Hello comunity,
I think all of you are familiar with intel’s tick tock strategy for CPU releases. I think we could identify a similar pattern with Apple SoC which became quite a topic over last few years. I see this as ticks being focused on energy efficiency and die size on even A chips (4,6,8) which went into drastically redesigned much thinner and sleeker products, while odd ones (5,7,9) went all out on performance on an already established form factor.
It’s obvious that reality is way more complicated than this simple asumption - like process node availability and performance being big factor here which Apple can’t influence, or the fact that no matter how much “efficiency” was gained over generations, peak power compustion just went up and up.
After A9 Apple was forced to stay on same node, which caused size and power consumption go up as Apple really wanted to push high in performance. It’s been so “bad” in this regard that the GPU in A10 supposedly can’t sustain peak performance even in the 6th gen iPad ( a bit bigger than iPhone !).
Then 10nm came which I would describe as better node than expected. Along with supposed 35 % reduced power, all new architectures were aded on all levels which resulted in hugely impressive upgrade - A11 ( a bit of a clock shuttering isn’t it ?)
And now what A12 is supposed to be ?
12 for TOCK
- New 7nm node, which is promised to increase density by 60 percent over already dense 10nm. And of course supposed 40% less power (isoperf) which if true would be really big.
- Second gen Apple designs just crush it... Remember first Apple CPU’s ? Swift and Cyclone ? Shall we expect a similar uptake in performance for first and second gen Apple GPU ?
- 4 GB of Ram, with Apple kind of managment could keep whole folders of apps open.
- Only 6 inch iPhones this year - While A11 was happy in it’s 4,7” housing going bigger obviously gives room for more heat and battery resources.
12 for TICK
- New 7nm node. You know shrinking so much in 24 moths could end up in quite embarasing problems. Especially with such a high density, hotspots can appear limiting clock speed increases.
- No improvement to memory bandwidth. We have LPDDR4 since 2015 and chips are getting starved badly. What could possibly solve this problem ?
- 7nm being a two year node. Why not just take a break after such a great shrink as A11 ? There is another year left to play with 7nm.
- A12X. Apple suposedly focused to deliver A11 without making much effort for their iPad chips. With “A12X” rumored to target notebook devices and performance it’s obvious phones will get behind a bit.
For me personally considerng just how powerfull A11 already is, I would prefer a leaner more power efficient design - bettery batery life being a supreme goal of course. But Apple might have different plans.
What’s your opinion ?
I think all of you are familiar with intel’s tick tock strategy for CPU releases. I think we could identify a similar pattern with Apple SoC which became quite a topic over last few years. I see this as ticks being focused on energy efficiency and die size on even A chips (4,6,8) which went into drastically redesigned much thinner and sleeker products, while odd ones (5,7,9) went all out on performance on an already established form factor.
It’s obvious that reality is way more complicated than this simple asumption - like process node availability and performance being big factor here which Apple can’t influence, or the fact that no matter how much “efficiency” was gained over generations, peak power compustion just went up and up.
After A9 Apple was forced to stay on same node, which caused size and power consumption go up as Apple really wanted to push high in performance. It’s been so “bad” in this regard that the GPU in A10 supposedly can’t sustain peak performance even in the 6th gen iPad ( a bit bigger than iPhone !).
Then 10nm came which I would describe as better node than expected. Along with supposed 35 % reduced power, all new architectures were aded on all levels which resulted in hugely impressive upgrade - A11 ( a bit of a clock shuttering isn’t it ?)
And now what A12 is supposed to be ?
12 for TOCK
- New 7nm node, which is promised to increase density by 60 percent over already dense 10nm. And of course supposed 40% less power (isoperf) which if true would be really big.
- Second gen Apple designs just crush it... Remember first Apple CPU’s ? Swift and Cyclone ? Shall we expect a similar uptake in performance for first and second gen Apple GPU ?
- 4 GB of Ram, with Apple kind of managment could keep whole folders of apps open.
- Only 6 inch iPhones this year - While A11 was happy in it’s 4,7” housing going bigger obviously gives room for more heat and battery resources.
12 for TICK
- New 7nm node. You know shrinking so much in 24 moths could end up in quite embarasing problems. Especially with such a high density, hotspots can appear limiting clock speed increases.
- No improvement to memory bandwidth. We have LPDDR4 since 2015 and chips are getting starved badly. What could possibly solve this problem ?
- 7nm being a two year node. Why not just take a break after such a great shrink as A11 ? There is another year left to play with 7nm.
- A12X. Apple suposedly focused to deliver A11 without making much effort for their iPad chips. With “A12X” rumored to target notebook devices and performance it’s obvious phones will get behind a bit.
For me personally considerng just how powerfull A11 already is, I would prefer a leaner more power efficient design - bettery batery life being a supreme goal of course. But Apple might have different plans.
What’s your opinion ?