I agree this all about getting the process issues ironed out. But it doesn't look like it will be Cortex-A9-based.
I'd like to think it will be A15, but has the iPhone ever been ahead of the iPad in CPU? I think it will be a 32nm A5X, then the next iPad will be first to A6 with Cortex A15 cores.
When is the class action regarding this faulty Core ?
I'm sure there are some bottom-feeding attorneys fishing for consumers that will cry about being harmed by this.
If they advertise it as single core, and they're using a binned dual-core with the defective core disabled I could care less.
The only time I would have a problem with this are like the stunts that IBM pulls all the time with their mid-range platforms (iSeries, AS/400, etc..) where they would install fully-function multi-core CPU's in all their machines but software lock the extra cores which would then require an additional fee to "activate" each individual core.
I got into arguments with IBM over the years on this. They would spin their stories as "providing only the performance tailored to the customer" or other BS and I would turn around and tell them that they are selling me a fully-functional V8 motor with two of the spark plug wires pulled out.
If Intel pulled a stunt like IBM does, the market would riot.
I'm curious to see what Apple's response would be. Not that having extra cores would necessarily help stream a 1080p video any faster.
I'd like to think it will be A15, but has the iPhone ever been ahead of the iPad in CPU? I think it will be a 32nm A5X, then the next iPad will be first to A6 with Cortex A15 cores.
Three things:
1) The iPad's SoC has been ahead of the iPhone thus far, but this year is unusual as there was no 32nm process or Cortex-A15 that was ready for the iPad's release. Both of those are looking to be possible for Autumn.
2) The iPad had never been ahead of the iPhone in cellular connectivity until the iPad (3) despite having more room from the start for larger and more cellular chips, so it's possible that things will change as components permit or market pressure forces Apple's hand.
3) I don't know of any new Img Tech GPUs that could lower the power envelope but I don't think they will go with the 4 core system, which I think you are suggesting with the 'X' usage. The 4 cores were only needed for the 2048x1536 display. Maybe they will, but a much slower speeds but I'd think that two, higher performing cores make more sense.
I'm sure there are some bottom-feeding attorneys fishing for consumers that will cry about being harmed by this.
If they advertise it as single core, and they're using a binned dual-core with the defective core disabled I could care less.
<...>
I got into arguments with IBM over the years on this. <..>
Interesting point, which probably explains why Apple advertises it as a single Core. But by doing so, they are outside of specifications, and guilty of lying ! ----> in either cases, there is room for suing Apple !
Or really manufacturing in general. No process is absolutely perfect.
Quote:
Originally Posted by tipoo
And there ya go. Some people got upset for some reason in articles where myself and some others suggested it was die-harvested. Its a normal practice, no reason for Apple not to do it.
It is actually normal in many industries, that is one often has processes that seldom yield 100% if ever.
Quote:
More interesting is that this is 32nm rather than 45 like the A5X,
This is huge and totally ignored by the article. It means a lower power processor for the next series of Touches and iPhones. At 32 nm they would also have the option of cranking up the clock rate without penalty.
Quote:
I wonder if that means a silently updated 2012 iPad with lower power consuming 32nm chips.
I don't think there is a chance in hell. Maybe not for the expected reasons either. Apples processor consumption is such that they will likely need to employ multiple plants and processes to meet demand. Two years ago they where using 80% of Samsungs production capacity. Today I suspect that they are using all of Samsungs existing capacity plus the capacity of the new Texas plant. Thus the need to spread production across plants and processes.
In any event this does give us a ball park figure to work with. 32nm should allow Apple to hit 1.6GHz easy or to beef up the GPU in the next iPhone. It would be nice to have more detail with respect to this processor, especially if it is a simple die shrink or in some way enhanced.
They are ramping a brand new process in a brand new plant. Smart move by any measure.
Quote:
Originally Posted by SolipsismX
They are testing a new new process node with a low-volume product. I don't think we've seen Apple do that before, but it's also not surprising that they would as 32/28nm is expected to be in volume for the 6th gen iPhone.
Having one core going along for a free ride doesn't make much sense to me, at all . I can only trust that Apple knows what it is doing or that the second core can be turned on if needed. So, it's possible the theory of this A5 processor being a reject processor might be wrong. I'd like to see if someone can turn on that second core to prove it's functional. Apple is doing things that are hard to fathom. If the Apple TV sells for $99, why disable one core when two can be put in use. Is it possible there might be a heat issue? I would think that it would make more sense to use both of those cores and downclock both cores. I honestly tend to believe it's a yield problem with one core possibly not up to spec, but what do I know. I doubt that any Android vendor would ever consider disabling a working core because for them it's all about specs.
Interesting point, which probably explains why Apple advertises it as a single Core. But by doing so, they are outside of specifications, and guilty of lying ! ----> in either cases, there is room for suing Apple !
The Apple haters would jump all over this. Apple is cheating consumers with defective or reject components. The only thing is that if the device performs as specified, there's nothing for anyone to complain about. I just don't like that AppleTV doesn't support Hulu Plus but that has nothing to do with the hardware aspect.
Having one core going along for a free ride doesn't make much sense to me, at all . I can only trust that Apple knows what it is doing or that the second core can be turned on if needed. So, it's possible the theory of this A5 processor being a reject processor might be wrong. I'd like to see if someone can turn on that second core to prove it's functional. Apple is doing things that are hard to fathom. If the Apple TV sells for $99, why disable one core when two can be put in use. Is it possible there might be a heat issue? I would think that it would make more sense to use both of those cores and downclock both cores. I honestly tend to believe it's a yield problem with one core possibly not up to spec, but what do I know. I doubt that any Android vendor would ever consider disabling a working core because for them it's all about specs.
You might reconsider your thoguhts after actually reading the article and the previous comments.
And there ya go. Some people got upset for some reason in articles where myself and some others suggested it was die-harvested. Its a normal practice, no reason for Apple not to do it.
More interesting is that this is 32nm rather than 45 like the A5X, I wonder if that means a silently updated 2012 iPad with lower power consuming 32nm chips.
...but... but... but... I paid for two fully functioning cores. So, can I get half my money back?
My god man it couldn't be any clearer than has been explained in this thread. It is common practice in the industry to bin parts. Is that difficult to understand?
Quote:
Originally Posted by Constable Odo
Having one core going along for a free ride doesn't make much sense to me, at all .
Give me a break, this isn't rocket science at all.
Quote:
I can only trust that Apple knows what it is doing or that the second core can be turned on if needed. So, it's possible the theory of this A5 processor being a reject processor might be wrong.
It isn't a reject it is simply a different part number. It is no different than Intel baking a wafer and getting a range of processors that meet different performance levels. Or for that matter Intel turning off defective core and selling the chips as Celerons or whatever. Ever semiconductor manufacture bins parts, they always have and always will.
Quote:
I'd like to see if someone can turn on that second core to prove it's functional. Apple is doing things that are hard to fathom.
This is no time to be dense.
[quote[If the Apple TV sells for $99, why disable one core when two can be put in use. Is it possible there might be a heat issue? I would think that it would make more sense to use both of those cores and downclock both cores. I honestly tend to believe it's a yield problem with one core possibly not up to spec, but what do I know. I doubt that any Android vendor would ever consider disabling a working core because for them it's all about specs. [/QUOTE]
I'm mystified here as to your mental processes here. Honestly you should look for a job with the department of justice you would fit in with those idiots just fine.
Do we know if this is a Samsung chip? Were they planning for 32nm? If so, could this be from the new Texas plant? Texas is new and perhaps they are still refining the process and getting a high reject rate.
If Apple sells 3 million of these this year, and they are the "rejects," what does that imply about the number of non-rejected parts? Could that tell us anything about the product they will go into (for example, probably not the rumored iTV, because the numbers are so larger)?
If these chips are the rejects from the next-gen iPhone chips, isn't it kind of early for an October release, for Apple to be producing chips in quantity enough to get all these rejects?
I'm not surprised they are binning the A5 CPU's, and I'm not surprised they are using the dual-core rejects as single core in the Apple TV. All makes perfect sense. Wat I do find newsworthy is that apparently, Apple is using the Apple TV as some kind of test bed for 32 nm parts.
If true, I don't think this is very common in the semiconductor industry. Usually foundries like to optimize every process node to the maximum, and take big and discrete steps to newer process nodes. This means production runs of distinct designs will always be on the same node, anything else is R&D only. The switch is flipped when R&D yields satisfy some criterium, after which all production moves to the new node.
Apple is now using (and probably will be using of some time to come) variants of the A5 in almost every iOS device they sell, so if the can have the luxury to move it to 32 nm, with reduced risk because try can use dual-core rejects in the Apple TV, this could be a very smart move.
This is an assumption on the parts of ChipWorks. It may well be correct, or not. One problem with this premise - it is based on another assumption: that Apple is producing millions of smaller versions of A5 (APL2498) and are picking off some of the defective ones to use in AppleTV. So Apple is already manufacturing this 32 nm A5 in sufficient volume that one specific defective species is sufficiently large in quantity as to supply another product line? Pretty big assumption.
What you are saying is also right and I think no one but Apple and the vendor that fabricated them know about it.
However, if I am not mistaken, TSMC is the supplier of these devices. Based on the experience that I have had with TSMC their yield is usually low! and we were not even working at 32nm! Their yield should be much lower at 32nm but if they can recycle it like the way they have done it, then it would be worthwhile!
Comments
I agree this all about getting the process issues ironed out. But it doesn't look like it will be Cortex-A9-based.
I'd like to think it will be A15, but has the iPhone ever been ahead of the iPad in CPU? I think it will be a 32nm A5X, then the next iPad will be first to A6 with Cortex A15 cores.
When is the class action regarding this faulty Core ?
I'm sure there are some bottom-feeding attorneys fishing for consumers that will cry about being harmed by this.
If they advertise it as single core, and they're using a binned dual-core with the defective core disabled I could care less.
The only time I would have a problem with this are like the stunts that IBM pulls all the time with their mid-range platforms (iSeries, AS/400, etc..) where they would install fully-function multi-core CPU's in all their machines but software lock the extra cores which would then require an additional fee to "activate" each individual core.
I got into arguments with IBM over the years on this. They would spin their stories as "providing only the performance tailored to the customer" or other BS and I would turn around and tell them that they are selling me a fully-functional V8 motor with two of the spark plug wires pulled out.
If Intel pulled a stunt like IBM does, the market would riot.
I'm curious to see what Apple's response would be. Not that having extra cores would necessarily help stream a 1080p video any faster.
I'd like to think it will be A15, but has the iPhone ever been ahead of the iPad in CPU? I think it will be a 32nm A5X, then the next iPad will be first to A6 with Cortex A15 cores.
Three things:
1) The iPad's SoC has been ahead of the iPhone thus far, but this year is unusual as there was no 32nm process or Cortex-A15 that was ready for the iPad's release. Both of those are looking to be possible for Autumn.
2) The iPad had never been ahead of the iPhone in cellular connectivity until the iPad (3) despite having more room from the start for larger and more cellular chips, so it's possible that things will change as components permit or market pressure forces Apple's hand.
3) I don't know of any new Img Tech GPUs that could lower the power envelope but I don't think they will go with the 4 core system, which I think you are suggesting with the 'X' usage. The 4 cores were only needed for the 2048x1536 display. Maybe they will, but a much slower speeds but I'd think that two, higher performing cores make more sense.
how long till someone in the jailbreak community tries to enable the other core?
3...2...1...
How long till someone in the jailbreak community tries to enable the other core?
They can try, and only a few will succeed.
I'm sure there are some bottom-feeding attorneys fishing for consumers that will cry about being harmed by this.
If they advertise it as single core, and they're using a binned dual-core with the defective core disabled I could care less.
<...>
I got into arguments with IBM over the years on this. <..>
Interesting point, which probably explains why Apple advertises it as a single Core. But by doing so, they are outside of specifications, and guilty of lying ! ----> in either cases, there is room for suing Apple !
is for example plans for external or internal embedding.
Internal=put it inside A own or others future products, ie TVs, screens/projtrs, desktops/laptops
External=sold as external add-on to retrofit to customers existing prod's as above
think "oversized HDMI plug w sneak cable to TV/screen/proj USB outlet for power
add antennas for WiFi & Bluetooth protruding from HDMIplug
"IOS enable your screen for 50bucks"
as precursor to build subscriberbase and momentum for "the real thing"
Would allow smarter Airplay enabling also for high-end audio gear
If you can do robust,smart & min overall footprint it has enormous impact on
reduced cost of acquiring new consumers and "convienience ball&chain" for existing
customers.
"Think big, start "small" "
And there ya go. Some people got upset for some reason in articles where myself and some others suggested it was die-harvested. Its a normal practice, no reason for Apple not to do it.
It is actually normal in many industries, that is one often has processes that seldom yield 100% if ever.
More interesting is that this is 32nm rather than 45 like the A5X,
This is huge and totally ignored by the article. It means a lower power processor for the next series of Touches and iPhones. At 32 nm they would also have the option of cranking up the clock rate without penalty.
I wonder if that means a silently updated 2012 iPad with lower power consuming 32nm chips.
I don't think there is a chance in hell. Maybe not for the expected reasons either. Apples processor consumption is such that they will likely need to employ multiple plants and processes to meet demand. Two years ago they where using 80% of Samsungs production capacity. Today I suspect that they are using all of Samsungs existing capacity plus the capacity of the new Texas plant. Thus the need to spread production across plants and processes.
In any event this does give us a ball park figure to work with. 32nm should allow Apple to hit 1.6GHz easy or to beef up the GPU in the next iPhone. It would be nice to have more detail with respect to this processor, especially if it is a simple die shrink or in some way enhanced.
They are testing a new new process node with a low-volume product. I don't think we've seen Apple do that before, but it's also not surprising that they would as 32/28nm is expected to be in volume for the 6th gen iPhone.
Two cores. Only 1 in use. Very anthropomorphic.
Having one core going along for a free ride doesn't make much sense to me, at all
Interesting point, which probably explains why Apple advertises it as a single Core. But by doing so, they are outside of specifications, and guilty of lying ! ----> in either cases, there is room for suing Apple !
The Apple haters would jump all over this. Apple is cheating consumers with defective or reject components. The only thing is that if the device performs as specified, there's nothing for anyone to complain about. I just don't like that AppleTV doesn't support Hulu Plus but that has nothing to do with the hardware aspect.
Having one core going along for a free ride doesn't make much sense to me, at all
You might reconsider your thoguhts after actually reading the article and the previous comments.
And there ya go. Some people got upset for some reason in articles where myself and some others suggested it was die-harvested. Its a normal practice, no reason for Apple not to do it.
More interesting is that this is 32nm rather than 45 like the A5X, I wonder if that means a silently updated 2012 iPad with lower power consuming 32nm chips.
...but... but... but... I paid for two fully functioning cores. So, can I get half my money back?
Having one core going along for a free ride doesn't make much sense to me, at all
Give me a break, this isn't rocket science at all.
I can only trust that Apple knows what it is doing or that the second core can be turned on if needed. So, it's possible the theory of this A5 processor being a reject processor might be wrong.
It isn't a reject it is simply a different part number. It is no different than Intel baking a wafer and getting a range of processors that meet different performance levels. Or for that matter Intel turning off defective core and selling the chips as Celerons or whatever. Ever semiconductor manufacture bins parts, they always have and always will.
I'd like to see if someone can turn on that second core to prove it's functional. Apple is doing things that are hard to fathom.
This is no time to be dense.
[quote[If the Apple TV sells for $99, why disable one core when two can be put in use. Is it possible there might be a heat issue? I would think that it would make more sense to use both of those cores and downclock both cores. I honestly tend to believe it's a yield problem with one core possibly not up to spec, but what do I know. I doubt that any Android vendor would ever consider disabling a working core because for them it's all about specs.
I'm mystified here as to your mental processes here. Honestly you should look for a job with the department of justice you would fit in with those idiots just fine.
If Apple sells 3 million of these this year, and they are the "rejects," what does that imply about the number of non-rejected parts? Could that tell us anything about the product they will go into (for example, probably not the rumored iTV, because the numbers are so larger)?
If these chips are the rejects from the next-gen iPhone chips, isn't it kind of early for an October release, for Apple to be producing chips in quantity enough to get all these rejects?
If true, I don't think this is very common in the semiconductor industry. Usually foundries like to optimize every process node to the maximum, and take big and discrete steps to newer process nodes. This means production runs of distinct designs will always be on the same node, anything else is R&D only. The switch is flipped when R&D yields satisfy some criterium, after which all production moves to the new node.
Apple is now using (and probably will be using of some time to come) variants of the A5 in almost every iOS device they sell, so if the can have the luxury to move it to 32 nm, with reduced risk because try can use dual-core rejects in the Apple TV, this could be a very smart move.
This is an assumption on the parts of ChipWorks. It may well be correct, or not. One problem with this premise - it is based on another assumption: that Apple is producing millions of smaller versions of A5 (APL2498) and are picking off some of the defective ones to use in AppleTV. So Apple is already manufacturing this 32 nm A5 in sufficient volume that one specific defective species is sufficiently large in quantity as to supply another product line? Pretty big assumption.
What you are saying is also right and I think no one but Apple and the vendor that fabricated them know about it.
However, if I am not mistaken, TSMC is the supplier of these devices. Based on the experience that I have had with TSMC their yield is usually low! and we were not even working at 32nm! Their yield should be much lower at 32nm but if they can recycle it like the way they have done it, then it would be worthwhile!