I'm guessing the 7s, 7s Plus, and 8 will all share the same A11 chip. There isn't really any reason to think otherwise.
It is amazing to think that the 7s/7s Plus will be typical "S" model phones that could have stood alone as worthy upgrades...and in the shadow of the 8 they will essentially just be the "good enough" model for users looking at price over everything else.
Exactly. I would think that the A11 for all of these new phones would be exactly the same. Unless Apple decides to up the speed that it runs at by a bit, if the cooling allows it, to set performance up by a bit for the 8. But probably not.
what I do wonder about is whether the new machine learning chip is ready for use, and if so, whether it will appear in all three phones, or possibly, just the 8.
If the AI and AR chips are in volume production, sure, it will happen for all three models, as these features drive services. There will be plenty of differentiation for the iPhone 8 otherwise.
I would hope so, but we don’t know. It’s possible that Apple would reserve this for the 8 this year, for differentiation, (and remember that these chips are NOT required for ARkit) and next year, could spread throughout the line, including iPad Pro models, and maybe even later, for the $329 iPad as well.
For the GPU, Imagination has already given us a time line. As such i would expect to see Apples new GPU this year in all new Iphones
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
I'm guessing the 7s, 7s Plus, and 8 will all share the same A11 chip. There isn't really any reason to think otherwise.
It is amazing to think that the 7s/7s Plus will be typical "S" model phones that could have stood alone as worthy upgrades...and in the shadow of the 8 they will essentially just be the "good enough" model for users looking at price over everything else.
Exactly. I would think that the A11 for all of these new phones would be exactly the same. Unless Apple decides to up the speed that it runs at by a bit, if the cooling allows it, to set performance up by a bit for the 8. But probably not.
what I do wonder about is whether the new machine learning chip is ready for use, and if so, whether it will appear in all three phones, or possibly, just the 8.
If the AI and AR chips are in volume production, sure, it will happen for all three models, as these features drive services. There will be plenty of differentiation for the iPhone 8 otherwise.
I would hope so, but we don’t know. It’s possible that Apple would reserve this for the 8 this year, for differentiation, (and remember that these chips are NOT required for ARkit) and next year, could spread throughout the line, including iPad Pro models, and maybe even later, for the $329 iPad as well.
For the GPU, Imagination has already given us a time line. As such i would expect to see Apples new GPU this year in all new Iphones
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
With FIFO, Apple could easily add more functionality to the SOC, so that is where I would expect to find the AI / AR / facial recognition die, if it happens at all. There might be something new on the actual A11 die as well, but I wouldn't have a clue what that might be.
I'm guessing the 7s, 7s Plus, and 8 will all share the same A11 chip. There isn't really any reason to think otherwise.
It is amazing to think that the 7s/7s Plus will be typical "S" model phones that could have stood alone as worthy upgrades...and in the shadow of the 8 they will essentially just be the "good enough" model for users looking at price over everything else.
Exactly. I would think that the A11 for all of these new phones would be exactly the same. Unless Apple decides to up the speed that it runs at by a bit, if the cooling allows it, to set performance up by a bit for the 8. But probably not.
what I do wonder about is whether the new machine learning chip is ready for use, and if so, whether it will appear in all three phones, or possibly, just the 8.
If the AI and AR chips are in volume production, sure, it will happen for all three models, as these features drive services. There will be plenty of differentiation for the iPhone 8 otherwise.
I would hope so, but we don’t know. It’s possible that Apple would reserve this for the 8 this year, for differentiation, (and remember that these chips are NOT required for ARkit) and next year, could spread throughout the line, including iPad Pro models, and maybe even later, for the $329 iPad as well.
For the GPU, Imagination has already given us a time line. As such i would expect to see Apples new GPU this year in all new Iphones
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
What we do know is that Apple has stated that they stopped using new Imagination IP after 2015. So the latest Apple GPUs still are based on that older 2015 Imagination tech. As Anandtech has mentioned during the past two years, is that Apple’s GPUs don’t exactly match up with the Imagination specs, and that from what little they can glean fron microphotos of the chip, they don’t exactly look like their tech either. Anandtech has been speculating that Apple has been using more of their own tech in the GPUs for some time, replacing parts of the GPUs with their own versions.
i wonder what performance would have been if Apple did use their latest (much improved) IP, along with their own. But I think it’s a business decision to stick with the older tech, to make the business easier to unwind.
so what we’ve read is that Apple will stop using their tech in a timeline that now, at this later date, is anywhere from 12 months to 21 months away. That doesn’t include new chips this year. So while Apple is likely to have even more of their own stuff in the new GPU, I doubt that it will be completely replaced this September. And from what Apple said, they had a new machine learning chip. That doesn’t sound like it’s part of the SoC. As far as AR developments in the GPU, again, that’s speculation by technical writers, wondering why Apple would leave the company producing the best mobile GPU designs, even though they’ve been having problems attracting other large OEMs other than Apple.
so, this is still very confusing. I’m sure, that with speculation around every side of this, someone will be right, or partly right, just out of luck of the draw. But I’m not willing to speculate myself which argument it will be. This is one thing that Apple is very good at hiding.
Anyway, time is getting short. It’s just about 4 weeks away before we know everything.
edit: I just reread this post, and I’ve come up with an idea that just may allow completely Apple GPUs this year. It’s possible that when talking about that somewhat far off timeline, it could mean that Apple’s older devices will still be using that older IP until they’re discontinued. In that case, it becomes very possible that the new GPU will be an all Apple model. But it also leaves a question. The oldest model in the line should be around at least three more years, likely four. But if Apple won’t be using any more of their IP in two years, where does that leave the older devices? Does that mean that Apple will be discontinuing them earlier than usual? If so, as these tend to be the oldest, or older models (other than the SE, which might be upgraded later this year, or early next year), does that mean that Apple is preparing new models that are less expensive from the beginning, with their own GPU?
I'm guessing the 7s, 7s Plus, and 8 will all share the same A11 chip. There isn't really any reason to think otherwise.
It is amazing to think that the 7s/7s Plus will be typical "S" model phones that could have stood alone as worthy upgrades...and in the shadow of the 8 they will essentially just be the "good enough" model for users looking at price over everything else.
Exactly. I would think that the A11 for all of these new phones would be exactly the same. Unless Apple decides to up the speed that it runs at by a bit, if the cooling allows it, to set performance up by a bit for the 8. But probably not.
what I do wonder about is whether the new machine learning chip is ready for use, and if so, whether it will appear in all three phones, or possibly, just the 8.
If the AI and AR chips are in volume production, sure, it will happen for all three models, as these features drive services. There will be plenty of differentiation for the iPhone 8 otherwise.
I would hope so, but we don’t know. It’s possible that Apple would reserve this for the 8 this year, for differentiation, (and remember that these chips are NOT required for ARkit) and next year, could spread throughout the line, including iPad Pro models, and maybe even later, for the $329 iPad as well.
For the GPU, Imagination has already given us a time line. As such i would expect to see Apples new GPU this year in all new Iphones
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
With FIFO, Apple could easily add more functionality to the SOC, so that is where I would expect to find the AI / AR / facial recognition die, if it happens at all. There might be something new on the actual A11 die as well, but I wouldn't have a clue what that might be.
The one thing that has intrigued me is the size of the A10x die. The A9x die at 16nm, is about 145 square mm. The A10X die at 10nm, is about 96 square mm.
this is doubly interesting. One is simply because it’s so much smaller. It’s a LOT smaller. And two, because in this much smaller die, they added another high performance and another efficiency core.
its true that this is very possibly the first 10nm part TSMC has made. As such, you don’t want to go too complex. But that leaves open the possibility that the A11 will also be much smaller. If so, then Apple can do a lot with this chip if they decide to not go much smaller with the die, but instead, to,add to what’s on it.
145 square mm is 2.3 times the area of 96 square mm. That’s absolutely huge. Apple can fit two A10x SoCs on that die, and have extra space. The difference between the A10 and the A11 should be the same, percentage wise.
I'm guessing the 7s, 7s Plus, and 8 will all share the same A11 chip. There isn't really any reason to think otherwise.
It is amazing to think that the 7s/7s Plus will be typical "S" model phones that could have stood alone as worthy upgrades...and in the shadow of the 8 they will essentially just be the "good enough" model for users looking at price over everything else.
Exactly. I would think that the A11 for all of these new phones would be exactly the same. Unless Apple decides to up the speed that it runs at by a bit, if the cooling allows it, to set performance up by a bit for the 8. But probably not.
what I do wonder about is whether the new machine learning chip is ready for use, and if so, whether it will appear in all three phones, or possibly, just the 8.
If the AI and AR chips are in volume production, sure, it will happen for all three models, as these features drive services. There will be plenty of differentiation for the iPhone 8 otherwise.
I would hope so, but we don’t know. It’s possible that Apple would reserve this for the 8 this year, for differentiation, (and remember that these chips are NOT required for ARkit) and next year, could spread throughout the line, including iPad Pro models, and maybe even later, for the $329 iPad as well.
For the GPU, Imagination has already given us a time line. As such i would expect to see Apples new GPU this year in all new Iphones
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
With FIFO, Apple could easily add more functionality to the SOC, so that is where I would expect to find the AI / AR / facial recognition die, if it happens at all. There might be something new on the actual A11 die as well, but I wouldn't have a clue what that might be.
The one thing that has intrigued me is the size of the A10x die. The A9x die at 16nm, is about 145 square mm. The A10X die at 10nm, is about 96 square mm.
this is doubly interesting. One is simply because it’s so much smaller. It’s a LOT smaller. And two, because in this much smaller die, they added another high performance and another efficiency core.
its true that this is very possibly the first 10nm part TSMC has made. As such, you don’t want to go too complex. But that leaves open the possibility that the A11 will also be much smaller. If so, then Apple can do a lot with this chip if they decide to not go much smaller with the die, but instead, to,add to what’s on it.
145 square mm is 2.3 times the area of 96 square mm. That’s absolutely huge. Apple can fit two A10x SoCs on that die, and have extra space. The difference between the A10 and the A11 should be the same, percentage wise.
"That's absolutely huge"
Gee, I wonder what somebody could do with 145 square mm in a Mac Book form factor?
Why is Apple trying to release the 7s and 8 in the same year?
I think the reason is that Apple knows that there best and safest bet is to stick with "what works". Refining an already best seller is the best way to maximize ROI. They can increase the power, battery life and screen quality in the iPhone 7S and it would be a great upgrade. They may even be able to get some of the 7 Plus features into the standard-sized 7S. That alone would be wonderful.
But to do something extraordinary, they need to leap and that involves risk and expense. The iPhone 8 has probably been in the works for years and may be why things have become rather sedate of late. But to bound, that means new technology at lower yields and higher prices. So I expect that Apple will announce the 7S and it will be great. But then they will announce "one more thing" and show off the next evolution in iPhone. This insulates them to a degree by making the iPhone 8 ultra-premium and desired at a higher price in order to reduce inventory shock. I would expect the iPhone 7S to be available "today" and the iPhone 8 will be available in mid-October. I expect that iPhone 7S orders will be shipped promptly and be a huge windfall for the company. iPhone 8 demand will likely overrun supply and push delays into 2018. I expect the iPhone 8 to be revolutionary and may even require developers to make some additional leaps. I would suspect that the iPhone 8 will do something new with AR that brings it from the fringe of society into the mainstream similar to what the iPod did for digital music. It will also be a sharp deviation from the standard iPhone formula by dropping TouchID for PearlID, full screen, intuitive 3D gesture systems, etc.
I think it is certainly possible the OLED screen will leave Apple power budget for enhancing the CPU clock speed a bit.
Ah, gee, here we go again.
Look, it’s not true that OLEDs are more efficient. The OLED manufacturers keep saying that. The manufacturers who use them keep saying that, but there’s no evidence proving that.
the whole thing starts with the concept that there’s no power being used when a pixel is black (actually, a tiny amount is used). But power draw goes up as pixels get whiter. At a point at about 30 to 35% grey level, the draw between the most modern OLEDs and LCDs are about the same. But while LCD power draw remains the same at all brightness levels, OLED power draw becomes much higher than LCD power draw at maximum brightness. That’s why most OLED based GUIs are black with white text - to save power.
Comments
How much of the new AI tech that will be done on the new GPU is unknown. There have been various rumors some indicating a separate chip but i kinda doubt that would happen in iPhone. It isnt impossible but i suspect that Apple would prefer integration into the SoC. What might be a factor though is an AI chip that requires a different process than FinFET. In any event integration saves power vs. performance. In effect what im saying is that iPhone could see two different processors this year, one with and one without additional AI support.
i wonder what performance would have been if Apple did use their latest (much improved) IP, along with their own. But I think it’s a business decision to stick with the older tech, to make the business easier to unwind.
so what we’ve read is that Apple will stop using their tech in a timeline that now, at this later date, is anywhere from 12 months to 21 months away. That doesn’t include new chips this year. So while Apple is likely to have even more of their own stuff in the new GPU, I doubt that it will be completely replaced this September. And from what Apple said, they had a new machine learning chip. That doesn’t sound like it’s part of the SoC. As far as AR developments in the GPU, again, that’s speculation by technical writers, wondering why Apple would leave the company producing the best mobile GPU designs, even though they’ve been having problems attracting other large OEMs other than Apple.
so, this is still very confusing. I’m sure, that with speculation around every side of this, someone will be right, or partly right, just out of luck of the draw. But I’m not willing to speculate myself which argument it will be. This is one thing that Apple is very good at hiding.
Anyway, time is getting short. It’s just about 4 weeks away before we know everything.
edit: I just reread this post, and I’ve come up with an idea that just may allow completely Apple GPUs this year. It’s possible that when talking about that somewhat far off timeline, it could mean that Apple’s older devices will still be using that older IP until they’re discontinued. In that case, it becomes very possible that the new GPU will be an all Apple model. But it also leaves a question. The oldest model in the line should be around at least three more years, likely four. But if Apple won’t be using any more of their IP in two years, where does that leave the older devices? Does that mean that Apple will be discontinuing them earlier than usual? If so, as these tend to be the oldest, or older models (other than the SE, which might be upgraded later this year, or early next year), does that mean that Apple is preparing new models that are less expensive from the beginning, with their own GPU?
The one thing that has intrigued me is the size of the A10x die. The A9x die at 16nm, is about 145 square mm. The A10X die at 10nm, is about 96 square mm.
this is doubly interesting. One is simply because it’s so much smaller. It’s a LOT smaller. And two, because in this much smaller die, they added another high performance and another efficiency core.
its true that this is very possibly the first 10nm part TSMC has made. As such, you don’t want to go too complex. But that leaves open the possibility that the A11 will also be much smaller. If so, then Apple can do a lot with this chip if they decide to not go much smaller with the die, but instead, to,add to what’s on it.
145 square mm is 2.3 times the area of 96 square mm. That’s absolutely huge. Apple can fit two A10x SoCs on that die, and have extra space. The difference between the A10 and the A11 should be the same, percentage wise.
Gee, I wonder what somebody could do with 145 square mm in a Mac Book form factor?
Or how about a 15 inch iPad Studio Pro at 5K?
But to do something extraordinary, they need to leap and that involves risk and expense. The iPhone 8 has probably been in the works for years and may be why things have become rather sedate of late. But to bound, that means new technology at lower yields and higher prices. So I expect that Apple will announce the 7S and it will be great. But then they will announce "one more thing" and show off the next evolution in iPhone. This insulates them to a degree by making the iPhone 8 ultra-premium and desired at a higher price in order to reduce inventory shock. I would expect the iPhone 7S to be available "today" and the iPhone 8 will be available in mid-October. I expect that iPhone 7S orders will be shipped promptly and be a huge windfall for the company. iPhone 8 demand will likely overrun supply and push delays into 2018. I expect the iPhone 8 to be revolutionary and may even require developers to make some additional leaps. I would suspect that the iPhone 8 will do something new with AR that brings it from the fringe of society into the mainstream similar to what the iPod did for digital music. It will also be a sharp deviation from the standard iPhone formula by dropping TouchID for PearlID, full screen, intuitive 3D gesture systems, etc.
Look, it’s not true that OLEDs are more efficient. The OLED manufacturers keep saying that. The manufacturers who use them keep saying that, but there’s no evidence proving that.
the whole thing starts with the concept that there’s no power being used when a pixel is black (actually, a tiny amount is used). But power draw goes up as pixels get whiter. At a point at about 30 to 35% grey level, the draw between the most modern OLEDs and LCDs are about the same. But while LCD power draw remains the same at all brightness levels, OLED power draw becomes much higher than LCD power draw at maximum brightness. That’s why most OLED based GUIs are black with white text - to save power.