or Connect
AppleInsider › Forums › Mobile › iPhone › Apple's 64-bit A7 SoC 'set off panic' for chipmakers
New Posts  All Forums:Forum Nav:

Apple's 64-bit A7 SoC 'set off panic' for chipmakers - Page 3

post #81 of 131

Sorry, I have a much higher standard for Apple, especially with the price I am paying for a low-end laptop like Macbook. If I purchased any other low-end brand laptop for a very cheap price, then I probably wouldn't complain. It seems like you (and many others) gave Apple a free pass just because it is only an "esthetic problem." If you have such low standard for the price you paid Apple and if you have the mentality of the American auto industries back in the 80s and 90s, that is fine by me. I've never said the crack tempered the laptop functionality and that is not the point.

 

Again, you questioned rather or not I actually own the product, like the way you assume that the iOS 7 crash is my own fault. That does not help the discussion at all. I still own the white plastic Macbook (2007) and I returned it once for a new case and now the bottom part started to crack. I've never questioned your experience just because your experience is different than mine.

 

Quote:

Originally Posted by BigMac2 View Post
 

 

Agree It is always easier to blame some else for their own mistake. Also I do agree the volume of sales do not represent the quality of a product, but it make much more easy to identify a major design flaw if every device is affected with.  Apple has never got a free pass before, every new product Apple comes with is scrutinize by so many groups so happy to get something bad to say about Apple and their product.

 

I totally disagree with you about calling those MacBook being plastic junk just because of a esthetic problems, have you got one of those MacBook? I've got many at my office, I still do have one in perfect shape, for the price and the time (2007), it was one of the best small laptop on the market, the cracking case issues was purely esthetic and never tempered the laptop functionnallity.

post #82 of 131
Quote:
Originally Posted by old-wiz View Post

Hmm...So Samsung is building the A7....and how much of it will they copy since they will have have access to the chip? How much of Xynos will be copied from the A7? this will make a great patent war...

 

They didn't have access to the A7 architecture, and none of it will be copied. At least for 2014 where most of the 64-bit CPUs will be Cortex-A53 on 28 nm, or -A57 if they have a 20 nm node ready in 2H 14. Most of the CPU designs will end up in the same place: 4-issue, OOE, lots of caches, dual and quad core, and on-die power regulator for active idle. A Haswell type architecture is just about where they are all heading.

post #83 of 131
Quote:
Originally Posted by sog35 View Post
 

 

Theres this thing called Glasses.  They help with vision.  You may want to research it and come back to us in a week.  While you are at it buy a Nexus7 and report back to us.

 

There is also this other thing called a bigger font which is easier to see, especially on a retina display.

 

Sticking with the low resolution screen of a 3GS does not seem like a very good solution.

Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
post #84 of 131
Quote:

Originally Posted by rgwychu View Post

 

Sorry, I have a much higher standard for Apple, especially with the price I am paying for a low-end laptop like Macbook. If I purchased any other low-end brand laptop for a very cheap price, then I probably wouldn't complain. It seems like you (and many others) gave Apple a free pass just because it is only an "esthetic problem." If you have such low standard for the price you paid Apple and if you have the mentality of the American auto industries back in the 80s and 90s, that is fine by me. I've never said the crack tempered the laptop functionality and that is not the point.

 

Again, you questioned rather or not I actually own the product, like the way you assume that the iOS 7 crash is my own fault. That does not help the discussion at all. I still own the white plastic Macbook (2007) and I returned it once for a new case and now the bottom part started to crack. I've never questioned your experience just because your experience is different than mine.

 

I do questioned your over reacting impression of calling things broken or crap base on anecdotal impression.   Apple haven't got a free pass for this, as you related your self, Apple made a recall and replaced the top case free of charge, I don't think those issue is enough to utterly call it crap.  Same apply on your assertion about the A7 and iOS7.

 

I've seen so much worst problem in the past, like the unstable for real Quadra 700, the bad power supply iMac G5, the ram amnesic PowerBook G4, the leaked PowerMac G5, the hinge cracking PowerBook Ti, etc..... Nothing is perfect, Apple included, but on the long run Apple is better a fixing things than other.


Edited by BigMac2 - 12/17/13 at 12:24pm
post #85 of 131
Quote:
Originally Posted by THT View Post
 

 

They didn't have access to the A7 architecture, and none of it will be copied. At least for 2014 where most of the 64-bit CPUs will be Cortex-A53 on 28 nm, or -A57 if they have a 20 nm node ready in 2H 14. Most of the CPU designs will end up in the same place: 4-issue, OOE, lots of caches, dual and quad core, and on-die power regulator for active idle. A Haswell type architecture is just about where they are all heading.


I was being a little facetious... with regard to Samsung (and others) copying Apple..

post #86 of 131
Quote:
Originally Posted by EricTheHalfBee View Post

Remember when Chipworks "examined" the A7 die to get an idea how it works? I guarantee you Samsung would know a lot more than Chipworks would since they make the processor. They might not know how it would perform or how many instructions it can do per clock (6, BTW), but they'd definitely know it was 64 bit along with how much cache it had and many other details.

But how? I'm still cloudy on how they would know a chip is 32 or 64-bit. We're not talking about a node change which is something that needs to be performed at the fabrication level and is completely up to Samsung but an instruction set of a chip. What about Chipwork's images say "This is 64-bit"? Samsung would know that Apple included 4MB of RAM on the chip and other physical features but I don't see how Samsung would know it's 64-bit just by looking at it.

if they really did know all about this chip itt really makes that part of Samsung uncharacteristically tightlipped and ethical when they could have leaked the news and let Samsung make an announcement when Apple starts ramping up production before the iPhone 5S was announced.

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply
post #87 of 131
Quote:
Originally Posted by BigMac2 View Post
 
Quote:

Originally Posted by rgwychu View Post

 

Sorry, I have a much higher standard for Apple, especially with the price I am paying for a low-end laptop like Macbook. If I purchased any other low-end brand laptop for a very cheap price, then I probably wouldn't complain. It seems like you (and many others) gave Apple a free pass just because it is only an "esthetic problem." If you have such low standard for the price you paid Apple and if you have the mentality of the American auto industries back in the 80s and 90s, that is fine by me. I've never said the crack tempered the laptop functionality and that is not the point.

 

Again, you questioned rather or not I actually own the product, like the way you assume that the iOS 7 crash is my own fault. That does not help the discussion at all. I still own the white plastic Macbook (2007) and I returned it once for a new case and now the bottom part started to crack. I've never questioned your experience just because your experience is different than mine.

 

I do questioned your over reacting impression of calling things broken or crap base on anecdotal impression.   Apple haven't got a free pass for this, as you related your self you've got it repair from Apple free of charge, I don't think those issue is enough to utterly call it crap.  Same apply on your assertion about the A7 and iOS7.

 

Sure, Apple didn't get a free pass just because people like me are "overreacting based on anecdotal impression". Wow, Apple is willing to repair their defective product for free because people like me are "overreacting based on anecdotal impression". It's a real wonder of how you view things. Go tell Anandtech and iPad Air users that their article is just based on some anecdotal impression just because you don't have the same issue. The fact that Apple took care of the Macbook plastic crack, which the media didn't make a big deal out of it at all, just proves my point. You can't say an issue doesn't exist just because the media didn't make a big deal out of it.

 

Paying hundreds of dollars more for the same hardware quality that you get from other laptops. If a product that sucks you extra hundreds of dollars for nothing shouldn't be called junk, then I'm not sure what should be qualified as junk. What difference does it make if I saved that hundreds of dollars to buy a used laptop that doesn't work? I guess hundreds of dollars don't really worth much to you.

Yeah, plastic Macbook had Mac OS X, so did Macbook Pro, only if somebody told me that it was worth the price. And I bet that somebody wouldn't be people like you who tell plastic Macbook owners that the crack is just an anecdotal impression.

 

If what you have to say is "you are wrong because of this technical reasoning", then I am all ears and I respect that. However, all you are doing so far is "There must be something wrong with you because your experience is different than mine. Your experience is based on anecdotal impression because it is different than mine. The same applies to those who post their experience that are different than mine on any online forum, including Anandtech's article." I'm not sure how that kind of reply itself is not based on "anecdotal impression." I guess that is your best shot.


Edited by rgwychu - 12/19/13 at 6:57am
post #88 of 131
Quote:
Originally Posted by THT View Post

My take on benefits of 64-bit is that it's a holistic one for the consumer. The big benefit will be for developers (Apple software included) and customers get a downstream effect of better software.

Virtual address space goes to whatever terabyte or exabyte levels it will be, instead of the 4 GB it is today. This will make it easier for them to make more complex software. So, for obvious things like video, 3D, big computation applications, developers can get going designing their software now. It'll allow for desktop level implementations of web browsers, which is likely the largest user-facing, highest usage application on computers today.

Then, for hardware, it opens the door for Apple to using their ARM SoC in "bigger" hardware, like the supposed 13" iPad or a new 12" laptop. If Apple keeps going at the pace of SoC development where CPU and GPU performance is doubling year over year, they can used it for Macintoshes in a year or so. iMacs have a curvy bulge in the back about 1.5" thick or more. With an 5 Watt TDP SoC, an iMac could be 0.5" thin. With 5 Watt TDP, instead of the 7.7" x 7.7" Mac mini, it'll be like the Apple TV or Airport Express, just 4" x 4" and 0.75" high. Lastly, the laptops can get thinner and lighter. Instead of a 3 lb 0.68" thick 13" MBA, it'll be 2 lb and 0.5" thick instead.

Thanks for the reply. I would be interested in a MBA with an ARM chip. I would keep an iMac at home to do the heavy lifting and use the MBA for just browsing, email, etc. on the go.
post #89 of 131
Quote:
Originally Posted by THT View Post

 

The problem with Silvermont is that it's still trying to fit inside a box of not competing with Core while arriving a little bit late. They are essentially a new entrant in the mobile world and have to push out the incumbent. Silvermont is at parity with ARM SoCs. To push out the incumbent, they have to be better, both in performance and perf/watt.

 

x86 compatibility for Windows tablets is a big deal.  Windows on ARM is pointless.

 

Core is going to be Ultrabooks and up.  Silvermont will get marketed as Atom, Celeron and Pentium for lower end products.

 

Whether they get smartphone design wins is still up in the air.  I'm not certain that Intel cares that much even now.  They sure as hell care that ARM doesn't encroach on laptops and servers and would like to win in tablets.  

 

But if they really cared about being real players in smartphone SoC market they could have been making ARM SoCs this whole time.  I'm thinking the margins simply haven't been all that enticing except for maybe the top tier SoCs used in flagship phones.

post #90 of 131
Quote:
Originally Posted by nht View Post
 

 

x86 compatibility for Windows tablets is a big deal.  Windows on ARM is pointless.

 

Core is going to be Ultrabooks and up.  Silvermont will get marketed as Atom, Celeron and Pentium for lower end products.

 

Whether they get smartphone design wins is still up in the air.  I'm not certain that Intel cares that much even now.  They sure as hell care that ARM doesn't encroach on laptops and servers and would like to win in tablets.  

 

But if they really cared about being real players in smartphone SoC market they could have been making ARM SoCs this whole time.  I'm thinking the margins simply haven't been all that enticing except for maybe the top tier SoCs used in flagship phones.

 

I think Windows compatibility is the biggest deal for having a x86 on a tablet and I'm sure Microsoft is lobbying really hard to push this solution, I think this is also why there is near zero interest on having an Intel SoC in a smartphone.


Edited by BigMac2 - 12/17/13 at 1:54pm
post #91 of 131
Quote:
Originally Posted by rgwychu View Post
 

 

Sure, Apple didn't get a free pass just because people like me are "overreacting based on anecdotal impression". Wow, Apple is willing to repair their defective product for free because people like me are "overreacting based on anecdotal impression". It's a real wonder of how you view things. Go tell Anandtech and iPad Air users that their article is just based on some anecdotal impression just because you don't have the same issue.

 

Paying hundreds of dollars more for the same hardware quality that you get from other laptops. If a product that sucks you extra hundreds of dollars for nothing shouldn't be called junk, then I'm not sure what should be qualified as junk. What difference does it make if I saved that hundreds of dollars to buy a used laptop that doesn't work? I guess hundreds of dollars don't really worth much to you.

Yeah, plastic Macbook had Mac OS X, so did Macbook Pro, only if somebody told me that it was worth the price. And I bet that somebody wouldn't be people like you who tell plastic Macbook owners that the crack is just an anecdotal impression.

 

So how many of these other "same hardware quality" laptops are still around six years later?

 

You are obviously still using the MacBook or why else would you be whining so much about it.

Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
post #92 of 131
Quote:
Originally Posted by st88 View Post
 

Meanwhile, Intel's x86-64 Bay Trail (22nm Silvermont + Ivy Bridge GPU) is already on the market. In 2014 Intel is planning their tick-tock with Cherry Trail (14nm Airmont + Broadwell GPU) and Willow Trail (14nm Goldmont + Skylake GPU).  Intel will also be launching a successor to their current LTE chip with LTE-Advanced in 2014.

 

So Intel is on a slow decline, and are dead in mobile.

post #93 of 131
Quote:
Originally Posted by hill60 View Post
 

So how many of these other "same hardware quality" laptops are still around six years later?

That depends on the condition of the laptop you are talking about...

 

My wife's six year old laptop (the brand is MSI if you've ever heard of it) is still running fine functionally, especially if you don't consider the esthetic issues, which prove to be trivial for some.


Edited by rgwychu - 12/17/13 at 1:44pm
post #94 of 131
Quote:
Originally Posted by rgwychu View Post
 

That depends on the condition of the laptop you are talking about...

 

My wife's six year old laptop (the brand is MSI if you've ever heard of it) is still running fine functionally, especially if you don't consider the esthetic issues, which prove to be trivial for some.

 

So an aesthetic issue, like say, a crack in a case?

Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
post #95 of 131
Quote:
Originally Posted by ipen View Post
 

I wouldn't call it "innovate".  It's certainly the next step.  Apple just took the lead and came out with the 64bit faster than others.  Good work.

 

64-bit in and of itself isn't the innovation, but rather why it was done (massive performance improvements without increased power consumption) and how it was done (Apple controls the ecosystem and the entire platform so recompiling to 64-bit is often no more than flipping a switch).  If that doesn't qualify as innovation then I don't know what is.  On the other hand the word "innovation" is used so much these days it's almost lost all meaning.  For example a bigger screen is not something I would call innovation.

post #96 of 131
Quote:
Originally Posted by kForceZero View Post
 

 

64-bit in and of itself isn't the innovation, but rather why it was done (massive performance improvements without increased power consumption) and how it was done (Apple controls the ecosystem and the entire platform so recompiling to 64-bit is often no more than flipping a switch).  If that doesn't qualify as innovation then I don't know what is.  On the other hand the word "innovation" is used so much these days it's almost lost all meaning.  For example a bigger screen is not something I would call innovation.

 

A bigger screen originally intended to hide a bigger battery for power hungry systems is closer to the truth.

 

Practicality not innovation.

Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
Better than my Bose, better than my Skullcandy's, listening to Mozart through my LeBron James limited edition PowerBeats by Dre is almost as good as my Sennheisers.
Reply
post #97 of 131
Quote:
Originally Posted by rgwychu View Post
 

My wife's six year old laptop (the brand is MSI if you've ever heard of it) is still running fine functionally, especially if you don't consider the esthetic issues, which prove to be trivial for some.

 

Yeah, they were great.

 

Seven and a half pounds of Intel Centrino + Vista goodness. I like the giant vent holes on the side. 

 

So much better than a MacBook.

post #98 of 131
Quote:
Originally Posted by BigMac2 View Post
 

 

I think Windows compatibility is the biggest deal for having a x86 on a tablet an I'm sure Microsoft is lobbying really hard to push this solution, I think this is also why there is near zero interest on having an Intel SoC in a smartphone.

 

Is this really true? Windows compatibility is likely huge for things like enterprise desktops (where specific applications have been design to help the business run), but in the tablet space, we're talking consumer usage and office automation usage , right? In that space, those usages are becoming more and more web-based and office automation is becoming more and more commoditized.

 

Microsoft's problems in tablets aren't really x86 SoC or legacy app related. I still think overall Surface hardware wasn't that great (a year late and bulky), the live tile UI wasn't that great (too cold and foreign for Windows users), and MS didn't really commit to making what they have great.

 

Intel's problem is that they just aren't that interested. They don't view $20 ARM SoCs as life threatening or mobiles as threatening their business yet.

post #99 of 131
Quote:
Originally Posted by THT View Post
 

 

They didn't have access to the A7 architecture, and none of it will be copied. At least for 2014 where most of the 64-bit CPUs will be Cortex-A53 on 28 nm, or -A57 if they have a 20 nm node ready in 2H 14. Most of the CPU designs will end up in the same place: 4-issue, OOE, lots of caches, dual and quad core, and on-die power regulator for active idle. A Haswell type architecture is just about where they are all heading.

 

They sure did have access to it. See below...

 

Quote:
Originally Posted by SolipsismX View Post


But how? I'm still cloudy on how they would know a chip is 32 or 64-bit. We're not talking about a node change which is something that needs to be performed at the fabrication level and is completely up to Samsung but an instruction set of a chip. What about Chipwork's images say "This is 64-bit"? Samsung would know that Apple included 4MB of RAM on the chip and other physical features but I don't see how Samsung would know it's 64-bit just by looking at it.

if they really did know all about this chip itt really makes that part of Samsung uncharacteristically tightlipped and ethical when they could have leaked the news and let Samsung make an announcement when Apple starts ramping up production before the iPhone 5S was announced.

 

You have to understand how processors are designed. Engineers use software to lay out the processor. The end result is sort of like a blueprint, if you will (or like an AutoCAD drawing, only much, much more complex and detailed). Samsung would use this blueprint to manufacture the processor. It wouldn't be too hard to look at the registers (which are just memory) and determine from the number and layout of transistors how many they are and how wide they are.

 

Chipworks wouldn't be able to tell if a processor is 64bit since they are looking at a finished device. Samsung actually has the blueprint for the device. Samsung would also need access to the instruction set so they could perform testing of the chips themselves (unless Apple does this, which I doubt).

 

However, it would be very difficult for Samsung to duplicate the A7 and customize it/improve it (for example, making a quad core) since they don't have access to the original files used to make the "blueprint". Remember the output from whatever software Apple uses is intended to fab plants. It's not the same as the front end the engineers are looking at when they actually design the processor.

 

 

This is when I laugh at people who claim Samsung is primarily responsible for the A7 - they're not. The sheer amount of work Apple engineers would invest to design the A7 would completely dwarf the amount of work Samsung would do to actually start fabbing the A7.

 

So Apple doesn't have to worry about Samsung copying their processors, unless they made a literal 100% copy and called it something else. And that's something that would be easy to prove in court (and a very stupid move for Samsung to do). Who would trust them for anything if they took their "copying" to such an extreme?

post #100 of 131
Quote:
Originally Posted by pdq2 View Post
 

 

Yeah, they were great.

 

Seven and a half pounds of Intel Centrino + Vista goodness. I like the giant vent holes on the side. 

 

So much better than a MacBook.

Their latest ultrabook looks nice

http://www.msimobile.com/level3_productpage.aspx?cid=116&id=387

post #101 of 131
Quote:
Originally Posted by EricTheHalfBee View Post

It wouldn't be too hard to look at the registers (which are just memory) and determine from the number and layout of transistors how many they are and how wide they are.

What about the registers and width (of pointers) say it must be 64-bit? I know from both Chipworks and AnandTech that Apple's A5X in the iPad 3 used a very wide memory interface and it was only 32-bit and, I think, 512MB RAM. According to Mike Ash the number of registers did double but can't Apple also do that without going to 64-bit? Note that in ARM64 Mike Ash notes these are still half what you see in other modern processors.

"There are two parts of the CPU that "X-bit" usually refers to: the width of the integer registers, and the width of pointers. Thankfully, in most modern CPUs, these widths are the same. "64-bit" then typically means that the CPU has 64-bit integer registers and 64-bit pointers."

"ARM64 doubles the number of integer registers over 32-bit ARM. 32-bit ARM provides 16 integer registers, of which one is a dedicated program counter, two more are given over to a stack pointer and link register, and the other 13 are available for general use. With ARM64, there are 32 integer registers, with a dedicated zero register, link register, and frame pointer register. One further register is reserved for the platform, leaving 28 general purpose integer registers."

The floating-point registers in AR-64 are 128-bit and there are only 32 of them which tells me it's possible Scotty, I mean Samsung, didn't know.

"ARM64 also increases the number of floating-point registers available. The floating point registers on 32-bit ARM are a bit odd, so it's tough to compare. It has 32 32-bit floating point registers which can also be viewed as 16 overlapped 64-bit registers, and there are 16 additional independent 64-bit registers. The 32 total 64-bit registers registers can also be viewed as 16 overlapped 128-bit registers. ARM64 simplifies this to 32 128-bit registers, which can also be used for smaller data types, and there's no overlapping."

By no means am I saying you're incorrect but I haven't yet been convinced that what Samsung was given to fab for Apple clearly shows they were going to release a 64-bit CPU.
Quote:
So Apple doesn't have to worry about Samsung copying their processors, unless they made a literal 100% copy and called it something else. And that's something that would be easy to prove in court (and a very stupid move for Samsung to do). Who would trust them for anything if they took their "copying" to such an extreme?

But what about Samsung stealing their thunder by announcing a 64-bit ARM chip before Apple did? Of all the things Samsung has done that would seem so minimal.

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply
post #102 of 131
Quote:
Originally Posted by Just_Me View Post
 

Their latest ultrabook looks nice

http://www.msimobile.com/level3_productpage.aspx?cid=116&id=387

 

umm...suuure.

 

Or, you could get a MacBook Air, with the same processor and memory, and save a hundred bucks.

 

But then, you would miss out on the excitement of MSI:

 

Quote:
"From rough to refinement, we found MSI from nothing. Our logo's evolution witnesses the advances we have made, and we will keep improving ourselves all the way. With you customers’ support, can push us moving forward to the next stage!"
post #103 of 131
Is this the first arm 64 bit chip to hit market?

Most people so far, you don't need 64 bit, it's a gimmick, next year, why did it take android so long.
post #104 of 131
Quote:
Originally Posted by EricTheHalfBee View Post
 

 

They sure did have access to it. See below...

 

 

You have to understand how processors are designed. Engineers use software to lay out the processor. The end result is sort of like a blueprint, if you will (or like an AutoCAD drawing, only much, much more complex and detailed). Samsung would use this blueprint to manufacture the processor. It wouldn't be too hard to look at the registers (which are just memory) and determine from the number and layout of transistors how many they are and how wide they are.

 

And when Apple hands over the masks for laying out the transistors and metal layers on the wafer, what is the Samsung fab engineer seeing? They aren't seeing a blueprint. It's doubtful to me that Apple is handing over the circuit design or "blueprint" for the SoC. They are handing over circuit masks to etch the transistors. Apple and Samsung work to make sure the masks are designed correctly for the process involved, but I doubt that Samsung Semi knows at any point time what those masks are building.

 

Samsung would have to steal a wafer and put it under a microscope. Basically the same thing Chipworks did. This is not some day at the park type of activity. It would be act of espionage. Not worth the cost for what you can figure out based on images of the product.

post #105 of 131
All this talk about Samsung looking at Apple's blueprints and processors...

Samsung also makes their OWN processors... Samsung has their OWN engineers.

Doesn't anyone at Samsung know how to move to 64-bit? Does Samsung really have to look at someone else?
post #106 of 131
Quote:
Originally Posted by Curtis Hannah View Post

Is this the first arm 64 bit chip to hit market?

Most people so far, you don't need 64 bit, it's a gimmick, next year, why did it take android so long.

 

Yes.

 

X-Gene, a micro-server 64-bit ARM part, was announced to be available on Oct 30, 2013. Using that date, Apple likely shipped 10m A7 SoCs before Applied Micro even took their first order. It's quite likely Apple will have shipped 50m A7 SoCs by end of 2013. Who knows how many Applied Micro will have shipped by Dec 31. Their product page still says "pre-order".

 

As for smartphone SoCs, the linked article is likely right, at the earliest, the industry thought it was going to be 2H 14 for the first ARM 64-bit SoCs and 2015 is really the year when 64-bit ARM ships in significant numbers in the smartphone space. Apple shipping a 64-bit part in Sep of 2013 was a surprise to everyone, including ARM Holdings themselves.

 

However, Intel does have 64-bit Baytrail (for tablet, netbook, and cheak PC markets) available today and Merrifield will be coming for smartphones in 1H 2014. Unfortunately, Windows 8 Baytrail systems will be 32 bit systems since the 64-bit connected standby implementation is running late. Won't see 64-bit x86 Windows tablets until 1Q 14 at the earliest. I have no idea where WP8 is at w/respect to 64 bit support.

 

As for Android, who knows. They have to navigate a gigantic 32-bit, 64-bit driver minefield and the middleware layers (dalvik VM, NDK, APIs) have to be converted. Intel apparently has done quite a bit of work for x86-64 Android, but I don't know how far along that effort is. Who knows, maybe a Merrifield phone will be the first 64-bit Android device. For 64-bit Cortex-A53 and -A57 ARM SoCs, probably 2H 14 at the earliest.

 

It's important to put it in perspective though. Like quad-core or 4k or 20+ megapixels or f/2.0 or 802.11ac/ad or QHD (2560x1440) or LTE-advanced, it's only one vector by which a company competes, with all of them basically overserving customers. Apple may have big aspirations for its 64-bit ARM platform, so it does represent some very interesting possibilities for 2014 hardware if they are able to either get 20 nm parts out or use 10-20 W TDP parts.

post #107 of 131
Quote:
Originally Posted by Michael Scrip View Post

All this talk about Samsung looking at Apple's blueprints and processors...

Samsung also makes their OWN processors... Samsung has their OWN engineers.

Doesn't anyone at Samsung know how to move to 64-bit? Does Samsung really have to look at someone else?

 

Yes. All Exynos SoCs to date use CPU designs from ARMH and GPU designs from ARMH or ImgTec. Samsung does indeed have great SoC designers, but a custom design will be a whole new level for them. They appear to have plans to do a custom ARM CPU, but it would be 2015 before you see that. For 2014, if they can fit a Cortex-A57 into a smartphone TDP, they'll likely do that. Otherwise they stretch a Krait (Snapdragon 805) or a Cortex-A15. Their Exynos 5250, 5410, 5420 adventure with Cortex-A15 and big.LITTLE hasn't been the smoothest roll out, so hopefully, they learned and will do better in 2014.

 

Apple and Qualcomm currently get the benefit of the doubt because they have already shipped 2 custom CPU cores. There's even speculation that a custom GPU is next for Apple. Nvidia? Not a lot of wins there.

 

Edit: Forgot to say that MediaTek is a big player too. They could surprise.

 

Edit 2: Forgot that AMD is also looking to produce a 64-bit ARM SoC. Probably 2014. Who knows with them.


Edited by THT - 12/17/13 at 3:28pm
post #108 of 131
To those who keep using the term Samsung as in one company , please read carefully about their company structure. They are a enormous company. We are not even talking about division. Each one of them have their own operation chef and they are highly competitive against each other. To those who only knows about their crap mobile phone and wanna be products. They also build ships, air defence system , sell insurance, banking and even own kimchi plantation (indirectly) but that just a tiny bit of Samsung holding business interest. After the law sue start between Apple and Samsung mobile division. Their chef of foundry business fly to US personally to smooth thing over (it was wildly report in supply chain news) so yes they know and no not everyone knows. And I can bet any money Apple put a huge penalty cause if trade secret been leak in bold Helvetica font right on the front page of the contract ... 64 bit interger even in won will be the worth of a few Samsung 1wink.gif
post #109 of 131
I just came back from Apple tech talks London (cheers for the free lunch and free beers)

Well, to those mommy's boy never pay taxes and bitch about standard and over priced product. Moved along, nothing for the like of you here.

To those who are interested. Just from what I read between the lines - the term "the most forward thinking" is an extremely understatement. from a pure software POV to see their road map. Boy, I have to say, we have just caught a glimpse of what Apple want to do in the next 10 years. Developers know what I am talking about 1wink.gif
post #110 of 131
Quote:
Originally Posted by THT View Post

Yes. All Exynos SoCs to date use CPU designs from ARMH and GPU designs from ARMH or ImgTec. Samsung does indeed have great SoC designers, but a custom design will be a whole new level for them. They appear to have plans to do a custom ARM CPU, but it would be 2015 before you see that. For 2014, if they can fit a Cortex-A57 into a smartphone TDP, they'll likely do that. Otherwise they stretch a Krait (Snapdragon 805) or a Cortex-A15. Their Exynos 5250, 5410, 5420 adventure with Cortex-A15 and big.LITTLE hasn't been the smoothest roll out, so hopefully, they learned and will do better in 2014.

Apple and Qualcomm currently get the benefit of the doubt because they have already shipped 2 custom CPU cores. There's even speculation that a custom GPU is next for Apple. Nvidia? Not a lot of wins there.

Edit: Forgot to say that MediaTek is a big player too. They could surprise.

Edit 2: Forgot that AMD is also looking to produce a 64-bit ARM SoC. Probably 2014. Who knows with them.

Thank you! That explains it well.
post #111 of 131

Quote:

Originally Posted by THT View Post
 

 

Is this really true? Windows compatibility is likely huge for things like enterprise desktops (where specific applications have been design to help the business run), but in the tablet space, we're talking consumer usage and office automation usage , right? In that space, those usages are becoming more and more web-based and office automation is becoming more and more commoditized.

 

Microsoft's problems in tablets aren't really x86 SoC or legacy app related. I still think overall Surface hardware wasn't that great (a year late and bulky), the live tile UI wasn't that great (too cold and foreign for Windows users), and MS didn't really commit to making what they have great.

 

Intel's problem is that they just aren't that interested. They don't view $20 ARM SoCs as life threatening or mobiles as threatening their business yet.

 

Don't get me wrong, I do share the same feelings of yours. 

 

There is failed Windows tablet attempts for more than 15 years now, but the dream of having a full Windows on a mobile device is a hard one to kill in people imagination.  To my understanding the reason why all Windows tablet are still a market failure come from the fact that none are better than a conventional laptop at the same price point, and legacy UI or apps cannot be used without a keyboard and mouse which neglect the tablet form factor attributs.  The sadly truth is nothing can replace the keyboard and mouse input efficiency in legacy apps, and having a mandatory keyboard and mouse on a tablet is a worse compromise.

 

Intel and Microsoft has big interested into protecting their interdependent IPs.  In the past, most enterprise and home users had a WinTel computers, since few years ago things started to change and while most people still got a PC at home, most of them sees better value to keep their aging PC and got a new tablet instead.  Microsoft already lose his monopole of the web browsing machines, their OS and Office suite will follow soon enough. And without windows nobody will depend on x86 architecture anymore. 

post #112 of 131
Quote:
Originally Posted by Michael Scrip View Post

All this talk about Samsung looking at Apple's blueprints and processors...

Samsung also makes their OWN processors... Samsung has their OWN engineers.

Doesn't anyone at Samsung know how to move to 64-bit? Does Samsung really have to look at someone else?

 

Their most powerful phone out at the moment is the Note, which still doesn't have a need for (and wouldn't benefit much) from 64-bit architecture.

 

The big 'knob' is memory.  Anything over 4 gig and you absolutely need a 64-bit architecture or you are going to need two clock cycles for one memory address.  Since the Note has 3 gig, 32-bits are just fine, and it beats the 64-bit iPhone in quite a few performance metrics while the iPhone certainly returns the favor and clobbers the Note in other benchmarks.  Both are amazingly zippy devices.

 

I think where a lot of people went wrong on the iPhone is saying that because the iPhone doesn't have 4 gig of memory, it doesn't need a 64-bit bus.  If you go over 4 gig, yes, you absolutely need a 64-bit bus (and Samsung will likely go there once it needs to), but just because you only have one gig doesn't mean you can't benefit.  Just use the lower 32 bits for memory access and you can still use the remaining 32 bits for other things.  Op-codes, feeding registers, whatever.  How optimized the iPhone is to do that I don't really know, but it certainly remains well positioned for future growth.

post #113 of 131
Quote:
Originally Posted by THT View Post
 

 

And when Apple hands over the masks for laying out the transistors and metal layers on the wafer, what is the Samsung fab engineer seeing? They aren't seeing a blueprint. It's doubtful to me that Apple is handing over the circuit design or "blueprint" for the SoC. They are handing over circuit masks to etch the transistors. Apple and Samsung work to make sure the masks are designed correctly for the process involved, but I doubt that Samsung Semi knows at any point time what those masks are building.

 

Samsung would have to steal a wafer and put it under a microscope. Basically the same thing Chipworks did. This is not some day at the park type of activity. It would be act of espionage. Not worth the cost for what you can figure out based on images of the product.

 

You do realize that there are companies that specialize in reverse engineering a chip just by getting a sample of one? They can take the physical chip and by removing layers and imaging the chip they can actually construct a schematic diagram of all the transistors. I went back to Chipworks and found their capabilities are far beyond what I thought. This is exactly what they are capable of doing.

 

When Chipworks looked at the A7 it was more of a publicity stunt. They image the chip to get a basic view of the major components, but stop there. They have the ability to reverse engineer a processor and produce a schematic from it. Of course, they're not going to invest the huge $$$ it would take to completely reverse engineer a processor like the A7 "just for kicks". Simply providing the basic images brings a lot of attention to what they do and what their capabilities are.

 

Why would someone want to reverse engineer a competitors processor? It's not so they can copy. One use is to spot any possible infringement of your own IP in their designs. Another would be to "snoop around" and see where they are at in terms of their technology & expertise in comparison to your own.

 

 

I guarantee you Intel, TSMC, Samsung and other large fabs would have this capability. But why would Samsung need to reverse engineer an A7 when they basically have the schematic already (or an ability to create one from the mask)? If you have a schematic then you know pretty much all you need to determine if it's a 64 bit chip or not.

 

I think it's impossible that Samsung didn't know the A7 was 64 bit. I think the only thing they were in the dark about is the release schedule and iOS 7 already being coded for 64 bit.

post #114 of 131
Quote:
Originally Posted by BigMac2 View Post
 

 

Have you ever seen the 5 inch Dell Streak phablet? There is a limit for how big a mobile phone could be before losing it's mobility usefulness, I'll be embarrassed to use one of those overgrown size phone in public, I don't want to show off my phone in public, I wanted it has discreet as possible.


I like the look of the Galaxy S4 - and the size of the screen

- just not sure I want to jump ship to the world of Android!

post #115 of 131
Quote:
Originally Posted by SolipsismX View Post


But how? I'm still cloudy on how they would know a chip is 32 or 64-bit. We're not talking about a node change which is something that needs to be performed at the fabrication level and is completely up to Samsung but an instruction set of a chip. What about Chipwork's images say "This is 64-bit"? Samsung would know that Apple included 4MB of RAM on the chip and other physical features but I don't see how Samsung would know it's 64-bit just by looking at it.

if they really did know all about this chip itt really makes that part of Samsung uncharacteristically tightlipped and ethical when they could have leaked the news and let Samsung make an announcement when Apple starts ramping up production before the iPhone 5S was announced.

 

It would be pretty implausible for Samsung not to know the A7 was 64-bit.  A big part of device manufacturing is an insane amount of QC and rigid specs on just about every film property imaginable.  Film stress, refractive index, particles.... name it.  Wafers are continually under SEMs, FIBs, and TEMs which will easily show you anything you want to know about a device. 

 

Prior posts about there being 'over a billion transistors on a chip, no way they'd find 64 bits!' are a little laughable.  First of all a transistor pretty much has 3 IO's, (source, gate and drain) so it would be a pretty poor place to look in the first place.  All they'd have to do is look at one register and.... hey look, there's 64 lines going in instead of 32......  Done.

 

I don't believe your scenario of Apple surprisingly quickly flipping a pilot line into production is likely either.  These things take a ton of very specific capital equipment that takes an incredible amount of resources to set up (both in time and money).  If all they had was a pilot line and Apple said 'Ha!  Start making millions of these!  Samsung would probably realistically say, "We can get ramped for that kind of volume in 6-8 months if *everything* goes well" and even that is unrealistically optimistic.

 

I also find your suggestion that Samsung was, for once, being 'ethical' pretty unlikely too.  Ha ha. 

I think they just plain didn't think it was all that noteworthy. PC's have been 64 bit for years, they just didn't go there until they needed to.  Samsung was probably like 'Hey look, this thing is 64 bit, what on earth do they need that for at this stage of the game?'

 

There was no surprise shown even during the manufacturing process- the surprise came after it was released and in consumers hands and Apple brilliantly proclaimed "We are the first to amazing 64-bit!"  And the crowd goes wild!!  And the non-Apple crowd made it even worse by saying 'they really don't need this yet' and kind of scratching their bums in general.  Too late.  Its already a lot like the articles reference to Spinal Tap.  Apple built an amplifier that goes all the way up to 11!!!  Now everyone wants an Amp that goes all the way up to 11!  If you are in the Amp business, your next Amp *better* go up to 11 whether you think its needed or not.

 

64 bit is certainly not far away from being needed and Apple is well poised to continue to improve performance accordingly, but so will the competition.  Who is right, who is wrong?  Obviously on an Apple fansite I'm guessing what most peoples answer is, but with the 'amazing breakthough of 64-bit' is Apples latest offering 2x as fast as Samsung's latest offering?  Nope.  Okay, 50% faster?  Nope.  25%  Nope.  C'mon!  10%???  Nope.  Faster at all??  On some tests, yes, on others, no.   Does that make Samsung amazing because they can build 32-bit devices that are as fast as Apple's 64-bit devices?  Nope.

 

I just hope they both keep on trying to outdo each other =)

post #116 of 131
Quote:
Originally Posted by Frood View Post

It would be pretty implausible for Samsung not to know the A7 was 64-bit.  A big part of device manufacturing is an insane amount of QC and rigid specs on just about every film property imaginable.  Film stress, refractive index, particles.... name it.  Wafers are continually under SEMs, FIBs, and TEMs which will easily show you anything you want to know about a device.

Anything you want to know? Does Foxconn know everything about the iPhone simply because they build it? Why can't Apple's contract with Samsung allow them to use their people to deal with any verification? I've seen it before in other industries in order to limit and compartmentalize what manufactures can see.
Quote:
Prior posts about there being 'over a billion transistors on a chip, no way they'd find 64 bits!' are a little laughable.  First of all a transistor pretty much has 3 IO's, (source, gate and drain) so it would be a pretty poor place to look in the first place.  All they'd have to do is look at one register and.... hey look, there's 64 lines going in instead of 32......  Done.

That goes against what Mike Ash stated. I'll repeat again, "With ARM64, there are 32 integer registers" and "32-bit ARM are […] 32 32-bit floating point registers […] ARM64 simplifies this to 32 128-bit registers.'

Quote:
I don't believe your scenario of Apple surprisingly quickly flipping a pilot line into production is likely either. These things take a ton of very specific capital equipment that takes an incredible amount of resources to set up (both in time and money).  If all they had was a pilot line and Apple said 'Ha!  Start making millions of these!  Samsung would probably realistically say, "We can get ramped for that kind of volume in 6-8 months if *everything* goes well" and even that is unrealistically optimistic.

Huh? Where did I say that? And why does it take a ton of specific capital that would scream it's a 64-bit chip? We're not talking about a smaller node here.
Quote:
I also find your suggestion that Samsung was, for once, being 'ethical' pretty unlikely too.  Ha ha. 
I think they just plain didn't think it was all that noteworthy. PC's have been 64 bit for years, they just didn't go there until they needed to.  Samsung was probably like 'Hey look, this thing is 64 bit, what on earth do they need that for at this stage of the game?'

I find it implausible that it would neither be noteworthy (we're noting it now and it's been talked about since it was announced) and that no one at Samsung would see how this would affect their mobile devices or feel there was anything to be gained from usurping Apple by simply saying, "Hey, the S5 will be 64-bit." This is a company that doped their mobile processors for a 4% gain in performance when certain benchmarks were run.

As I stated to Eric, I am not discounting any possibility but I also haven't read any argument on AI that 1) is convincing to show Samsung would have to know by simply having it's plant utilized regardless of how much control Apple had over the production, and 2) that if they did know they would think it's business worthy to do the same (until after Apple announces it).

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply
post #117 of 131
Quote:
Originally Posted by THT View Post

 

Is this really true? Windows compatibility is likely huge for things like enterprise desktops (where specific applications have been design to help the business run), but in the tablet space, we're talking consumer usage and office automation usage right? In that space, those usages are becoming more and more web-based and office automation is becoming more and more commoditized.

 

There is an asston of enterprise apps that isn't web-based yet.  Enterprise buys tablets as well and there's a certain usefulness if you could have both in one thing instead of having to carry your laptop as well as an iPad.

 

Apple is moving in this direction by making iOS a lot more capable.  Microsoft is trying to move in this direction by making Windows suck less on a tablet and IMHO floundering at it this go around.

 

Still, I'd love an 11" MBA that could become a tablet as needed...

post #118 of 131
Quote:
Originally Posted by EricTheHalfBee View Post
 

 

You do realize that there are companies that specialize in reverse engineering a chip just by getting a sample of one? They can take the physical chip and by removing layers and imaging the chip they can actually construct a schematic diagram of all the transistors. I went back to Chipworks and found their capabilities are far beyond what I thought. This is exactly what they are capable of doing.

 

When Chipworks looked at the A7 it was more of a publicity stunt. They image the chip to get a basic view of the major components, but stop there. They have the ability to reverse engineer a processor and produce a schematic from it. Of course, they're not going to invest the huge $$$ it would take to completely reverse engineer a processor like the A7 "just for kicks". Simply providing the basic images brings a lot of attention to what they do and what their capabilities are.

 

Why would someone want to reverse engineer a competitors processor? It's not so they can copy. One use is to spot any possible infringement of your own IP in their designs. Another would be to "snoop around" and see where they are at in terms of their technology & expertise in comparison to your own.

 

 

I guarantee you Intel, TSMC, Samsung and other large fabs would have this capability. But why would Samsung need to reverse engineer an A7 when they basically have the schematic already (or an ability to create one from the mask)? If you have a schematic then you know pretty much all you need to determine if it's a 64 bit chip or not.

 

I think it's impossible that Samsung didn't know the A7 was 64 bit. I think the only thing they were in the dark about is the release schedule and iOS 7 already being coded for 64 bit.

 

Really don't agree with you that Samsung has any schematics or blueprints that give them any information. When Apple is working with Samsung Semi, Apple's SoC designers are figuring out how to fab the chip, to etch the transistors. Samsung Semi simply had no architecture information to glean from this type of work.

 

If you are playing the espionage angle, then that automatically means Samsung design know anything about the architecture. They are only doing it for "intelligence" purposes and any kind of information gleaned from it automatically means they are a year late. There are also legal risks incumbent in it.

post #119 of 131
Interesting Q&A from an ARM architect on AnandTech.


I won't pretend that I understood any of the heavy questions and answers.

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply

This bot has been removed from circulation due to a malfunctioning morality chip.

Reply
post #120 of 131
Quote:
Originally Posted by Frood View Post
 

 

Their most powerful phone out at the moment is the Note, which still doesn't have a need for (and wouldn't benefit much) from 64-bit architecture.

 

The big 'knob' is memory.  Anything over 4 gig and you absolutely need a 64-bit architecture or you are going to need two clock cycles for one memory address.  Since the Note has 3 gig, 32-bits are just fine, and it beats the 64-bit iPhone in quite a few performance metrics while the iPhone certainly returns the favor and clobbers the Note in other benchmarks.  Both are amazingly zippy devices.

 

I think where a lot of people went wrong on the iPhone is saying that because the iPhone doesn't have 4 gig of memory, it doesn't need a 64-bit bus.  If you go over 4 gig, yes, you absolutely need a 64-bit bus (and Samsung will likely go there once it needs to), but just because you only have one gig doesn't mean you can't benefit.  Just use the lower 32 bits for memory access and you can still use the remaining 32 bits for other things.  Op-codes, feeding registers, whatever.  How optimized the iPhone is to do that I don't really know, but it certainly remains well positioned for future growth.

 

The only performance metric the Note 3 beats the iPhone 5S in is multi-core benchmarks and some GPU benchmarks, and barely at that. Meanwhile, the iPhone 5S's single threaded performance is about 40 to 50% faster. Single threaded performance lifts all boats. There are very few applications that can actually use 4-cores, and even less so on smartphones.

 

And maybe you shouldn't use the word "bus" in the context of 64-bit either. Apple's SoC have used 32, 64 and 128-bit busses already. Register would be the more appropriate word.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: iPhone
AppleInsider › Forums › Mobile › iPhone › Apple's 64-bit A7 SoC 'set off panic' for chipmakers