhanced. Through the partnerships with with IBM and these Companies{Motorala,Cisco,Xilix,Sony,Microsoft,Nentindo,etc..} for example IBM+MOTO=AltaVec/VMX,IBM+Ciso=7??NP/97?np,IBM+Sony=cell,IBM+Nentindo=VMX128,IBM+Xilix={programable logicTo POWER5/POWER6, etc. instead of having someone argue that PowerPC=POWER because they have no concern for the history, but would rather rewrite history every week to get more WikiPOINTS at the detriment of society. I suggest to all those apple users who knew nothing about POWER in 1993, only of PowerPC. Find material about POWER between the years 1988-1992{hardcover}, and also be mindful of the revision numbers if your reading redbook's because I've seen names change from one revision to another. — Preceding unsigned comment added by 66.27.52.158 (talk • contribs) 16:59, 9 December 2006 (UTC)
Massive redundancy
[edit]Why must the article go over the history of the 801, POWER, PowerPC, PowerPC AS and the various implementations of each ISA? Why not go over the history of CISC as well? Surely the 801 was developed as a result of the analysis of complex architectures. This article only needs to state that the origins of the Power ISA are in the POWER architecture. The history of the Power ISA starts in 2006, when IBM and Freescale collaborated on Power ISA 2.03. Anything else before then is irrelevant and deserves mention in a footnote. Any comments or objections before I begin removing content covered elsewhere? Rilak (talk) 12:35, 30 May 2009 (UTC)
Broken link
[edit]The 2.06 specification link is broken, 2.06B_V2 is the new version. http://www.power.org/resources/downloads/PowerISA_V2.06B_V2_PUBLIC.pdf —Preceding unsigned comment added by 72.12.219.122 (talk) 15:22, 2 September 2010 (UTC)
Conformance and {backwards,forwards} compatibility
[edit]The "Specifications" section speaks of, for example, "Cores that comply with the Power ISA v.2.03" being compliant with the Power ISA v.2.04 as well. Presumably this means that:
- v.2.04 didn't introduce any non-backwards-compatible changes (e.g., changing an instruction opcode, or removing an instruction);
- any new features introduced in v.2.04 aren't required parts of the v.2.04 spec, so that a v.2.03 processor is v.2.04-compliant but omits those optional features.
Is that correct for all the versions of the spec listed in that section? Guy Harris (talk) 23:55, 29 September 2013 (UTC)
- Yes, this seems to be a correct assumption according to what I have gathered skimming through the documentation. -- Henriok (talk) 10:46, 16 January 2014 (UTC)
"Compliant cores" list and references to the previous Power ISA version
[edit]For each version of the Power ISA, there's a set of "compliant cores"; for most versions, the lists begin with "Cores that comply with the Power ISA {previous version}.
This is confusing; people often edit that entry to refer to the current version.
- I can't understand why this is confusing though. A new spec adds some new stuff and also includes products compliant with the previous version. That's it. It's like getting a child. The newborn is a member of the family going forward, but not a member of the previous version of the family. The older members will continue be members. -- Henriok (talk) 10:44, 16 January 2014 (UTC)
- I guess people think it's not an item in the list of compliant cores, but a somewhat-redundant statement that this is a list of cores that comply with the version in question, leading them to think the version number should be the same as the version number for that sub-section, and therefore to make edits such as this one and these ones and this one. If it could be stated in such a way as not to confuse people in that fashion, it might be better. Guy Harris (talk) 10:56, 16 January 2014 (UTC)
- I agree that people are confused, I don't know why though. You are probably better at formulating something better. Any suggestions? -- Henriok (talk) 12:23, 16 January 2014 (UTC)
- I guess people think it's not an item in the list of compliant cores, but a somewhat-redundant statement that this is a list of cores that comply with the version in question, leading them to think the version number should be the same as the version number for that sub-section, and therefore to make edits such as this one and these ones and this one. If it could be stated in such a way as not to confuse people in that fashion, it might be better. Guy Harris (talk) 10:56, 16 January 2014 (UTC)
- I changed it to "All cores that comply with previous versions of the Power ISA" for all the items. That:
- phrases it in a way that doesn't look as if it's a heading for a list of cores (so people don't think it refers only to the cores listed below it);
- doesn't give a version number (so there's nothing to "fix", and so that it explicitly acknowledges the transitivity of compliance - for example, if all cores compliant with v2.04 are compliant with v2.05, and all cores compliant with v2.05 are compliant with v2.06, then all cores compliant with v2.04 are compliant with v2.06);
- so hopefully that'll be less likely to be "fixed". Guy Harris (talk) 22:23, 18 January 2014 (UTC)
Does that sentence imply that the version in question of the Power ISA doesn't add additional restrictions (e.g., requiring that some bits in an instruction must be zero when they were undefined in a previous version), doesn't add additional features using bits previously specified as undefined, and doesn't make some previously optional features requirements?
- A new spec adds new features but doesn't add new requirements. At least that's how it's been from 2.03 through 2.07. No cores have been excluded since the great culling when the Power ISA v.2.03 was set and left various PowerPC and POWER cores like RS64, POWER2, 7xx, 74xx and 6xx behind. At least, not what I have seen. -- Henriok (talk) 10:44, 16 January 2014 (UTC)
And what causes particular processors to be put into the list for a particular version? Does it have to use one of the new features introduced in that version, or does it just have to be released when that version was the current version? Guy Harris (talk) 18:57, 15 January 2014 (UTC)
- A new core is verified against a certain version of the ISA, and becomes compliant to that version. It might be compliant to earlier version too, but that's not tested, and for that reason it won't be compliant with that version. -- Henriok (talk) 10:44, 16 January 2014 (UTC)
There will probably be a really new specification some time in the future (v.3.0?) where a new set of requirements are intoduced and there's a new culling. In that case we might see that a v.3.0 of the ISA keeps some cores compliant but leave others behind. But that doesn't seem to have happened in the last few versions (since v.2.03). -- Henriok (talk) 10:44, 16 January 2014 (UTC)
Cell Information
[edit]I don't understand why Cell belongs in the glossary section. Cell is it's own thing, and the PPC core is just there to manage the DSP units. If Cell belongs in the list, then so do a lot of other things that have been bolted onto a PPC core as well, but that would distract from what the purpose of the list is. I'm going to remove that row, but if someone else feels that it should be there, I won't contest it. — Preceding unsigned comment added by 71.200.57.8 (talk) 08:32, 26 June 2014 (UTC)
- I've put Cell BE here since IBM's grouping it together with everything else that Power Arch, and the general purpose instruction set of Cell BE is Power Arch. Power Arch includes facilities to bolt on accelerators, DSPs, GPUs, FPGAs and other exotic execution architectures while still remaining Power Architecture. I'm not trying to get the SPUs ISA to be Power for the same reason why the QUICC engine isn't, the FPGA part of Xilinx FPGAs and various integrated GPUs isn't. -- Henriok (talk) 12:47, 27 June 2014 (UTC)
- Got it. Thanks for taking the time to explain that.
Do Not Remove Operating Systems
[edit]AmigaOS 4, AROS, and MorphOS have been removed by IP address with weasel worded excuses as being "Obscure hobby operating systems known to a few thousand people which means they are not notable and they are not encyclopedic."
This has never been an issue before and each of these operating systems have established notability with Wikipedia.
- The accumulation of Amiga and Amiga related things has made it an issue. It's overrepresented in this article and as such it will cause readers looking to get familiar with this processor architecture to think that this family of OSs is significantly related and relevant in this space. That's why I'm questioning its encyclopedic merit. 71.200.57.8 (talk) 09:49, 28 June 2014 (UTC)
- There is no accumulation. It is not overrepresented. It does not promote Amiga. The brief mentionings of Amiga clones and operating systems do not harm the article in any way. In Correct (talk) 14:59, 28 June 2014 (UTC)
- I just came across the article, since I recently heard about the architecture. I came to look at the talk page because some parts of the article seem like opinion/advertisement, so I wondered if it was mentioned. However, reading this I wanted to add that operating systems offering even basic support for an architecture, especially one that isn't as broadly used as x86 is certainly something that is noteworthy in an article. I think operating system support is a fact about a noteworthy article, rather than something that must be noteworthy itself. For example that an Intel CPU has a certain instruction or L1 cache might not be noteworthy, however it is an important fact. I do think operating system/kernel support and having complete lists around this is something very elementary to mention about a CPU architecture. Also the notion that various hobby operating systems and a family or class of operating system has at least shown interest in the architecture seems like information one would want to get from an encyclopedic article of a CPU Architecture. Also, since those to my understanding are separate operating systems this seems like a lesser over-representation than Linux Distributions on the Raspberry Pi (which I also don't consider to be something that should be changed). Completeness in terms of operating systems supporting an architecture is something I'd expect in an encyclopedia. I also did not have impression that Amiga is related, since I am sure the article would explicitly mention something like this. If it really becomes excessive I'd consider clustering lists of operating systems (if that makes any sense - not sure if they are clones, forks or something else). As someone who isn't affiliated with Amiga and came here to learn about the architecture I hope I was able to give a rather neutral perspective on this.--Athaba (talk) 10:50, 12 November 2017 (UTC)
Obscure and hobby are opinions and are unacceptable reasons. As for the "few thousand people" excuse, there needs to be an EXACT number and it needs to be proven.
- I understand your point about hobby systems, but I think the obscurity issue is valid because it relates to whether or not the entry belongs in an encyclopedia, specifically in this article. If you took a survery of the IT landscape you'd find that virtually nobody has ever heard of MorphOS or Hyperion's Amiga OS 4. It's impact and notability in industry is approximately zero. It's only known to a handful of enthusiasts. How is this even remotely relevant in an encyclopedia article about the Power Architecture? Should we list every developer board that featured a PPC chip? Should we list every single research kernel that targeted PPC? I see that L4 isn't anywhere on the page. 71.200.57.8 (talk) 09:49, 28 June 2014 (UTC)
- Then I would suggest to separate the list of hardware and list of operating systems into a new article. Also I do not believe surveys are accurate. In Correct (talk) 14:59, 28 June 2014 (UTC)
AROS barely runs on the Power Archecture, but the other two must stay on this article.
- People make the same arguments about Amiga OS 4. Users widely report that the OS doesn't fully support the hardware it runs on. For instance the Amiga One X1000 is only able to use one core on the PA6T. None of the Amiga/Amigaoid operating systems take full advantage of the various PPC processors they run on. The very essence of the architecture of this family of OSs fundamentally prohibits being able to actually utilize the features that make Power Architecture a modern and relevant CPU design. They are the worst examples of an OS for this architecture that could possibly be presented. 71.200.57.8 (talk) 09:49, 28 June 2014 (UTC)
- But you just said nobody heard of AmigaOS 4, etc. Where are all these reports you speak of? In Correct (talk) 14:59, 28 June 2014 (UTC)
In Correct (talk) 03:45, 27 June 2014 (UTC)
In response to the edit comment by IP 71.200.57.8, "There's no need to list every single OS offshoot under the Amiga umbrella, we wouldn't do it for every linux variant)" It appears that "you" would do exactly that. See List of Linux distributions.Jacona (talk) 12:58, 27 June 2014 (UTC)
- I'm not talking about linux distributions, I'm talking about variants. A linux distro has a GNU user land and a Linux kernel. Hence, GNU/Linux. You wouldn't call Android a linux distribution. It has a modded kernel and a non GNU user land. 71.200.57.8 (talk) 09:49, 28 June 2014 (UTC)
I don't edit much on Wikipedia and thus don't know the ropes as well as some of you guys on here. I think it would be useful to get another editor in here to weigh in on this, somebody who's removed from the subject. At least one person commenting in this talk page, and participating on article edits, has obvious bias as indicated on his user page. I'm not interested in participating in an "edit war" or hurting anybody's feelings. I'd just like to help make the article better, it needs some work. 71.200.57.8 (talk) 10:51, 28 June 2014 (UTC)
- I will see what I can do about that. What other changes should be made to this article? The only thing you have done to this article is removed the mentions of Amiga clones and operating systems. In Correct (talk) 14:59, 28 June 2014 (UTC)
- My edit history says otherwise. Perhaps you didn't pursue my commits? — Preceding unsigned comment added by 71.200.57.8 (talk) 02:10, 30 June 2014 (UTC)
Historically there no denying that Amiga has an important part in home computing history and Amiga is not dead, and it has a small but sprawling little community that 71.200.57.8 seems to be aware of. Just like a democracy where in-proportionate efforts are made to accommodate minorities, my view on what an encyclopedia is is to give a wide and representative view of a subject, and not think too much about the proportions of importance within the subjects that comprise a realm of knowledge or article. One could argue that IBM i and AIX is so niche that if you ask most IT professionals they have never heard of them, or Power Arch itself for that matter. Should we delete them from this article too because of that reason? No, ignorance is a reason to keep information in a an article, not delete it. I agree that AmigaOS does not carry as much weight in the context of Power arch as lets say Linux, but the Amiga family must be allowed to be represented. In some respects, Amiga OS and MorphOS is more Power Arch than most of the OS's that listed, since they are pure Power Arch operating systems. Power is the only platform they live on. Most of the still listed operating systems are not pure Power and most of I have never even heard of, and I try too keep up with this article. But ignorance is why we should keep things, so I check to see if they indeed are supported on Power, currently or historically, and if they are, that's a keep. -- Henriok (talk) 16:38, 29 June 2014 (UTC)
- There seems to be some confusion here regarding the Amiga stuff. There were not any links to any Amiga systems nor were any of the entries referring to the Amiga operating system that shipped with Amiga computers. Indeed, the Amiga platform went under before it made the jump off of 68k, though there was a range of third party PPC accelerators which became very popular. 71.200.57.8 (talk)
- So, we're not talking about the Amiga here, we're talking about two "Amiga like" operating systems, one of which is x86 centric with an alpha level PPC port. And one official spiritual successor OS that isn't actually compatible with the majority of Amiga software, is basically a rewrite and is only allowed to call itself "AmigaOS" as long as it doesn't increase the major of its version number and it only runs on computers that are not allowed to be called Amiga. Ironically the unofficial "Amiga like" OS that is PPC centric, is more compatible with actual Amiga software and has better hardware support for the devices it's capable of running.
- I wholly agree with the notion that Amiga had a profound an notable impact on the commuting landscape, but the Amiga was never a "Power Architecture" system, and the new generation of Amiga related OS's and devices were never successful, they are virtually unknown, and have had zero impact on industry outside of a small handful of enthusiasts. 71.200.57.8 (talk)
- " >> No, ignorance is a reason to keep information in a an article, not delete it. " Whether something stays or goes should depend on how it contributes to the purpose of the article. A list of everything that ever ran on a PPC chip would be better suited for another page. If we had such a page (we probably do) we could list every single OS variant and every single revision of every board that's relevant to that space and it would be 100% on topic and relevant. 71.200.57.8 (talk) — Preceding undated comment added 23:19, 29 June 2014 (UTC)
- Nobody said that the Amiga systems were of Power Architecture. There are the Amiga clones from this article which despite "nobody" knowing about them, they are still very encyclopedic. Their developers (perhaps "hobbyists") made complete systems which as far as I am aware of, The AmigaOne and Sam computers are the only PowerPC computers still being made ever since Apple stopped selling PowerPC. The systems are relevant to this article. They have also were the topic of news articles. In Correct (talk) 23:59, 29 June 2014 (UTC)
- That a news article exists referencing Amiga OS 4 or MorphOS establishes that it's notable enough to possibly be included in wikipedia (because it's citable), but that doesn't mean that it's notable in all articles. I'll concede that it's notable enough here because that's the consensus. My last edit to the article reflects that. I think that it needs to be consolidated. Before the edits, there were around 8 lines referencing Amiga related things under implementations. I think that would lead readers to conclude that Amiga related operating systems and hardware have a much larger influence, and are more relevant to this topic than they really are. The only thing that I left out was a ref to the Sam systems, and that was because one of those boards is in an AmigaOne package, and I couldn't find a page that covers both. Maybe that could be another mini-project. The Acube page is anemic and could be a good entry point for info on both the 460 and 440 for other articles to point to. 71.200.57.8 (talk) 04:47, 30 June 2014 (UTC)
- Nobody said that the Amiga systems were of Power Architecture. There are the Amiga clones from this article which despite "nobody" knowing about them, they are still very encyclopedic. Their developers (perhaps "hobbyists") made complete systems which as far as I am aware of, The AmigaOne and Sam computers are the only PowerPC computers still being made ever since Apple stopped selling PowerPC. The systems are relevant to this article. They have also were the topic of news articles. In Correct (talk) 23:59, 29 June 2014 (UTC)
- AROS is certainly not exclusive to Power architecture. Development of AROS on any Power hardware is non-existent. (And all other platforms too). Amiga OS 4 is a victim of an embarrassing trademark problem that you mentioned. That fact shouldn't keep Amiga OS 4 out of this article, although I don't know what AmigaOS 4 has to offer Power Architecture. In Correct (talk) 23:59, 29 June 2014 (UTC)
- AROS recent nightlies are available for PPC, but it's linux hosted. PPC native builds exist for some of the systems, but they are old as you mention. If the consensus is that Amiga OS 4 and Morphos belong in the implementation entry of this article, then I'd wager that AROS has to be on there too. I'd like to point out that there is a git mirror of the AROS svn and it's showing significant activity. ~31 commits in the past month involving around half a dozen developers. It's still enjoying active development. — Preceding unsigned comment added by 71.200.57.8 (talk) 04:25, 30 June 2014 (UTC)
- I'll have to take back what I said earlier about PPC native AROS builds. There's a recent one for the SAM440 dated 2014-06-13. I checked out the ISO and it is indeed native. 71.200.57.8 (talk) 06:48, 30 June 2014 (UTC)
- So what about all these other operating systems in this article? I certainly have not heard of them and I am not sure who, if anybody, uses them. Is it only the Amiga subject you are attacking? In Correct (talk) 23:59, 29 June 2014 (UTC)
- An edit dispute isn't an attack, that's why we're having this polite discussion about the issue in this talk page. I'm familiar with nearly all the items mentioned under the OS header and I'd safely propose that there is an order of magnitude more people who have used those systems than have ever heard about these three Amiga family OSs combined. 71.200.57.8 (talk) 06:48, 30 June 2014 (UTC)
- So what about all these other operating systems in this article? I certainly have not heard of them and I am not sure who, if anybody, uses them. Is it only the Amiga subject you are attacking? In Correct (talk) 23:59, 29 June 2014 (UTC)
Which brings me to another point.. time. Should we remove BeOS? At which point should we remove OSX from this article? There certainly no active development on Power of either at this point, and in 10 or 20 years time, there might not be people left to remember that those operating systems was once all the rage on Power, or even remember Power Arch itself for that matter. Should we somewhere in the future delete facts from articles and articles themselves that at the time doesn't seem to be notable? What do we know about AROS's, AmigaOS's and MorphOS's future? They might be the future BeOS or OpenStep that is or is not picked up like a phoenix to be granted a second and spectacular life, and wouldn't it bee a great idea to have them included so that there is a historical context in which people can hang their new interest? Is it really a good idea to have articles or portions of them removed just because they did seem to be notable at an earlier point but isn't now? Much of my work on Wikipedia is to gather what historical information that's left to be found and collect them here so they can live on forever, for future archeologists and enthusiasts to be found. Is the PowerPC 620 notable today? No. Was it in the middle of the 1990's? Yes, most certainly. Are AmigaOS and MorphOS particularly notable for Power today? They are actively developed operating systems on a hardware plattform that's actively developed, marketed, used and sold, so yes. -- Henriok (talk) 16:38, 29 June 2014 (UTC)
- I grouped all the Amigaoid OSs on one line as is done for the BSD OSs. Left the Peg in the list (was never in dispute anyway since it was multi platform). AmigaOne line gets one entry, if folks want to see the line up, they can click on it's informative article. AmigaOne line references Amiga OS 4. My last edit leaves a reference to all three "new generation" Amiga platforms and references the hardware for the two commercial OSs. In my opinion, all the bases are covered now, and in a way that doesn't give undue weight. Thoughts? 71.200.57.8 (talk) 02:07, 30 June 2014 (UTC)
- I really miss MorphOS. That's an operating system that exists, is under active development, have certified hardware, and an important history. Otherwise I'm happy. -- Henriok (talk) 13:44, 30 June 2014 (UTC)
An article does not live in the current, it lives in the entire span of the subject's life, and after the subject's dead, then it'll live in in a historical context. Who knows what interesting and important facts or people in history that's been forgotten just because someone didn't have the time or resources to record them? What a loss! We have the time and the resources, so let's not cut things just because in might seem insignificant at the moment. And if it still seem insignificant, make it significant, don't erase it so that ignorance can prevail. -- Henriok (talk) 16:38, 29 June 2014 (UTC)
"List Of" Articles
[edit]So far I have not found any articles listing Power Architecture hardware and operating systems. If they exist, they must be linked to this Power Architecture article. In Correct (talk) 00:07, 30 June 2014 (UTC)
- How do you do the link? For instance, http://en.wikipedia.org/wiki/List_of_single-board_computers on wikipedia has an entry for the PPC "P-Cubed" single board computer. 71.200.57.8 (talk) 02:31, 30 June 2014 (UTC)
- All you have to do is surround the title of the article with brackets. [[ ]] In Correct (talk) 03:21, 30 June 2014 (UTC)
I have written the IBM POWER microprocessors and PowerPC-based game consoles articles which are list like. They are linked to this article via the category mechanism embedded in the Power Architecture template. -- Henriok (talk) 13:50, 30 June 2014 (UTC)
OS/2
[edit]I never heard about OS/2 running on POWER CPUs....OS/2 uses the 4 privilege levels of the protected mode of IA-32 - and this is not available on POWER. --93.220.218.110 (talk) 17:06, 17 December 2015 (UTC)
- Workplace OS, which had an OS/2 personality and ran on both x86 and PPC, didn't use them; perhaps it ran the non-ring-0 and non-ring-3 stuff as server processes. Guy Harris (talk) 20:30, 17 December 2015 (UTC)
External links modified
[edit]Hello fellow Wikipedians,
I have just added archive links to one external link on Power Architecture. Please take a moment to review my edit. If necessary, add {{cbignore}}
after the link to keep me from modifying it. Alternatively, you can add {{nobots|deny=InternetArchiveBot}}
to keep me off the page altogether. I made the following changes:
- Added archive http://web.archive.org/web/20110727184746/http://www.power.org/resources/downloads/PowerISA_V2.06_PUBLIC.pdf to http://www.power.org/resources/downloads/PowerISA_V2.06_PUBLIC.pdf
When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at {{Sourcecheck}}
).
An editor has reviewed this edit and fixed any errors that were found.
- 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.—cyberbot IITalk to my owner:Online 04:42, 1 March 2016 (UTC)
- That document appears to have disappeared down the memory hole at power.org - and the Wayback Machine is collaborating with them, by redirecting the 2009 archived URL to that 2011 archived URL, which doesn't work. I just nuked the reference entirely. Guy Harris (talk) 07:06, 1 March 2016 (UTC)
"More details and documentation..."
[edit]Currently the articles summary has a line saying More details and documentation on the Power Architecture can be found on the IBM Portal for OpenPOWER. with the latter part linking to the official website. To me this sounds like a teaser with an action for advertisement and that's nothing I'd expect to have in an article. Such information should go into the Eternal Links section of maybe into the site box if it really can be considered an official website of the architecture (I am not sure about how much Power and OpenPOWER relate). Otherwise the summary feels like an (informative) ad, especially because the expected encyclopedic details are supposed to be reached by scrolling down rather than going to the company website.--Athaba (talk) 10:56, 12 November 2017 (UTC)
Power Architecture on Wikipedia
[edit]There's a discussion at Wikipedia talk:WikiProject Computing#Power Architecture on Wikipedia about this article. 99Electrons (talk) 00:10, 22 February 2019 (UTC)