Talk:Apple A4/Archive 1

Archive 1

Citation needed

Citation needed on it being an arm processor. It seems that way, given apple has stated iPhone compatibility, but apple has also emulated / JITed older platforms. 70.79.18.89 (talk) 20:03, 27 January 2010 (UTC)

So what? iPhone apps run ob iPad with better performance than on iPhone. If iPad emulated iPhone's ARM CPU, the performance would be significantly lower, not higher. --KAMiKAZOW (talk) 16:24, 28 January 2010 (UTC)

It does not seem to be an ARM compatible processor. The iPad SDK specifically mentions building universal binaries so that the binaries can be run on the iPad as well as the iPhone and the iPod touch (which have ARM processors). 94.222.15.238 (talk) 22:34, 27 January 2010 (UTC)

The Simulator probably uses universal binaries to run on x86 and ARM. This way it doesn't have to emulate an ARM processor when you debug your applications on a Mac. So this is no proof. --83.52.39.81 (talk) 23:41, 27 January 2010 (UTC)

Agree. It's probably based on P.A. Semiconductor's PWRficient processor line, which uses its own Power Architecture, not ARM. Knellotron (talk) 22:45, 27 January 2010 (UTC)

ARM also doesn't list Apple as a licencee: http://www.arm.com/products/licensing/licencees.html. Highly likely to be a PowerPC based CPU. —Preceding unsigned comment added by 62.70.27.104 (talk) 08:26, 28 January 2010 (UTC)

Those are only public licensees. Apple doesn't need to be disclosed as licensee. --KAMiKAZOW (talk) 16:24, 28 January 2010 (UTC)

There is an ARM licensee that has not been disclosed publicly. Everybody assumes it is Apple. ARM employees are giving strong hints. But because of NDAs nobody can confirm this officially. --83.52.39.81 (talk) 20:43, 28 January 2010 (UTC)

One can use the universal binaries to build optimized versions for iPod touch G1 and G2 and iPhone G1 and 3G that used a ARM11 based CPUs, while G3 and iPhone 3GS uses ARM7 based. It would be extreme unlikely and quite unnecessary to use Power Architecure processors in a device like this. At these slow speeds ARM shines a magnitude above every other ISA in respect to performance/watt and size. -- Henriok (talk) 19:31, 28 January 2010 (UTC)

PowerVR core

Apple has been a licensee from Imagination Technologies a long time. I don't see how citation 2 pdf gives any proof that it is a powerVR core though it probably is! A better citation is needed! —Preceding unsigned comment added by 213.63.151.102 (talk) 23:36, 27 January 2010 (UTC)

No, *Samsung* has been a licensee of Imagination Technologies for a long time and it's Samsung processors which are in the iPhone & iPod Touch. The SoC in the iPad is developed by Apple directly (not using Samsung), which is why Apple being a licencee is significant! —Preceding unsigned comment added by 62.70.27.104 (talk) 07:30, 28 January 2010 (UTC)

No. Not developed by Apple. Licensed from ARM. —Preceding unsigned comment added by 173.206.38.28 (talk • contribs) 05:57, 29 January 2010

http://www.brightsideofnews.com/news/2010/1/27/apple-a4-soc-unveiled---its-an-arm-cpu-and-the-gpu!.aspx - This reference says the A4 uses the Mali-55. This is highly unlikely as the Mali-55 is an extremely low end part and doesn't have the power to driver the high resolution of the iPad (compared to the feature phones it's designed for). —Preceding unsigned comment added by 84.215.49.26 (talk) 15:30, 28 January 2010 (UTC)

http://www.blogcdn.com/www.engadget.com/media/2010/01/a4-apple-chip-top-1.jpg picture of the A4 —Preceding unsigned comment added by 213.63.151.102 (talk) 00:21, 28 January 2010 (UTC)

The ARM Mali 50 or 55 does not support OpenGL ES 2.0, only 1.1 - the iPhone 3GS already has OpenGL ES 2.0 and it's highly doubtful that Apple would step back on that. -- http://www.arm.com/news/16536.html —Preceding unsigned comment added by 149.169.73.65 (talk • contribs) 19:21, 28 January 2010

If its indeed a Mail GPU, then it is at least a Mali-200 or possibly a Mali-400. If it wasn't for the reference stating that is is a Mali GPU, then I would have ben certain that they'd use a high-end PowerVR core like in the iPhones and iPod touches. Especially since Apple have invested capital in Imagination Technologies, owns 3.5% of the company and has one member on its board. -- Henriok (talk) 19:38, 28 January 2010 (UTC)

Could not agree more. The mali 50 isn't even used in ref boards. mali 200 or 400, if not a PowerVR core. —Preceding unsigned comment added by 213.63.151.102 (talk • contribs) 19:34, 28 January 2010

Speculation

This article consist of a tremendous amount of speculation. It's really just the first line that adheres to Wikipedias rules for original research and referenced information. That said.. at this point we must use this kind of speculation while actual data is scarce. Can we put a box up where we acknowledge that most of the article is essentially unsubstantiated? -- Henriok (talk) 08:52, 28 January 2010 (UTC)

 Done. Mushroom (Talk) 19:37, 28 January 2010 (UTC)

I'm with Mushroom on this. I didn't like all that speculation anyways, and I'll keep reverting this article back to reliably sourced, purely confirmed information. -- Henriok (talk) 08:30, 29 January 2010 (UTC)

ARM core speculation

This New York Times article claims it's an ARM CPU:

"For example, Apple’s coming iPad tablet computer will run on an ARM chip."

No citation, though... Tarcieri (talk) 23:15, 22 February 2010 (UTC)

While I think we can be reasonably sure it's an ARM core, the article cited ([1]) says that it's in the same family as Snapdragon and Tegra which are ranges that include a variety of cores including Cortex-A8 (in use in the current generation of iPhone SoCs) and Cortex-A9. I'd therefore class articles like that as speculative and thus not something you can base a call on which particular ARM core variant is being used. Therefore I've removed any specific ARM core mention from the article. Bjrice (talk) 09:52, 29 January 2010 (UTC)

Crunchgear cites Apple Insider, which in turn cites the Bright Side of News blog, which is the only real source I've seen for these claims. Apart from being an unreliable source, that blog does not even claim to have proof that the Apple A4 is an ARM chip. It just states that the ARM CEO revealed to them that Apple is an ARM licensee, which may or may not be relevant when it comes to the Apple A4. There is not a single statement from someone at ARM or Apple that confirms this theory, it is pure speculation, but since it sounds quite likely it has been cited as a fact by many other (unreliable) blogs. Personally I think the theory holds its ground, however there is not enough information right now to keep those statements in the article. So, concluding: well done! Mushroom (Talk) 10:21, 29 January 2010 (UTC)

Macworld is reporting that A4 is ARM as well, and doing so definitively, but there's no indication of where they found that out: [2]. -- Hawaiian717 (talk) 17:25, 29 January 2010 (UTC)

Interesting. In that second paragraph it links to the same Bright Side of News blog post I mentioned above, so apparently that's the source. It's odd how no one actually bothers to verify their sources these days, and I didn't expect that from Macworld. A commenter pointed that out:

BTW, the BSN article linked to is speculative bull, pure link bait. No one knows what's in the A4, except that it must be an ARM architecture to allow running iPhone apps.

That's exactly my point. It's obvious that it's ARM-based, but we have no reliable sources stating it so we can't put it in the article, and we can also make no assumptions on the specific kind of ARM core that is in the A4. Mushroom (Talk) 17:45, 29 January 2010 (UTC)

I just want to remind anyone reading this, that even if it's obvious that A4 is ARM based, it would still be Original Research to reach such conclusions in the eyes of Wikipedia. -- Henriok (talk) 21:06, 29 January 2010 (UTC)

True. I have added a warning for editors who might not have looked at the talk page. Mushroom (Talk) 21:49, 29 January 2010 (UTC)

Finally, retractions begin: [3] Mushroom (Talk) 23:02, 29 January 2010 (UTC)

PC World is also reporting that it's an ARM CPU, and doesn't link to the BSoN article, although they don't link to any other sources either: [4] Tarcieri (talk) 17:29, 2 February 2010 (UTC)

PC World and Macworld reprint each others articles on their sites. In this case, PC World is rerunning the same Macworld article. Even states that it's from Macworld right in the byline. -- Hawaiian717 (talk) 06:12, 3 February 2010 (UTC)

Notability and other concerns...

• I don't believe that the A4 is notable at the moment. I don't see in-depth coverage of this SoC. I see lots of fanboi hype and speculation, but I don't see in-depth coverage of the likes of the Microprocessor Report or scholarly papers. What we have here is one sentence stating that the A4 is a SoC that contains a CPU and GPU. Can't this be covered in the iPad article?

It's notability can't be measured by the presence of in depth coverage of the device. This is Apple we talk about, we might not get any official info at all, just like the Broadway (microprocessor). But there's ample interest in this device and has been for some time. I think it would be a great feat for Wikipedia to remain a sane source for information about it, even if much info is scarce. The scarcity of data does not make the article less important, or the topic less notable, I't actually more of the opposite.-- Henriok (talk) 12:18, 30 January 2010 (UTC)

• The "Similar devices" section appears to consist entirely of original research as no citations are given.

The data is on the respective device page, we don't have to cite references here for that. Must we use references to back up similarity of to other topics? While I might agree with you that a lay person would have no idea about the similarities to other devices, a person with rudimentary domain knowledge would have no trouble at all to pick out similar devices. Reading the linked articles on this page would probably suffice to claim similarity, even if it would take some reading. But that's what the links are there for, no? -- Henriok (talk) 12:18, 30 January 2010 (UTC)

I should have been more specific, sorry. What concerns me about that section is not the facts about the processors, its the claim that they are similar. Stating that all SoCs with a CPU and GPU are "similar" is like saying that processors with n pipeline(s) are all similar. That is just one aspect. I don't believe that I have seen such comparisons elsewhere. It could be revised to say something such as "other devices with integrated CPU and GPU for portable applications are..." Of more concern is the claim that all devices listed have similar performance, amongst other things. Not much is known about the A4, so how can it be claimed that the performance of those devices are similar to the A4? Rilak (talk) 08:41, 31 January 2010 (UTC)

I agree about the performace statement, and I have removed it. -- Henriok (talk) 14:02, 31 January 2010 (UTC)

• I believe that the importance rating given to this article is inappropriate. I am interested in knowing just what led to this article being rated as of mid-importance? There are elementary topics in computer engineering related to CPUs that are ignored and rated as of no importance. Funny that a proprietary SoC that we know nothing about, that is only used in one unavailable product from one company can be considered to be more important than basic theory. While this concern does not affect the article's content, it highlights, in my opinion, Wikipedia's occasional inability to judge things against the big picture and susceptibility to media hype.

I agree that it's not that important in the general computing scope, but it's without a doubt of "Mid" importance i the Apple scope. -- Henriok (talk) 12:18, 30 January 2010 (UTC)

To address these issues, I suggest that this article be redirected to the iPad article and its talk page blanked. If consensus is against redirecting, then I suggest the similar devices section be removed, and importance ratings revised to no importance. If in the future the A4 does receive in-depth coverage, the redirect can be replaced by an article.

Any comments? Do you agree or disagree? Rilak (talk) 05:21, 30 January 2010 (UTC)

We don't generally blank talk pages when redirecting pages. It is likely that as more information is released that a re-split would happen. That said, I think a merge would be a good idea until more material is available. Chris Cunningham (not at work) - talk 08:40, 30 January 2010 (UTC)

Yes, I know. I meant to say that we should get rid of the project tags if this article gets redirected. Rilak (talk) 07:37, 31 January 2010 (UTC)

I obviously don't think It's a good idea. Many other devices have separate articles for their CPU, like Xenon of the Xbox 360, Broadway of the Wii, EE of the PS2 and there are a lot of articles of similar devices (which I have listed in the article). This article is of the same importance to Tegra, Snapdragon, OMAP and Cortex and those have separate articles. -- Henriok (talk) 12:18, 30 January 2010 (UTC)

I am not saying that proprietary ICs don't deserve articles, I am saying that the A4 has little coverage and thus questionable notability. I don't see what game console processors have to do with the A4. Are you saying that since they have not have their notability challenged, that the A4 does not deserve it either? If so, I think that it should be said that just because an IC is proprietary, and used in only one device, it does not equate to inherit notability. For example, the Emotion Engine has significant coverage. It has been described at Hot Chips and the ISSCC, and there are three conference papers in ISSCC proceedings about Emotion Engine alone. The Microprocessor Report has detailed coverage of it, and there is an article in the IEEE Micro about Emotion Engine programming. Why shouldn't the A4 be judged in a similar way? Rilak (talk) 08:04, 31 January 2010 (UTC)

This goes to notability to the issue/article. I did a Google search for "Apple A4" ≈ 200k pages and "Apple A4" vs Tegra ≈ 15k pages, vs Snapdragon ≈ 9k pages, vs Cortex ≈ 19k pages, vs. Atom ≈ 20k pages. This matter is surely discussed. And if we agree that the lack of hard info about the device isn't a concern, there is no lack of pages finding the A4 notable, and not comparisons to other similar devices either. I hope Apple will discuss the A4 at ISSCC, but since Apple's being Apple, don't hold your breath. It doesn't make the device any less notable either. -- Henriok (talk) 14:10, 31 January 2010 (UTC)

I don't think that equates to notability. For all we know, its mostly people repeating iPad specifications, or adding them to their forum signatures. It should be demonstrated that pages containing "Apple A4" is about the A4, not iPad specifications. I don't think discussion, speculation, or anything associated with hype equates to notability. And given that we still know nothing about the A4 other than it is a SoC with integrated CPU and GPU, I don't see how any valid comparisons to similar devices can be made, other than educated guesses by sufficiently qualified people.

I also wish to request further clarification regarding your claims about the A4's notability compared to other devices. I did a search that required a page contain both "ARM" and "Cortex" and I got 1,470,000 results. Considering that both terms are not exclusively used to refer to the processor, I did a search for "ARM Cortex" wich returned 1,780,000 results. Searching for "NVIDIA Tegra" returns the same number of results. Searching for "Apple A4" returned 213,000 results. It would appear from this, that my assertion that the A4 is not notable, and is not as notable as the ARM implementations, is confirmed. Rilak (talk) 02:29, 1 February 2010 (UTC)

I get >240k hits now so it increases with 30k articles in less than half a day. I don't have to make the case compared to other devices, just notability in itself, "In general, notability is measured by whether the topic has received significant coverage in reliable secondary sources". These are not Apple centric sources, focusing on the chip and not iPad. This is significant coverage to me:

• iPad powered by custom 1GHz Apple A4 chip - Engadget
• Apple's A4: ARM CPU, GPU - OSNews
• iPad's Identity Crisis and Apple's A4 CPU Showstopper - HotHardware
• Apple’s A4 processor is in ARM family tree with Snapdragon, Tegra - Crunchgear
• Apple iPad First Device to Use "Apple A4" Processor - Gizmodo
• Inside the iPad: Apple's new 'A4' chip - CNet
• Apple's A4 IPad Chip Could Find a Home in IPhones - PCWorld
• Apple Launches Its Own CPU, the 1GHz A4 – Tom's Hardware
• Apple A4 is not a CPU, it's a 'system-on-a-chip' - ZDNet
• Apple's A4 iPad chip could find a home in iPhones - InfoWorld
• The Mysterious Apple A4 Chip - Where's MSFT's and GOOG's Chip? - Silicon Valley Watcher
• Eagerly awaiting info on Apple's A4 chip for the iPad and how its PA Semi buy fits in ... - EDN

The EDN one is a blog and it is really someone saying, "I am eager to hear more about the A4". The difference between this and someone on a forum saying the same thing is that this is hosted by EDN. Rilak (talk) 03:56, 2 February 2010 (UTC)

The InfoWorld article doesn't add anything we don't know and is just a bunch of analysts speculating on things. Rilak (talk) 03:59, 2 February 2010 (UTC)

The CNet article, in typical CNet fashion, is short, lacks any detail, is mostly speculation, and digresses to irrelevant topics. Rilak (talk) 04:01, 2 February 2010 (UTC)

And its a blog. Rilak (talk) 04:02, 2 February 2010 (UTC)

The Silicon Valley Watcher is mostly just a recap of recent history, it isn't about the A4.

The rest of the publications are most likely the same as the four I summarized above. I won't bother reading them since the trend is "let's speculate". Does this constitute as "significant coverage"? If it does, then should we not know much more about the A4? Even if we do somehow prove notability, why do we even need an article that basically consists of one sentence: "The A4 is a SoC containing a CPU and GPU." Why can't we merge this into the iPad article? Presently, the A4 is used only in the iPad. It makes sense to put what we know about it in that article so readers won't have to waste time visiting an article whose contents are more like a footnote. Rilak (talk) 04:11, 2 February 2010 (UTC)

I was trying to make a case for notability, I you don't seem to contend that. So the article stays. We know that info is scarce, so we won't find them in the sources I provided, and that's why Wikipedia has a stub flag. So the article stays. Of course we should include the info we got into the iPad article, but what we know already is included there. And, since we talked last time, the hits on Google has risen by another 30k, which further strengthens the case for notability. I'm eager to know more too, but lack of info is not a criteria for deletion, only if there's NO info, but there is. -- Henriok (talk) 09:11, 2 February 2010 (UTC)

I was trying to prove that the sources you have provided failed to indicate any notability but the lack of in-depth coverage. Your argument, if I am not mistaken, is that there will be in-depth coverage in the future. What if there isn't?

AFAIK, the stub tag is not for articles whose topic may be notable in the future. Its for articles lacking content. Articles may lack content because it has received little attention, or because they are not notable.

Additionally I am not trying to get this article deleted. I am trying to advocate that Wikipedia does not need separate articles for every single little thing when we can cover a group of very closely related topics in a single article. Why? So that our readers won't have to visit multiple articles, and so they won't have to waste time with disappointing articles such as this. You have made no argument as to why we should have an article whose contents amount to a footnote. Lack of notability does not mean that content needs to get deleted, so what has deletion got to do with this?

As for the increase in Google results, I don't believe that it strengthens the case for the A4's notability. How do you know that the increase in in-depth coverage and not mere mentions or just people discussing the iPad on forums?

The question that we all need to ask is what benefit will a reader get from visiting this article? My answer is that the readers will not benefit. The iPad article already covers the A4, and the A4 info at present is only useful in the context of the iPad. I've said it before, this article should redirect to iPad for the reasons that I have previously mentioned in previous posts, and in this one.

Additionally, it should be obvious, but this discussion has reached a sort of stalemate. Therefore, I'll make a new proposal. If there is no additional substantial information about the A4 after ISSCC (after February) from any reliable or sort-of-reliable publication, we redirect this article to the iPad. Rilak (talk) 04:44, 3 February 2010 (UTC)

Lack of info does not equate lack of notability. This article lacks data, nothing else. And lack of data is not a criteria for merging nor deleting, only if there's NO data, but there is. Anyhoo.. I think it would be fair to at least wait until the release of the iPad. At that point, people will have the chase to crack it open, look inside, poke and probe it, NDAs are lifted and such. -- Henriok (talk) 09:27, 3 February 2010 (UTC)

OK. I find this to be very silly. Firstly, lack of content is a direct consequence of either laziness on the part of WP editors or the lack of significant coverage, thus lack of notability. Since this article has the attention of multiple editors, has received ~100 edits, and it still contains next to nothing. Articles containing three short sentences are candidates for merging. This entire discussion has been a waste of time, so this will be my last involvement in this discussion. Rilak (talk) 03:35, 4 February 2010 (UTC)

You miss a third reason:This article isn't anemic due to laziness nor lack of coverage, but lack of information from Apple. Apple (the only reliable source at this point) has said next to noting, but what they have provided, we have included. The info that Apple has provided has been covered by A LOT of sources (12 provided above). Articles containing three sentances might be considered candidates for merging unless certain conditions apply. I cite two of the reasons given at Wikipedia:Merging:

"{Merge} If a page is very short and is unlikely to be expanded within a reasonable amount of time." Reasonable ammount of time? Yes, Apple and their partners will disclose more info as iPad starts shipping, NDAs are lifted and so forth. We know when this is likely to be: late March, early April 2010. It's reasonable to wait at least until then.

"{Don't merge if} The topics are discrete subjects and deserve their own articles even though they may be short." Discrete topics? iPad and Apple A4 are clearly discrete topics. There are ample examples where components of gadgets have articles of their own even if the component is unique to that gadget: GameCube/Gekko, Wii/Broadway, PS2/EE, PSP/UMD, System z10/z10 and A-10/GAU-8. It's a fairly good assumption that A4 is the first of a family of processors for several products from Apple in the future, so the articles will diverge even more.

-- Henriok (talk) 13:53, 4 February 2010 (UTC)

I may have said that I will no longer continue this discussion, but your response requires a response. A table may be constructed from a panel of wood, a frame, and four columns of wood for the legs. Do we have a separate article for each component? No. The guideline does not apply. A short article is ~1000 words. And as I have said before, this article's useful content can be summed up in a footnote. That said, I will no longer continue this discussion. And I would like to request that you don't reply to this comment. Rilak (talk) 06:23, 5 February 2010 (UTC)

As more details emerge, this article will get richer content. I don't think there is a point in merging these pages, as it seems pretty certain they would be going to be split again in the future. I think we should just:

• Remove the "Similar devices" section
• Tag the article as a stub
• I'll let the importance rating of this topic up to your judgment. It is just a tool to drive ongoing work on wikipedia - as such, the "importance rating" is not really important by itself.

Cochonfou (talk) 12:00, 30 January 2010 (UTC)

It's highly likely that Apple will use this part or variants of it in future devices, so while it might make sense to merge it with the iPad article in the short term, in the longer term it probably deserves its own spot. Bjrice (talk) 02:09, 31 January 2010 (UTC)

Not that I necessarily disagree, but the latter is certainly conjecturing. --Cybercobra (talk) 03:16, 31 January 2010 (UTC)

The Crystal Ball rule only applies to content of articles. It would be absurd if it applied to us discussing strategies regaring articles. -- Henriok (talk) 03:36, 31 January 2010 (UTC)

I'm just saying we're anticipating notability. Again, I'm not campaigning for actually doing something about it. --Cybercobra (talk) 06:44, 31 January 2010 (UTC)

Similar devices section

Half of this article is dedicated to summarizing similar chips. I know there's not much known about the A4, but I don't think inflating this article by discussing similar chips is necessary. If it's going to stay, this section belongs either on System-on-a-chip or something like Comparison of system-on-a-chip platforms. -- Hawaiian717 (talk) 22:19, 1 February 2010 (UTC)

I agree, there is no need for that section here. Mushroom (Talk) 00:53, 2 February 2010 (UTC)

Remove PA Semi

Can we remove all references to PA Semi also. According to this, the A4 was designed by Apple's in-house VLSI team, the same team responsible for the north bridge chipset of the PowerPC Macs that they were making for YEARS. Apple is not new to the chips business, they were just not a supplier of their custom silicon to other companies.

This is the first source stating that it's not the PA Semi group doing this, but there's other sources stating the contrary. What are we going to do about it? We could include both sides, or no sides. I opt for the former, and that's the proposed way to write about opposing views. Something like "There's a doubts about who's designed the A4 inside Apple. It's either the engineers from former P.A. Semi, or the engineers that's been designing chips inside Apple for years." and sourcing both statements. I hope that Apple have formed a unified group though. Little point in having separate design departments. –– Henriok (talk) 16:54, 7 February 2010 (UTC)

Oh, I'm not so sure about that. It can go either way for me. I think they have a single design department anyway too, but working on several projects in the same time. Maybe the new guys are working on the SoC of the next iPhone. Imagine CPU, I/O, memory controller, graphics AND wifi and 3G networking all in single die. Wow!

That's not really a contradiction - I doubt Apple put the PA Semi guys in a department all of their own, where they fiddle their thumb since the acquirement. Lars T. (talk) 00:11, 8 February 2010 (UTC)

Does Apple have an ARM Architecture license, or just an ARM core license?

If Apple, like Marvell has a full architectural license from ARM , instead of just licensing the core, as most ARM implementers do, then they could in theory design a brand new implementation of a 4-core ARM Cortex-A9 MPCore that is both faster, and more energy efficient, than the default ARM Cortex-A9 MPCore implementations as licensed from ARM. If they did, and their implementation is considerably better, it could really be a "game changing" fact. Mahjongg (talk) 00:44, 9 February 2010 (UTC)

Perhaps. We don't know. No one is telling. ARM's CEO said that they have a large secret architecture licensee, and allegedly hinted at Apple.[5] I would guess that they have an architecture license, but we don't know for sure. -- Henriok (talk) 10:13, 9 February 2010 (UTC)

Interesting link! I see that Apple brought in Dan Dobberpuhl, the architect of the StrongARM processor. I have little doubt that Apple is indeed the mystery "architecture licensee", and that the A4 is a 4-core ARM multicore processor, but time will tell. Mahjongg (talk) 12:31, 9 February 2010 (UTC)

I highly doubt that a 4-core A9 would give them the battery life or the price point they achieved. I'd be far more inclined the believe that it's a souped-up A8. However I also believe that Apple is an architecture licensee and may indeed bring their own stuff to market in the future but developing a CPU core is not a quick task. This is all my own opinion though so not really something that can be put on the page. =) Bjrice (talk) 23:42, 10 February 2010 (UTC)

A cortex-A8 is what is in the iPod touch, I think they have something with a bit more oomph in the iPad, the "4" in A4 could stand for the four cores in a ARM Cortex-A9 MPCore, if they can power down cores independently they could use one core and hardware video decoding to achieve 10 hours of video playback (which is what they are actually claiming). They (allegedly) signed the license with ARM in Q2 of 2008, so its possible. They also have been working on low power processor technology for years before that. But I agree its all just speculation at this point. Mahjongg (talk) 00:55, 11 February 2010 (UTC)

On second thought, the name A4 may have been chosen because of the G4, but in the G4 the four stands for fourth generation, I doubt the A4 has seen three previous generations. Mahjongg (talk) 01:13, 12 February 2010 (UTC)

I think you're reading too much into the name. =) The CPU core also isn't the sole predictor of total system performance, especially in an SoC like this. There are a huge number of things they could've done to increase overall performance such as increasing bus speeds and bandwidth, moving blocks around to allow faster communication, speeding up other components and, shockingly enough, optimising the software. =) Bjrice (talk) 05:20, 13 February 2010 (UTC)

All true, well we will see what emerges when the devices some onto the market. Mahjongg (talk) 19:31, 13 February 2010 (UTC)

Came across this [6] claiming that it cost Apple a cool billion (that is a short scale billion, thus a thousand million. not a million million, for you non American readers) to create the A4. If that has any truth in it then its almost certain that Apple built its own ARM code compatible processor, instead of using any of the existing ARM implementations. That or it's something even more outrageous that can run ARM code. But thats just me speculating. Mahjongg (talk) 23:42, 24 February 2010 (UTC)

It runs ARM compiled apps without recompile

and faster than the iPhone. Then, it is not an emulator, and hence does have an ARM core. If it has an ARM core, it must somehow have been licenced from ARM. Is there not a single person working on this lemma with a technical understanding? Put at least the ARM info back.

It's like knowing that this Toyota drives on gasoline, but refusing to state that it has a combustion engine, since Toyota has not yet confirmed. Thyl Engelhardt213.70.217.172 (talk) —Preceding undated comment added 12:26, 19 February 2010 (UTC).

yeah, well I'm totally convinced it uses the ARM architecture, and fairly certain it is (based on) the latest and greatest Cortex incarnation, and somewhat certain it will be a multi-core, And I'm speculating that its their own implementation, instead the one from ARM Holdings that everybody else (except Marvell) is using, but the problem still remains that unless Apple starts talking, or people get an iPad and start to take it apart and analyze it, there won't be any reliable sources we can quote. Mahjongg (talk) 16:34, 19 February 2010 (UTC)

I am not calling out for any speculations, but there is no need for Apple talking; there is simply no alternative that it would not be ARM compatible. Thyl213.70.217.172 (talk) 16:43, 19 February 2010 (UTC)

I agree, but this is wikipedia, unless there is a "reliable source" that states so we cannot even mention the most obvious facts. Mahjongg (talk) 23:53, 19 February 2010 (UTC)

Every single thing that i posted in the early version will eventually be confirmed. PowerVR chip (check), Samsung is manufacturer (is next check), etc... etc... —Preceding unsigned comment added by 85.243.209.90 (talk) 00:15, 24 February 2010 (UTC)

Oh, I full expect that to happen, it doesn't make a bit of difference though. The moment a reliable source confirms any of these it can be added instantly. I read some news speculating that "the iPad is using ARM, because Apple is Hiring ARM experts",[7] unfortunately the article at Apples site did not mention the iPad [8]. That was a downer... Someday soon, al of these things will be confirmed I agree, I can't wait, I'm very curious myself. Keep on scanning the net for those reliable sources! Mahjongg (talk) 23:27, 24 February 2010 (UTC)

Ah, that [9] didn't take long at all!

On the other hand, from the announcement till a reliable source took the trouble to state the obvious fact that the iPad must somehow have a processor able to execute ARM code very fast, still took a remarkable long time! And its still didn't come from the horses mouth (Apple) what technology actually is used in the A4 . I wonder... why all the reluctancy. Mahjongg (talk) 21:24, 25 February 2010 (UTC)

A reliable source?

Yesterday, Ars Technica ran an article on the iPad processor: »The A4 and the A8: secrets of the iPad's brain« In the article, Jon Stokes writes:

But it turns out that the the A4 is a 1GHz custom SoC with a single Cortex A8 core and a PowerVR SGX GPU. The fact that A4 uses a single A8 core hasn't been made public, but I've heard from multiple sources who are certain for different reasons that this is indeed the case. (I wish I could be more specific, but I can't.) In all, the A4 is quite comparable to the other Cortex A8-based SoCs that are coming onto the market, except that the A4 has even less hardware. The iPad doesn't have much in the way of I/O, so the A4 itself can do away with the I/O that it doesn't need.

…

If they were involved at all in the A4 design, and it's still not 100 percent clear that they were, it's likely that the P.A. Semi team made its biggest contribution to the A4 in the area of dynamic power optimization.

Perhaps this would be useful for the article?

Dan Villiom Podlaski Christiansen (talk) 18:31, 1 March 2010 (UTC)

Sounds very speculative at the time, especially as the claim that the A4 is "based" on the Cortex A8 is very vague, what does "based on" actually mean in this context? Also, its not a claim that is very far from existing expectations that the A4 would be based on one of the latest incarnations of ARM (the cortex family). So it is vague and doesn't add much at this time, except for confirming what little we already know. The little factual information that is stated ("stripped of unneeded interfaces") is especially speculative. It's not needed to remove an interface to keep the power down for example, as "unneeded interfaces" can be powered down, just as well as removing them from the silicon. For example a "camera interface", needs, compared to the CPU or GPU, a minutely small die space area. Actually the most interesting part of the whole article is the claim that the G4 is based on the Cortex-A8 single core incarnation, while it was only a short time ago that a reliable source even mentioned it was an ARM compatible CPU. However, the fact that it likely is an ARM Cotex-A8 based CPU is already incorporated into the article. Mahjongg (talk) 21:53, 1 March 2010 (UTC)

My sources (which I can't link to and hence can't use for any citation) say it's a Cortex-A8, for what it's worth. Bjrice (talk) 23:57, 1 March 2010 (UTC)

Yeah, well that its a Cortex A8 is already in the main (iPad) article. Mahjongg (talk) 00:09, 3 March 2010 (UTC)

Power consumption

Recent vandalism, (changing the CPU to a power-PC with the remark/excuse that a powerPC could "easily be less than 5 Watt") sure made me laugh. Less than 5 watt isn't in any way "frugal" in this context. The power consumption of the CPU (plus GPU) (depending on the tasks performed) is in many cases probably only a something like half a Watt! Remember, the iPad has a 25 Watt/hour battery. That means that the total consumption (according to Apple) of the device even while showing a Video, (probably the most power hungry task) is just 2,5 Watt (25 watt/hour /10 hours = 2.5 Watt ) That is including the backlighting for the LCD and other power using electronics (like the GPU). But clearly the CPU alone can go a lot lower than that, the 140 Hour playing music means that the total power consumption of the CPU can at least be throttled back to just under 0,18 Watt (this must be music playback with the LCD backlighting turned off, as power needed for the backlight alone typically uses something like that amount). Hopefully we will have some dependable power consumption figures soon, so we can see just how frugal the A4 itself is. Mahjongg (talk) 03:50, 23 March 2010 (UTC).

Reliable source?

Is it really acceptable to first remove a reference [10] from a reliable source (Ars technica}, and then after its gone to remove the referenced text with the comment "its all just speculation". It may or may not be speculation, but if its speculation its speculation from a reliable source. The A4 is claimed to be an "single Cortex A8 core and a PowerVR SGX GPU" (In fact the exact wording is "But it turns out to be an single Cortex...", which isn't stated as speculation at all) in this reliable source, so that fact should be in the article, if need be with some reserve, but still.. Mahjongg (talk) 17:16, 25 March 2010 (UTC)

I think Ars and HAnnibal are reliable sources, but as the article is written, it's hearsay at best. _I take his word for it, but I don't thinkWikipedia should. Let's give it a week or two so an actual iPad can be analyzed, and we can be certain what the A4 really is. There will probably be a lot of NDAs lifted by then. Perhaps Hannibal's sources can be more open by then. -- Henriok (talk) 22:04, 25 March 2010 (UTC)

That is not what I meant, I do not deny that is probably hearsay, I claim that wikipedia isn't interested in the "truth", it is only interested in what is said by reliable sources! Go and read Wikipedia:Truth. So if you say but I don't thinkWikipedia should. then you are certainly working against Wikipedia's rules. I agree that in a few weeks all will be clear, but in the mean time WP readers have a right to hear what experts on the subject think about it now. The removal of a perfectly good reference is certainly not done. I will restore it. Mahjongg (talk) 00:30, 26 March 2010 (UTC)

Excellent. I concur with that essay, and I can certainly live with what it says. I had not read it before and interpreted Wikipedias policies in another way. It was most helpful. -- Henriok (talk) 08:40, 26 March 2010 (UTC)

"Consensus" its an Cortext-A8 ?

I found an expert who claims the processor in the A4 might just as well, be a single-core Cortex-A9, and that this would even fit the current facts better that assuming it must be a Cortex-A8 "because it is a single core". I added this fact to the article with this sentence:

Although there are also sources that claim it might be the single core version of the ARM Cortex-A9 
because it would explain the speed doubling compared to the Cortex-A8 in iPhones.

Now user Bjrice has removed this with this edit [11] claiming that "there is consensus it is an A8", without adding new evidence for it. Wat do you think, is there really "consensus" when an expert thinks the known facts are better explained with a single-core cortext-A9. All we really know is that it seems to be a single core ARM processor, so it doesn't seem to be the multi-core Cortex-A9, but the Cortext-A9 also comes in a single core version. [12]. And its higher performance per clock cycle matches the observed performance of the A4 better than the Cortex-A8. I'm going to put back the sentence, and its ref, also because it has other important information. Mahjongg (talk) 11:43, 15 April 2010 (UTC)

I don't think there is consensus. The higher frequency and performance can just as well be thanks to some magic done by folks from PA Semi and Intrinsity. -- Henriok (talk) 14:00, 15 April 2010 (UTC)

I don't mean there might be consensus that its an Cortex-A9, I meant that the person who removed my text did that with the excuse that the "consensus" is that its an ARM Cortex-A8. That "consensus" came about after the conclusion was drawn that the CPU must be a single core, so many people made the automatic assumption that it could NOT be a Cortex-A9, because they assumed that all Cortex-A9 were multicores. But actually that isn't true, as the A9 is not per definition a multicore processor, singlecore versions of it do exist. So someone at PC world argued that the conclusion that the consensus of the A4 containing a Cortex-A8 could be false, [13]. He argued that looking at the facts it was more probably that it actually is a single core version of the Cortex-A9.

"This performance increase over the iPhone 3GS is exactly in line with what I've expected, 
based on a single-core ARM Cortex-A9 CPU running at 1GHz," 

"The speed increases [in the benchmarks] aren't possible with just the clock speed increase of the A4, 
he added, using the 67% boost provided by the iPad's 1GHz A4 to argue that the SoC integrates an 
ARM Cortex-A9, not an A8."

It would be much more a stretch of the imagination that the speed increase is due to "some magic done by folks from PA Semi and Intrinsity" than it would be to accept that its a Cortex-A9. However what could go some way into explaining a speed increase is the 64-bit wide databus, so i'm still not 100% convinced its a Cortex-A9, but I do think the chances that it is Cortex-A9 are better than that its an A8. Only a full analysis of the processor core can give us a definitive verdict. In any case I don't think you can argue that this new info must be suppressed "because there is consensus it MUST be a Cortex A8". Mahjongg (talk) 16:43, 15 April 2010 (UTC)

One person reckons it's an A9 based on merely the fact that they can't explain the performance increase based on what appears to be a complete lack of understanding of CPU design. It's actually far easier to tweak an existing design, say the previous iPhone 3GS part, than it is to implement based on an entire new core architecture. If you do it right, you can get phenomenal speed boosts. Witness what happened when Intel decided to start mucking around with the Pentium Pro architecture again and ended up with Core. There are several other places, primarily the iFixit/Chipworks stuff but also people who have done analysis on the OS and generated apps, that conclude fairly conclusively that it's A8. If we want to go in to other areas, parts with Cortex-A9 are only just starting to ship to designers. There are no devices containing A9s on the market yet. If Apple have been doing design work on the A4 for a significant time in lead-up to this, they won't have been using an A9 simply because it wasn't ready yet. Now I realise that last part is my conjecture and probably counts as original research but since a Google search for "apple a4 cortex a9" still has the debunked "Bright Side Of News" article as the top hit and then the iFixit article, the consensus among people who aren't hand-waving is that it's an A8. Lastly, I have a source who is probably similar to the one Ars Technica has who has given me rock-solid proof that it's an A8. I'm not going to get into an edit war over it, but I feel that just because one person says it and that you want to believe doesn't mean that the page should contain misleading information. Bjrice (talk) 23:01, 15 April 2010 (UTC)

Okay, I start to understand your position on this a little better. Still "one person who appears to have a complete lack of understanding of CPU design"? Thats a harsh verdict for Aaron Vronko, a person who is considered to be a "hardware guru". Still I don't know the guy, and must say I don't have a high regard for PC World. But he does have some good points, the fact that the A4 contains a single core does not -automatically- mean it must be an A8, and a 33% greater increase in performance than what is to be expected from the increase of clock-speed alone isn't simple to do with some "tweaks", and would indeed be explained by the specified 2.50 DMIPS/MHz of the A9 instead of the 2.0 DMIPS/MHz of the A8. Still I wouldn't claim to know for sure who to believe, or that I'm a "CPU expert" myself, so who am I to decide who to believe on this, I do know that Vronko is not "infallible", as (for example) he predicted the iPad would not sport a LCD, but either e-ink or a hybrid display similar to the one in the OLPC XO-1 obviously he was wrong. I did not know (for sure) that nobody has yet released a product with the single core Cortex-A9, but I do know that some have used the multicore of the A9 (for example in the Nvidea Tegra, so how much harder can it be to use the single core version? Is Apple really incapable of this feat? Maybe they got help directly from ARM. We don't know what kind of relation Apple has with ARM, remember that at one time (during the era of the Newton MessagePad) they were more or less directly responsible for the creation of the ARM6 processor, so Apple and ARM go a long way back. But this, I a confess is a bit of conjecture too. In the end, we don't know -for sure- that the A4 contains a Cortex-A8, so at the very least there should be a -little- bit of room for doubts, at least until Apple coughs up the facts, or in another way the facts about what is in the A4 become irrefutable. Mahjongg (talk) 01:03, 16 April 2010 (UTC)

The thing is that an increase of that magnitude doesn't have to come from retooling the CPU core. It can come from a myriad of things including die-shrinks, improvement in bus speeds and a whole host of other factors. Total system performance is dependent on far more variables than just the speed of the CPU. The NVIDIA Tegra 2 (as opposed to Tegra, which is ARM11) is only just starting to appear and as far as I'm aware there are products planned around it but none as yet available. I have solid knowledge that it is an A8, it's just not in a form I can add as a ref to the article which is why I don't want to get into an edit war. If I could link to it, I'd put it up and that'd be that. My understanding is that we will see an A9-based part and at some point (heading off into complete conjecture) we may even see their own core design a la StrongARM/Xscale/etc. I'm fairly sure that right now the A9 pulls too much current and was not easily usable by Apple for the A4 and there was plenty that could be done with the designs they were already using. —Preceding unsigned comment added by Bjrice (talk • contribs) 04:46, 16 April 2010 (UTC)

Fair enough, but in principle, unless you can provide proof (provide a link to a reliable source) its an A8 it is against wikipedia policy to repress information that DOES have a reliable source, see WP:truth.

The threshold for inclusion in Wikipedia is verifiability, not truth

Mahjongg (talk) 12:10, 16 April 2010 (UTC)

I would rate the iFixit/Chipworks sources far above Vronko as sources of verifiability. Bjrice (talk) 11:38, 17 April 2010 (UTC)

A4 based on hummingbird?

According to the latest new information Apple acquired chip-maker intrinsity [14] just before the iPad launch. Interestingly intrinsity in collaboration with Samsung were the first that managed to improve the basic cortex-A8 design so that it could be clocked at 1 GHz. Their 1GHz cortex-A8 processor was dubbed the hummingbird [15]. So speculation now is that the A4 is based on a modified hummingbird cortex-a8 cpu. 16 April Mahjongg.

Okay, I made a few edits to point out that the " performance enhancements to the Cortex-A8 developed by Intrinsity and Samsung" were in fact used to be able the clock the A4 higher than a standard Cortex A8 design, which can only be clocked up to 650MHz. There are only two companies that actually made a 1GHz version of the A8, Qualcomm which spent three years and $300 million developing the "Snapdragon", and Intrinsity, which together with samsung developed their own 1GHz Cortex A8 core they dubbed "Hummingbird". Now the things is, Intrinsity is no stranger to dealings with Apple, they used to design faster versions of the Power PC for Mac's. It seems that in close cooperation between them Samsung and Apple they were intrinsic to the development of a 1GHz version of the Cortex A8 that also had a powerful video-co-processor. It is obvious that intrinsity is so intrinsic to Apple that they simply bought the company, so that the intrinsity engineers are no Apple engineers. Obviously the acquisition of intrinsity happened after the launch of the A4, but that does not mean intrinsity was not involved in the design of the A4. So I edited a few sentences in the article explaining the influence intrinsity has had on the A4, namely making it possible to lock it at 1GHz. However these edits were removed by User:Lars T. as "speculation" .

However, In fact according to this reference [16]

”Actually, there’s no speculation,” Strauss said in the report. 
”It’s only the Intrinsity folks who could have taken it up to a gigahertz. Period.” 

The removal of the text seems to be based on the assumption that intrinsity coudn't have been involved in the design of the A4 because

 the purchase only happened in this April, so obviously the chip can not have been designed by Instrinsic <sic>

Which is obviously false reasoning. But even if it were "purely speculation", I only wrote what reliable sources were saying. Therefore I will revert Lars T's edits. Any comments? Mahjongg (talk) 00:46, 26 April 2010 (UTC)

The comment that the Cortex-A8 design can "only be clocked up to 650MHz" is wrong. For instance: the Texas Instruments OMAP3530 (Cortex-A8 core) is clocked at 720MHz, the Freescale i.MX51 is clocked at 800MHz [17]; and T.I. have announced a 45nm 1GHz OMAP3640 part [18]. Ptoboley (talk) 12:12, 26 April 2010 (UTC)

The source I sited apparently talks about the unmodified IP from ARM, its obvious that all these companies did something to the A8 design to make it possible to clock the chip higher than the default design by ARM. Maybe this must be re-worded to make this point clearer. Mahjongg (talk) 12:29, 26 April 2010 (UTC)

Sorry for chiming in late on this, but I think it needs to be emphasized that this is just speculation. And, no, the AppleInsider article really doesn't count as a reliable source (particularly given their use of the word "likely"). It's a rumor site citing a rumor article in Spectrum. Intrinsity is not the only company to produce high-performance ARM-derived parts. Anyway, the current phrasing of the article, which suggests a level of fact not borne out by your sources or anything else in the real world, is not good. I'm going to go ahead and drop that paragraph until we can figure out some verifiable way to include the actual information. —Roguelazer (talk) 08:53, 28 April 2010 (UTC)

No, its NOT "speculation". Its not at all difficult finding a (more) reliable source for this. Is ars technica reliable enough for you? [19].

Apple confirmed to The New York Times today what Linkedin profile updates have already indicated, 
with Intrinsity's employees naming Apple as their new employer. 
As for what Intrinsity brought to Apple that PA Semi couldn't, the answer is a CPU core.
Samsung and Intrinsity announced the Hummingbird core in the summer of 2009, 
and this is almost certainly the core on which Apple's A4 is based. 

Mahjongg (talk) 09:08, 28 April 2010 (UTC)

I'm not questioning that Apple bought Intrinsity. I'm questioning that Intrinsity designed the A4. Even that article only says "almost certainly". Until somebody from Apple comments (or somebody spends a few million dollars worth of time on an electron microscope and with a VLSI expert), it's still speculation and should not be stated as fact. —Roguelazer (talk) 09:30, 28 April 2010 (UTC)

I do agree that "almost certain" is not the same as "certain", but that is the level of accepted knowledge at this time, backed by a reliable source, so it should be in the article. (see WP:truth) But If you want to re-word it so there is still a smallish chance the experts have it wrong, go ahead. Mahjongg (talk) 09:35, 28 April 2010 (UTC)

I've re-worded to more clearly indicate that it's not confirmed. I'm still not altogether comfortable with having speculation in the article, but I guess this is better than nothing. I have a few more qualms with the rest of this paragraph that I haven't yet resolved. The first is that the citation on the second sentence has nothing to do with the content of the sentence. I agree, it's an interesting citation, but it doesn't indicate that this is a Cortex-A8. The second problem I have is that this isn't "compatible with the Cortex A8", it's compatible with ARMv7a. We really haven't got any way to say yet what the microarchitecture is. Not that that particularly problem isn't rampant in the article, but one issue at a time. I also don't think we actually know what the L2 count is; if we do, I'd like to see a cite on it. —Roguelazer (talk) 09:53, 28 April 2010 (UTC)

Don't worry about "having speculation" in the article, as long as you simply state what a reliable source is saying its compatible with how Wikipedia "works". Personally I think that there should probably should be a warning in the lede that Apple as sole source of definitive information has not confirmed most of these "facts". Mahjongg (talk) 11:11, 28 April 2010 (UTC)

page move to "Apple A4 Processor"

I oppose the undiscussed page move of this article from Apple A4 to Apple A4 Processor. The move was done under the pretext that this is how Apple names the chip, but the link given does not confirm this, in fact they call it the "Apple A4 chip". Its also factually incorrect to call it a "processor" as the Apple A4 contains much more that just the CPU (the "processor"). Among many other things it also contains a GPU, and many peripherals Because it is a "SoC", not just a processor . I propose the reversal of the page move, at least until the move is discussed first. Mahjongg (talk) 23:16, 18 June 2010 (UTC)

I concur -- Henriok (talk) 10:05, 19 June 2010 (UTC)

Upon reading your comments, I also agree. I have since unsuccessfully tried to revert my more. The reason I moved the page was because the title used to describe the A4 chip on the iPhone 4's Design page is the Apple A4 Processor – but if the chip contains a GPU then the page should be reverted back to Apple A4.

Scott Bywater (talk) 11:00, 19 June 2010 (UTC)

Reversion of move requested --Cybercobra (talk) 20:28, 19 June 2010 (UTC)

And done. --Cybercobra (talk) 01:47, 20 June 2010 (UTC)