RIM sees PlayBook OS as 10-year future for smartphones, tablets - Page 4

Originally Posted by Hiro

So I guess by your estimation Apple via it's Intrinsity purchase for modifying of the ARM core for power management doesn't add or subtract any gates on the chip? Yeah right. Wrong again.

And you still don't know how to read. The Reg didn't say there were only 4 licensee's, they gave 4 examples. And noted that 200 or so of ARMs customers don't like to change the cores, that means others of the remaining 400+ do like that option. The Reg is an entertaining & useful a low budget rumor/news reposter/editorial for frack's sake. Good for a lead, never take their info as the last word on anything or you are going to keep playing the part of the fool.

And oh what a fool you can be exposed as when playing ping pong with nht.

Originally Posted by samab

The only publicly known ARM architecture licensees are Marvell, Infineon, Qualcomm and Microsoft. There may be a fifth one that has not been publicly announced. But you were talking about hundreds and hundreds of architecture licensees --- which meant that you don't know what you were talking about.

http://www.bbc.co.uk/news/technology-10741448

Intrinsity does not have an architecture license --- therefore they didn't touch the core.

Do you know what Apple was doing with the Hummingbird? Apple was basically taking things out. The iphone/ipad don't have an HDMI output, so they took it out (and save a nickel here). They will never make a flip phone, they took out the 2nd LCD output (and save a nickel there). That's basically it. I am not saying that Apple's team is stupid. I am saying that given the time constraint for launching the A4, that's all they could do.

You can go and look at something like the RIM Playbook --- it has one LCD screen and one HDMI out. But the OMAP4 chip has 2 LCD screen output and one HDMI out. TI designed the chip so that in the event a phone manufacturer wants to make a smartphone with a flip phone form factor, they can use OMAP4. But RIM is basically wasting money on a CPU feature that they would never be able to use in the Playbook.

The more you talk, the more people will find out that you know absolutely nothing about this topic.

Originally Posted by Hiro

Attempted boomerang attack!

You do not know who has ARM licenses and nht directly contradicted you with the excerpts of the ARM conference call.

You now have two references that speak to public announcements of architecture licensees. But nht has delivered a ARM CEO announcement of a non-public licensee. There are also others doing defense work, and they aren't on your public list either. You can't keep citing second and third party reports of public announcements and impeach first party transcripts that include explicit announcements of partners that will remain unnamed.

I have no idea why you bring up the RIM Playbook as evidence for anything. RIM bought an off the shelf part. Big deal. Apple designs custom ASICS and has since the AppleII days. Apple has brought in significant ARM specific and CPU power management talent over the past two years. If we believe your premise that Apple could not possibly have an ARM architecture licensee, we would have to believe that Apple has no intention of doing any ARM CPU related design, because legally they could not, and more so wouldn't have the IP to even work with. Apple would be amongst the stupidest acquisition houses out there, buying 2 companies worth of talent for a niche they can't play in. I have a bit more faith in Steve and basic logic than that. Not to mention you keep pulling up tired and already contradicted sources.

Originally Posted by samab

So you are arguing I got it wrong because there is an unidentified fifth architecture licensee, when you were claiming that there were hundreds of architecture licensees.

And noted that 200 or so of ARMs customers don't like to change the cores, that means others of the remaining 400+ do like that option.

Originally Posted by samab

And nht is wrong anyway, because when they teardown the ipad, they also x-rayed the chip --- which has a stock A8 core inside.

Originally Posted by samab

So you are arguing I got it wrong because there is an unidentified fifth architecture licensee, when you were claiming that there were hundreds of architecture licensees.

Originally Posted by samab

I never said that the PA Semi guys are stupid, I said that given the time constraints they can't take a architecture license, design a compatible core/chip and have it out in a product in a couple of years.

Apple could very well be the fifth licensee, but so what? Do you know how long it took Qualcomm to design their own compatible core? When AIM (Apple, IBM and Motorola) fell apart, Qualcomm hired the entire embedded PowerPC chip team from IBM, gave them something like $300-400 million, and it took them 4 years to design the Scorpion core and the Snapdragon chip. And the IBM chip designers ain't stupid.

Originally Posted by Hiro

Again with the inability to read actual content. I quote myself:

See that little boldface part, it's the important part of the sentence you are ignoring. It says I don't know how many there are, but I know of at least one [and personally belive n is greater than 2]. Its mere existence shows you are not to be trusted because you are making things up again. Overreaching and fallacious restating of others points is a fatal indication of a losing argument.

No, it is not a stock A8, there is feature size reduction layout difference from the ARM 65nm reference design to the production 45nm part. Is it the same core as in a Hummingbird, looks like it. Is it an off the shelf totally unmodified ARM CortexA8, no.

Again with the inability to read actual content. I quote myself:

[IMG]And noted that 200 or so of ARMs customers don't like to change the cores, that means others of the remaining 400+ do like that option.[/IMG]

See that little boldface part, it's the important part of the sentence you are ignoring. It says I don't know how many there are, but I know of at least one [and personally belive n is greater than 2]. Its mere existence shows you are not to be trusted because you are making things up again. Overreaching and fallacious restating of others points is a fatal indication of a losing argument.



You keep talking about designing a new core, but that's not what most speculate Apple is trying to do, and definitely what I think they are trying to do. (Its not like they are designing totally custom weapon system CPU/SoC packages) They are in the game to modify the existing core designs for max power efficiency. A far tighter scope than designing a "new core" from architectural building blocks and exactly the kind of talent PASemi brought to the table. To do that you still need access to the full architecture so you can decide what splits where and how to test it when you make your decisions. This process has probably been going on since early 2008, and a 2009 audition by Intrinsity worked so good Apple felt compelled to gobble them up for total advantage and not having to share. A potential 2011 release would be nearly three years of work, and while it may only ship conservative (but impressive) first steps in Core power management those would be steps not possible for anyone else in the mobile space. I wouldn't expect radical game-changing Core power dissipation until some follow-on to the A9 series where Apple has been able to work with ARM from first inception of the Architectural revision. They will be able to make a lot of hay with the other parts of the SOC this time around though.

No IBM chip designers aren't stupid. But they are pig-headed. Power dissipation per unit of work has been abysmal in the Power series, IBM's premier line. They refused to make design tradeoffs that would allow Apple to field competitive laptops, and got very bunker mentality-ish over anything that didn't fit their Power Server line architecture. With the exception of funding a dedicated engineering effort like Microsoft did. So Apple walked and went Intel.

In the meantime IBM had a come to Jesus moment because their Power Servers aren selling enough to fund the PowerPC line on it's own. So now they do way overpriced supercomputers because there are no price controls or competitive forces in that arena and contract work. The contract work engineers are very receptive because they have to be, but it has taken each team of them several years to learn their craft all over again. So comparing results from a set of engineers successfully navigating through an identity crisis to a set of engineers with currently relevant goals and thought processes from the start doesn't work so well.

Originally Posted by samab

Let me quote the beginning:

I said "I know that only 4 companies to-date has signed ARM architecture license --- Qualcomm, Infineon, Marvell and Microsoft. They can design their own cores."

You said "You are showing yourself to be an absolute talk out your ass sideways idiot. Apple is an ARM licensor, as is TI- what do you think the OMAP series are? And Samsung is a licensor too--Hummingbird isn't made or designed by ARM, the CortexA8 core was licensed by Samsung so they could make the Hummingbird. There are also a handful of DoD and European defense contractors that contract for ARM IP directly too."

You have zero understanding from the beginning and yet you have been belittling me for days. I have been quite tolerant with you for awhile now.

You don't buy an architecture license and not build your own compatible core.

You don't go and buy 2 silicon firms with hundreds of silicon engineers just to tinker a few gates here and there. It cost a lot of money to get the architecture license up front and then you have to spend a lot of time and money to develop your own compatible core. Why do Qualcomm and Marvell and Infineon do it? Because you don't have to pay the "relatively" expensive cortex a8 license fee for every chip you make. Apple would be paying twice if they buy the architecture license and then tinker with a few cortex gates --- because they would still have to pay the per unit cortex license fee.

PA Semi designed 64 bit chips on the Power/PowerPC architecture before Apple bought them, so there was no running start.

The IBM team that designed the G5 belonged to the DESKTOP powerpc chip group, not the EMBEDDED powerpc chip group (which Qualcomm hired).

Originally Posted by Hiro

Given you kept changing the goalposts to suit your purposes I was correct in the rebuttal at that point in time. Your previous posts had been all over the map with exceptionally loose use of terminology. What are we supposed to be, psychic? Taking the revisionist history look at week old posts now that we have been debating a commonly-understood target makes it easy to snipe. Go back to where you were lax and say you meant something other that what it appeared you were referring to at that point. whoop-de-whorl, deal with the fact that in the current posts, debating an unambiguous topic, you are still failing and misstating facts.

Originally Posted by Hiro

Well, Qualcomm has the Snapdragon, which despite it's early promise seems to be a less than market beating chip. I don't know enough about Marvell, but they seem to not be keeping everyone else out of the mobile space either. But at least these folks did build their own cores.

Originally Posted by Hiro

Infineon wasn't designing their own cores, they were been grafting their stuff onto ARM cores, and now they are owned by Intel, so much for their meeting your criteria.

Originally Posted by Hiro

You really think MS is going to build CPUs? They never have before, they are good at contracting that out when needed, and there's no way they could be ready for the XBox follow on to the 360 with a new chip from scratch, so your analysis seems to be missing something with them.

Originally Posted by Hiro

So if you were being magnanimous and figuring MS and Infineon were eventually going to make ARM cores, I fail to see how you can discriminate Apple out of that same conversation when they have already shipped a custom ARM SOC. The power efficient core seems to be a reasonable outgrowth of that.

Originally Posted by Hiro

Your logic is abysmal here. I don't think anyone else on the planet thinks Apple is paying a full ARM architectural license to access the A8. That is undoubtedly A9 and on. You have to twist logic farther than a Klein bottle neck to get where you are.

Originally Posted by Hiro

Duh. Nobody said there was a running start. I said they had almost three years from the acquisition to work on something like an A9 derivative. Is it possible for you to respond to a sentence without making you own completely fantastical interpretation?

Originally Posted by Hiro

Company culture is company culture. And the embedded PowerPC group is just the group that gets to deal with the old designs and repurpose the already well amortized IP. They were all IBM PPC engineers in a culture that had the Power server line as it's pinnacle. All large organizations have cultures which are embedded, and it is the rare group that can survive bucking the trend. In retrospect with the milquetoast success Snapdragon has had to date, I might speculate that the whole hog dual core performance or bust decisions were borne of that culture and prevented a far more spectacular really low power chip with almost as good performance promise from emerging.

Originally Posted by samab

There is no changing of goalposts --- I originally said that if Apple wants to have a dual cortex A9 core for the ipad 2 and launch in April next year, it looks like that Apple/Samsung may not be able to make that on time (and that they have to source it from TI like RIM). Then suddenly a few days after I said it, rumors arose that Samsung placed a large order for the nVidia Tegra 2 chip to be used in their own Galaxy Tab 2.



Considering that 70-80% of the Android phones use the Snapdragon --- they are the market beating chip.



It's not that Infineon wasn't designing their own core --- it is that buying the architecture license is only step number 1. This is a company that barely dodged bankruptcy in 2009 and then had to sell their wireless business to Intel. Where the hell were they going to find $300-400 million dollars to develop their own core when they were facing bankrupty.



Nobody said that Microsoft is going to build CPU's. But now with an architecture license, Microsoft can hire say IBM to design their own ARM core if they want --- just like what they did with Xenon, which Microsoft owns the IP. Who the hell knows what Microsoft is up to? But they have over $40 billion in cash and they have plenty of long term projects.



As I stated earlier, Infineon just barely dodged bankruptcy and Microsoft has over $40 billion cash. Who knows what they would be doing?



This is how much you know NOTHING about this topic. It doesn't matter that you already paid for your architectural license --- that's just the instruction set to make a compatible core. ARM created implementations of the ARMv7 instruction set in the cortex series of core --- if Apple uses that ARM's implementation (even a small portion of it), then Apple has to pay a per unit license fee for that implementation.

Nobody pays twice when they can pay just once. If Apple is the fifth architecture licensee, then they will (1) in the short term, continue to make their own custom silicon with a stock cortex A9 chip and (2) in the long term, migrate to their own compatible core. There is no in-between option as you suggested because that would mean that they would have to continue to pay per unit cortex licensing fee.



Anything is possible, but the question is --- is it probable that they can do it in time? We just heard about Samsung being late with their own Orion chip and have to buy the nVidia chip. Even with the nvidia chip, quantity becomes a real problem --- when you look at LG's announcement of their dual core a9 smartphone that will only have enough to basically launch in South Korea only.



What planet are you from? Milquetoast success? They just name Qualcomm/Android (the Quadroid) as the new Wintel.

Originally Posted by Hiro

Wow, you really buy into the marketing hype. "Quadroid" A catchy term maybe, but not all that and a bag of chips.

70-80% of Android phones using it? Once again you are completely pulling numbers out your ass sideways. This is why I cannot even begin to accept 1% of what you say and nobody else should either. You give up on one set of fake numbers and then move on to manufacturing the next like it's just another afternoon cookie break.

Originally Posted by samab

More than 3/4 of the android phone models use Qualcomm chips.

http://money.cnn.com/2010/11/12/tech...roid/index.htm

The funny thing is that it seems the only non-Qualcomm based Android phones --- are the Verizon ones.

Note to all AI moderators, I have been very tolerant to Hiro's belittling so far. But I do have my limit and I am prepared to launch unlimited warfare to defend myself.

Originally Posted by Hiro

You are quick to reply. Do you wait at the keyboard for my every post? You do have my original text, but you sat on it for awhile didn't you? Because I changed it about three minutes later, softening it a bit, I thought that was probably the right thing to do. My later edit for a typo still beat your small post to the board.

Your only necessary defense is to stop with the misinformation, I am only responding to your continual posts of questionable truthiness. I do admit, not every nugget you post is made up, the Quadroid thing just appears to an attempt to be inflammatory in a different way. It worked for a few minutes, but you can see it wasn't quite as successful as you hoped.

As for the rest of it, you do still need to learn to read what is written rather than what you want to read though. That little weakness of yours leads to lots of mis-charcterizing posts. Along with reading, I expect you could also be better served by doing a little analysis and basic sanity checking.

Do you know the source of the 75-80% number.? Which was actually reported as 77% by the source? You might reconsider the veracity of the claim if you actually did a little 2+2 on who ships what, which carriers ship the most smartphones in total number (didn't I read really recently the name Verizon?) and that a business consultancy somehow has better numbers than any Street analyst? PRTM is trying to sell their own brand of Android services by a little opportunistic accounting.

I don't think verification and subsequent doubt of a posted on the internet data-let constitutes heinous ad hom behavior. It's just good data hygiene.

Notice that there are no other posters on the boards that have run to your defense stating my or nht exposing of your rampant inaccuracies. Not even any of the ever present sock puppets have appeared to have waded-in, and they would love nothing else other than seeing me get a virtual black eye.

I do so chuckle at the continued attempts to say I don't have any idea of what I am posting about, what an OS is, an ISA, a contract. It's all very engaging but oh so smacks of trying to deflect examination of the substance or lack thereof. Apparently I can also create "in-between options" from thin air. The only thing I know that can do that is the event horizon of a black hole, so I guess your in-between options are just virtual fantasies of your own eyes, or at least a heaping portion of wishful imagination.

To then think my (virtual) ideas show Apple would paying twice for something, and that would be the contradiction to break the chain is actually a nice shot at proof by contradiction. But to get there you have had to oversimplify both contracts and motivations. The oversimplifications place too great a weight on your argument an it collapses upon itself. The clear lightweight non-self-licking-lollipop is that Apple pays ARM for what Apple wants out of that particular contract. The fact Apple may have multiple business relationships with an old friend of a company doesn't mean they are paying multiple times for the same thing. They are paying ARM multiple times for different things on different timescales.

You really underestimate the concept of long-thought, which probably means you aren't old enough to have been shafted by long-thought yet. [Exercise: what long thought might I be referencing? Whatever the answer is can only be a guess unless you work in Apple and dine with Steve, neither of which I do. So I am only "guessing too", but not all guesses are created equal so give it a shot.] Oh well, either you will develop the sight or you won't, not for anyone else to say. But keep on keeping on, keep your eyes open and learn to not be so believing of the first article you come across. I promise you I will continue to be a deep reader, not a surface skimmer.

Originally Posted by Hiro

CortexA9 has been licensable as IP from ARM since early 2009.

Originally Posted by samab

It's like the xbox CPU's. Microsoft licensed the IBM powerpc architecture and then hire IBM to help them design the xenon cpu. The end result is that the CPU itself is a Microsoft's own IP. Who did most of the work? It's IBM.

Originally Posted by samab

I specifically stated that it is the same as IBM designing the Xenon chip and Microsoft owning the IP. Microsoft bought an architecture license for the PowerPC chip, then hire IBM to design the Xenon. Microsoft keeps the intellectual property for the Xenon --- IBM can't sell the Xenon to another customer and can't license the Xenon to another customer.

Originally Posted by samab

This is like AMD selling a x86 compatible CPU.

Originally Posted by Hiro

ARM licenses IP, not chips.

Originally Posted by samab

I know that only 4 companies to-date has signed ARM architecture license --- Qualcomm, Infineon, Marvell and Microsoft. They can design their own cores.

I know that all the other players license the cores and put their own extra stuff around it.