Editorial: As Apple A13 Bionic rises, Samsung Exynos scales back its silicon ambitions

2»

Comments

  • Reply 21 of 34

    How many affordable electric car makers are giving Tesla any real competition?
    Define 'affordable' ?
    What is affordable to you may well not be for me.
    Next year, we... well at least in Europe will see a good number of BEV's that are much more affordable for the majority of people than even the lowest priced Tesla Model 3.
    These will be produced in Volume. The VW ID.3 is just the start. The new Zoe, Opel/Vauxhall Corsa and Peugeot e208 (the latter two share the same platform) are just the start.
    Again, u are failing to understand the point, look at what Tesla has done!! The research that’s gone into their factories, heck the amounts of money they are BURNING!!! daily!
    u think Vauxhall, VW can replicate that?
    Tesla factories are not normal car factory conversions. These are built from the ground up.
    The software development and engineering and how it’s executed matters.
    Charging networks, Servicing networks, these are all things Tesla is BLOWING BILLIONS on.
    To folks, these things don’t important but that’s the difference between Tesla everyone one else.
    i think the comparison is on point. The actions Apple has taken with its Chip ambition are identical to what Tesla is doing to differentiate itself from all other EV OEMs
    lolliverchiawatto_cobra
  • Reply 22 of 34
    Fatman said:
    Imagine in the not too distant future ... Apple exclusive microLED displays, Apple’s own 5G chips, and Apple memory chips combined with TSMC 5nm fabrication starting with the A14 - now that would demolish the competition. No more costly reliance on Samsung and Qualcomm. 
    Apple will not escape Qualcomm reliance. Several years will use their modems and even after will pay patent royalties for sure. And getting rid of Samsung's displays won't be that super fast as well. Even I would wish. Important is Samsung got cut off Apple chip design process and manufacturing. It must be much harder to copy design
    muthuk_vanalingam
  • Reply 23 of 34
    blastdoor said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Great question.

    i wonder if it just isn’t worth it. Compare a pixel 4 or top Galaxy to the iPhone and it’s hard to see how the iPhone benefits from its better CPU by a margin that is large enough to warrant competitors making an investment large enough to compete. If Apple were taking serious market share from android then maybe the droid guys would try harder. But Apple has been in the 15 to 20 percent market share range for a long time.

    i think an even better question is — why does Apple bother making a cpu this good? I suspect the answer is that Apple also sells something called an iPad Pro, and they want the cpu speed for that product. Maybe also the Mac?

    if Apple didn’t sell iPads, I wonder if they’d invest less in cpu development...
    Apple makes better and better CPU's because a fast CPU is a power efficient CPU that extends battery life if you don't max it out.  That's why the iPhone 11's have such great battery life.  They don't HAVE to overclock the CPU to get max performance.  Also, there are functions that may be CPU intensive (AR?) that Apple is planning for a year or two down the road.  This makes current iPhones keep their value, and lifts the whole ecosystem.
    watto_cobra
  • Reply 24 of 34
    Don’t buy it

    considering a new scenario: standard ARM core is getting the “good enough” quality

    why I should develop a custom core if I can license a (relative) cheap one with Android already compiled and patched? (remember Android kernel MUST be compiled on a per chip design basis)
    thus saving A LOT of money on OS software upgrades letting my smartphone upgradable on a two year cycle with no cost (like Nokia with no android customization already does)

    I see Samsung going like Huawei for the Kirin
    leveraging a foundry just to be indipendent from Qualcomm 

    Apple with its own os is not comparable 
    qualcomm with its business model either

    sure you will be for a while a couple of step back (eg faceid and ar in the near future)
    but Android market is another stuff and if you really want a stronger chip you can go Qualcomm for the high end certain you will sell few top tier phones anyway

    i agree with the line “there are no volume” to sustain r&d (like for Tegra)
    the point is: is REALLY a problem for Android huge and un-loyal (with respect to the brand of a vendor) market?
    simply not
    edited October 2019
  • Reply 25 of 34
    radarthekatradarthekat Posts: 3,569moderator
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    It’s fine to make cheap devices.  I think the problem we who see Apple as being a great business have is twofold.  First, Android is a rip-off of Apple.  Second, the analysts and pundits like to compare Apple to its competitors based solely on unit numbers, and a unit of a cheap device is not economically equivalent to a unit of a premium device.  If everyone would simple state the facts and own their position in the market and thier history of how they got there, then I, for one, would have a lot less animosity toward the Android vendors.  
    watto_cobra
  • Reply 26 of 34
    avon b7avon b7 Posts: 6,500member
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Market leading processor design is fantastically expensive. Recall how Intel solidly monopolized the PC market for decades, with little progress made even by x86 cloners, not to mention all the alternative chip makers who took stabs at competing with things like Transmeta, or even the Power PC Alliance, for that matter.
    Then look at how difficult it has been to unseat Qualcomm anywhere but in the lower end chip market--although in mobiles that is largely modem-patent related.
    Apple is making it look easy to crank out custom chips, but that's because it has billions to work with an it has the confidence to sell tens of millions of a new product, such as Apple Watch or AirPods.
    As the previous article noted, there's a cycle of "inability to compete" when nobody can afford to speculatively build optimised silicon for powering a sophisicated watch, because nobody is building a sophisticated watch they can sell in volume, and nobody is building a software platform for a sophisticated watch because ...
    There are cheap devices markets, yes, $13 watch bands even. Tablets and TV boxes that cost $30. Those devices are so cheap they need to report what you are doing & advertise and sign you up for services or they are losing money.
    The point isn't that it is bad or undignified to make cheap devices. The point is that companies that are doing this are not going to be delivering the future, the same way that 16-bit PC makers quickly became irrelevant when tech shifted, or basic phone makers evaporated after iPhone, etc.
    Google and Samsung can alternate between suggesting that they are selling affordable phones because they're trying to save you money, and jacking up their phone prices to try to actually stay in business, but if you focus on volume you end up selling an uncompetitive product short term. If you focus on product, you make enough money to stick around.
    How many affordable electric car makers are giving Tesla any real competition?
    Enough companies do chip design, even expensive chip design. All kinds of chips for all kinds of needs. And from bleeding edge manufacturing processes to old manufacturing processes at other companies. Sometimes the chip design doesn't call for the best of the best. Apple actually doesn't design that many.

    If you go back to the PowerPC days, we, in this space, focus on the laptop/desktop versions but even way back then, PowerPC processors were being put into space probes and there was a huge market for embedded  low power PowerPC variants. IIRC, those chips were in around 50% of all cars at one point. Far more PowerPC chips were going into cars than Apple products. Then consoles, networking gear and highly specialised areas like the oil industry. That technology lives on (what might be called a branch of it anyway) in the Power Architecture. Famed for extreme reliability and a huge and hardened software toolkit to play off.

    It's an expensive business but other companies can afford to be in it, even if margins on final products aren't the highest in the business. Samsung, Qualcomm and Huawei all have fantastic custom silicon. Silicon that actually goes further than Apple's which is basically limited to CE coverage.

    It's expensive but the money is there.

    To answer the OP, other companies can (and do) achieve the same or better than Apple in chip design.

    Google, Nvidia, Samsung, Huawei, NXP are all in there with some excellent designs and covering a very broad range of use cases.

    Just one example, the Kirin A1:

    world’s first BT 5.1 & BLE 5.1 wearable chip
    - in-house designed audio DSP
    - BT-UHD support
    audio processing frequency at rates of up to 356 MHZ
    - new generation of Bluetooth signal selection algorithm brought about by intelligent adaptive frequency-hopping technology
    - Active and adaptive, real time signal interference reduction
    - in-house isochronous dual channel Bluetooth chip
    true stereo sound to wireless earbuds, connecting to both buds simultaneously, yet independently.
    - low latency (190ms)
    - low power
    - Ear canal simulation, the output frequency and intensity of the noise reduction signal are adjusted for different ear canal shapes, achieving active noise reduction effect, with open-fit design.

    All in a footprint of 4.3 mm x 4.4 mm.







    edited October 2019
  • Reply 27 of 34
    jbdragonjbdragon Posts: 2,246member
    sflocal said:
    blastdoor said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Great question.

    i wonder if it just isn’t worth it. Compare a pixel 4 or top Galaxy to the iPhone and it’s hard to see how the iPhone benefits from its better CPU by a margin that is large enough to warrant competitors making an investment large enough to compete. If Apple were taking serious market share from android then maybe the droid guys would try harder. But Apple has been in the 15 to 20 percent market share range for a long time.

    i think an even better question is — why does Apple bother making a cpu this good? I suspect the answer is that Apple also sells something called an iPad Pro, and they want the cpu speed for that product. Maybe also the Mac?

    if Apple didn’t sell iPads, I wonder if they’d invest less in cpu development...
    I get that, but why is market share important when the things that Apple is doing allows it to receive almost all the mobile profits in the industry?  That for sure tells us that Apple's investment in its own CPU path is paying off.

    I've also said this in the past.  Bring it down to simple levels.  Apple sells 1 iphone and makes a $50 profit.  Android(That's just everyone making Android phones) sell 1 ZILLION Android phones and breaks even, who is winning?  Well, Android users would say Android, they sold 1 zillion phones and Apple only sold 1.  Anyone else with a brain would say Apple won. Apple made a $50 profit and didn't have to work as hard to do it.

    Then you can scale it up.  Apple sells 10 phones and now makes a $500 profit.  Android sells 10 ZILLION phones and once again breaks even.  Again, who won?  

    Apple doesn't need to be #1.  Apple is making over 80% of the profits on phones.  While Samesung sells more phones that Apple, they don't make anywhere near as much as Apple doing it as most of their phones sold are in the low to mid-range zone where profit is very little.  Samesung makes most of its money from High-end phone sales.

    Generally, if you're #! in the market, you're making the most money.  That just hasn't been the case for Smartphones.   You have all these Android company's selling phones, Millions of them, but you look at their profits, and you go, that's it?    The only real winner with Android is Google.  They really don't care who is making phones using their OS so long as their apps and services are on the device.   Google Wins no matter what.

    watto_cobra
  • Reply 28 of 34
    melgrossmelgross Posts: 33,334member
    frantisek said:
    Fatman said:
    Imagine in the not too distant future ... Apple exclusive microLED displays, Apple’s own 5G chips, and Apple memory chips combined with TSMC 5nm fabrication starting with the A14 - now that would demolish the competition. No more costly reliance on Samsung and Qualcomm. 
    Apple will not escape Qualcomm reliance. Several years will use their modems and even after will pay patent royalties for sure. And getting rid of Samsung's displays won't be that super fast as well. Even I would wish. Important is Samsung got cut off Apple chip design process and manufacturing. It must be much harder to copy design
    I hate to tell you this, but every modem manufacturer is paying someone else for essential patents. That’s why those patents are essential. That’s why Qualcomm pays for some, Huawei pays for some. As many of those agreements are propriety, and therefor secret, we don’t know every contract out there. But it’s even possible that Huawei is paying Qualcomm, while Qualcomm is paying Huawei. We really don’t know. But now, Qualcomm has been ordered to open up their essential patent vault to everybody, so it will be interesting.
    watto_cobra
  • Reply 29 of 34
    melgrossmelgross Posts: 33,334member

    avon b7 said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Market leading processor design is fantastically expensive. Recall how Intel solidly monopolized the PC market for decades, with little progress made even by x86 cloners, not to mention all the alternative chip makers who took stabs at competing with things like Transmeta, or even the Power PC Alliance, for that matter.
    Then look at how difficult it has been to unseat Qualcomm anywhere but in the lower end chip market--although in mobiles that is largely modem-patent related.
    Apple is making it look easy to crank out custom chips, but that's because it has billions to work with an it has the confidence to sell tens of millions of a new product, such as Apple Watch or AirPods.
    As the previous article noted, there's a cycle of "inability to compete" when nobody can afford to speculatively build optimised silicon for powering a sophisicated watch, because nobody is building a sophisticated watch they can sell in volume, and nobody is building a software platform for a sophisticated watch because ...
    There are cheap devices markets, yes, $13 watch bands even. Tablets and TV boxes that cost $30. Those devices are so cheap they need to report what you are doing & advertise and sign you up for services or they are losing money.
    The point isn't that it is bad or undignified to make cheap devices. The point is that companies that are doing this are not going to be delivering the future, the same way that 16-bit PC makers quickly became irrelevant when tech shifted, or basic phone makers evaporated after iPhone, etc.
    Google and Samsung can alternate between suggesting that they are selling affordable phones because they're trying to save you money, and jacking up their phone prices to try to actually stay in business, but if you focus on volume you end up selling an uncompetitive product short term. If you focus on product, you make enough money to stick around.
    How many affordable electric car makers are giving Tesla any real competition?
    Enough companies do chip design, even expensive chip design. All kinds of chips for all kinds of needs. And from bleeding edge manufacturing processes to old manufacturing processes at other companies. Sometimes the chip design doesn't call for the best of the best. Apple actually doesn't design that many.

    If you go back to the PowerPC days, we, in this space, focus on the laptop/desktop versions but even way back then, PowerPC processors were being put into space probes and there was a huge market for embedded  low power PowerPC variants. IIRC, those chips were in around 50% of all cars at one point. Far more PowerPC chips were going into cars than Apple products. Then consoles, networking gear and highly specialised areas like the oil industry. That technology lives on (what might be called a branch of it anyway) in the Power Architecture. Famed for extreme reliability and a huge and hardened software toolkit to play off.

    It's an expensive business but other companies can afford to be in it, even if margins on final products aren't the highest in the business. Samsung, Qualcomm and Huawei all have fantastic custom silicon. Silicon that actually goes further than Apple's which is basically limited to CE coverage.

    It's expensive but the money is there.

    To answer the OP, other companies can (and do) achieve the same or better than Apple in chip design.

    Google, Nvidia, Samsung, Huawei, NXP are all in there with some excellent designs and covering a very broad range of use cases.

    Just one example, the Kirin A1:

    - world’s first BT 5.1 & BLE 5.1 wearable chip
    - in-house designed audio DSP
    - BT-UHD support
    - audio processing frequency at rates of up to 356 MHZ
    - new generation of Bluetooth signal selection algorithm brought about by intelligent adaptive frequency-hopping technology
    - Active and adaptive, real time signal interference reduction
    - in-house isochronous dual channel Bluetooth chip
    - true stereo sound to wireless earbuds, connecting to both buds simultaneously, yet independently.
    - low latency (190ms)
    - low power
    - Ear canal simulation, the output frequency and intensity of the noise reduction signal are adjusted for different ear canal shapes, achieving active noise reduction effect, with open-fit design.

    All in a footprint of 4.3 mm x 4.4 mm.







    Really, Huawei’s SoCs aren’t great, and you know it. The Kirin use standard ARM designs. That’s been noted on many different occasions. Like others, they=y are allowed to do two things when they have design licenses. One is to adapt the cores to a requires process technology. Two is they’re allowed to add proprietary modules to the chip. Bu5t they’re not allowed to modify the cores otherwise, change the logic, add instructions, etc.

    their latest 990 has added modems, which  is allowed. That’s not unusual for a company that also makes modems, it’s obviously what Apple will be doing once they have their own.

    i also don’t see all that Mitch in your list that hasn’t already been done by others, including Apple.
    watto_cobra
  • Reply 30 of 34
    avon b7avon b7 Posts: 6,500member
    melgross said:
    frantisek said:
    Fatman said:
    Imagine in the not too distant future ... Apple exclusive microLED displays, Apple’s own 5G chips, and Apple memory chips combined with TSMC 5nm fabrication starting with the A14 - now that would demolish the competition. No more costly reliance on Samsung and Qualcomm. 
    Apple will not escape Qualcomm reliance. Several years will use their modems and even after will pay patent royalties for sure. And getting rid of Samsung's displays won't be that super fast as well. Even I would wish. Important is Samsung got cut off Apple chip design process and manufacturing. It must be much harder to copy design
    I hate to tell you this, but every modem manufacturer is paying someone else for essential patents. That’s why those patents are essential. That’s why Qualcomm pays for some, Huawei pays for some. As many of those agreements are propriety, and therefor secret, we don’t know every contract out there. But it’s even possible that Huawei is paying Qualcomm, while Qualcomm is paying Huawei. We really don’t know. But now, Qualcomm has been ordered to open up their essential patent vault to everybody, so it will be interesting.
    We know they are paying each other. They are even fighting in court.
  • Reply 31 of 34
    avon b7avon b7 Posts: 6,500member
    melgross said:

    avon b7 said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Market leading processor design is fantastically expensive. Recall how Intel solidly monopolized the PC market for decades, with little progress made even by x86 cloners, not to mention all the alternative chip makers who took stabs at competing with things like Transmeta, or even the Power PC Alliance, for that matter.
    Then look at how difficult it has been to unseat Qualcomm anywhere but in the lower end chip market--although in mobiles that is largely modem-patent related.
    Apple is making it look easy to crank out custom chips, but that's because it has billions to work with an it has the confidence to sell tens of millions of a new product, such as Apple Watch or AirPods.
    As the previous article noted, there's a cycle of "inability to compete" when nobody can afford to speculatively build optimised silicon for powering a sophisicated watch, because nobody is building a sophisticated watch they can sell in volume, and nobody is building a software platform for a sophisticated watch because ...
    There are cheap devices markets, yes, $13 watch bands even. Tablets and TV boxes that cost $30. Those devices are so cheap they need to report what you are doing & advertise and sign you up for services or they are losing money.
    The point isn't that it is bad or undignified to make cheap devices. The point is that companies that are doing this are not going to be delivering the future, the same way that 16-bit PC makers quickly became irrelevant when tech shifted, or basic phone makers evaporated after iPhone, etc.
    Google and Samsung can alternate between suggesting that they are selling affordable phones because they're trying to save you money, and jacking up their phone prices to try to actually stay in business, but if you focus on volume you end up selling an uncompetitive product short term. If you focus on product, you make enough money to stick around.
    How many affordable electric car makers are giving Tesla any real competition?
    Enough companies do chip design, even expensive chip design. All kinds of chips for all kinds of needs. And from bleeding edge manufacturing processes to old manufacturing processes at other companies. Sometimes the chip design doesn't call for the best of the best. Apple actually doesn't design that many.

    If you go back to the PowerPC days, we, in this space, focus on the laptop/desktop versions but even way back then, PowerPC processors were being put into space probes and there was a huge market for embedded  low power PowerPC variants. IIRC, those chips were in around 50% of all cars at one point. Far more PowerPC chips were going into cars than Apple products. Then consoles, networking gear and highly specialised areas like the oil industry. That technology lives on (what might be called a branch of it anyway) in the Power Architecture. Famed for extreme reliability and a huge and hardened software toolkit to play off.

    It's an expensive business but other companies can afford to be in it, even if margins on final products aren't the highest in the business. Samsung, Qualcomm and Huawei all have fantastic custom silicon. Silicon that actually goes further than Apple's which is basically limited to CE coverage.

    It's expensive but the money is there.

    To answer the OP, other companies can (and do) achieve the same or better than Apple in chip design.

    Google, Nvidia, Samsung, Huawei, NXP are all in there with some excellent designs and covering a very broad range of use cases.

    Just one example, the Kirin A1:

    - world’s first BT 5.1 & BLE 5.1 wearable chip
    - in-house designed audio DSP
    - BT-UHD support
    - audio processing frequency at rates of up to 356 MHZ
    - new generation of Bluetooth signal selection algorithm brought about by intelligent adaptive frequency-hopping technology
    - Active and adaptive, real time signal interference reduction
    - in-house isochronous dual channel Bluetooth chip
    - true stereo sound to wireless earbuds, connecting to both buds simultaneously, yet independently.
    - low latency (190ms)
    - low power
    - Ear canal simulation, the output frequency and intensity of the noise reduction signal are adjusted for different ear canal shapes, achieving active noise reduction effect, with open-fit design.

    All in a footprint of 4.3 mm x 4.4 mm.







    Really, Huawei’s SoCs aren’t great, and you know it. The Kirin use standard ARM designs. That’s been noted on many different occasions. Like others, they=y are allowed to do two things when they have design licenses. One is to adapt the cores to a requires process technology. Two is they’re allowed to add proprietary modules to the chip. Bu5t they’re not allowed to modify the cores otherwise, change the logic, add instructions, etc.

    their latest 990 has added modems, which  is allowed. That’s not unusual for a company that also makes modems, it’s obviously what Apple will be doing once they have their own.

    i also don’t see all that Mitch in your list that hasn’t already been done by others, including Apple.
    Pretty much everything on the list is not done by Apple. That's why I gave it as an example.
  • Reply 32 of 34
    melgrossmelgross Posts: 33,334member
    avon b7 said:
    melgross said:

    avon b7 said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Market leading processor design is fantastically expensive. Recall how Intel solidly monopolized the PC market for decades, with little progress made even by x86 cloners, not to mention all the alternative chip makers who took stabs at competing with things like Transmeta, or even the Power PC Alliance, for that matter.
    Then look at how difficult it has been to unseat Qualcomm anywhere but in the lower end chip market--although in mobiles that is largely modem-patent related.
    Apple is making it look easy to crank out custom chips, but that's because it has billions to work with an it has the confidence to sell tens of millions of a new product, such as Apple Watch or AirPods.
    As the previous article noted, there's a cycle of "inability to compete" when nobody can afford to speculatively build optimised silicon for powering a sophisicated watch, because nobody is building a sophisticated watch they can sell in volume, and nobody is building a software platform for a sophisticated watch because ...
    There are cheap devices markets, yes, $13 watch bands even. Tablets and TV boxes that cost $30. Those devices are so cheap they need to report what you are doing & advertise and sign you up for services or they are losing money.
    The point isn't that it is bad or undignified to make cheap devices. The point is that companies that are doing this are not going to be delivering the future, the same way that 16-bit PC makers quickly became irrelevant when tech shifted, or basic phone makers evaporated after iPhone, etc.
    Google and Samsung can alternate between suggesting that they are selling affordable phones because they're trying to save you money, and jacking up their phone prices to try to actually stay in business, but if you focus on volume you end up selling an uncompetitive product short term. If you focus on product, you make enough money to stick around.
    How many affordable electric car makers are giving Tesla any real competition?
    Enough companies do chip design, even expensive chip design. All kinds of chips for all kinds of needs. And from bleeding edge manufacturing processes to old manufacturing processes at other companies. Sometimes the chip design doesn't call for the best of the best. Apple actually doesn't design that many.

    If you go back to the PowerPC days, we, in this space, focus on the laptop/desktop versions but even way back then, PowerPC processors were being put into space probes and there was a huge market for embedded  low power PowerPC variants. IIRC, those chips were in around 50% of all cars at one point. Far more PowerPC chips were going into cars than Apple products. Then consoles, networking gear and highly specialised areas like the oil industry. That technology lives on (what might be called a branch of it anyway) in the Power Architecture. Famed for extreme reliability and a huge and hardened software toolkit to play off.

    It's an expensive business but other companies can afford to be in it, even if margins on final products aren't the highest in the business. Samsung, Qualcomm and Huawei all have fantastic custom silicon. Silicon that actually goes further than Apple's which is basically limited to CE coverage.

    It's expensive but the money is there.

    To answer the OP, other companies can (and do) achieve the same or better than Apple in chip design.

    Google, Nvidia, Samsung, Huawei, NXP are all in there with some excellent designs and covering a very broad range of use cases.

    Just one example, the Kirin A1:

    - world’s first BT 5.1 & BLE 5.1 wearable chip
    - in-house designed audio DSP
    - BT-UHD support
    - audio processing frequency at rates of up to 356 MHZ
    - new generation of Bluetooth signal selection algorithm brought about by intelligent adaptive frequency-hopping technology
    - Active and adaptive, real time signal interference reduction
    - in-house isochronous dual channel Bluetooth chip
    - true stereo sound to wireless earbuds, connecting to both buds simultaneously, yet independently.
    - low latency (190ms)
    - low power
    - Ear canal simulation, the output frequency and intensity of the noise reduction signal are adjusted for different ear canal shapes, achieving active noise reduction effect, with open-fit design.

    All in a footprint of 4.3 mm x 4.4 mm.







    Really, Huawei’s SoCs aren’t great, and you know it. The Kirin use standard ARM designs. That’s been noted on many different occasions. Like others, they=y are allowed to do two things when they have design licenses. One is to adapt the cores to a requires process technology. Two is they’re allowed to add proprietary modules to the chip. Bu5t they’re not allowed to modify the cores otherwise, change the logic, add instructions, etc.

    their latest 990 has added modems, which  is allowed. That’s not unusual for a company that also makes modems, it’s obviously what Apple will be doing once they have their own.

    i also don’t see all that Mitch in your list that hasn’t already been done by others, including Apple.
    Pretty much everything on the list is not done by Apple. That's why I gave it as an example.
    Except that you’re wrong. And we don’t know everything Apple does in house, just what they let out. Sometimes we only find out years later.
    watto_cobra
  • Reply 33 of 34
    avon b7avon b7 Posts: 6,500member
    melgross said:
    avon b7 said:
    melgross said:

    avon b7 said:
    Still not sure why others can’t achieve what Apple can in chip design. And, while every editorial praises Apple’s approach (concentrating on premium devices only), there must be companies making cheap devices. There’s nothing wrong with being a cheap device producer. You obviously can’t make more money from cheap devices as they priced less than Apple’s margin on iPhones.
    Market leading processor design is fantastically expensive. Recall how Intel solidly monopolized the PC market for decades, with little progress made even by x86 cloners, not to mention all the alternative chip makers who took stabs at competing with things like Transmeta, or even the Power PC Alliance, for that matter.
    Then look at how difficult it has been to unseat Qualcomm anywhere but in the lower end chip market--although in mobiles that is largely modem-patent related.
    Apple is making it look easy to crank out custom chips, but that's because it has billions to work with an it has the confidence to sell tens of millions of a new product, such as Apple Watch or AirPods.
    As the previous article noted, there's a cycle of "inability to compete" when nobody can afford to speculatively build optimised silicon for powering a sophisicated watch, because nobody is building a sophisticated watch they can sell in volume, and nobody is building a software platform for a sophisticated watch because ...
    There are cheap devices markets, yes, $13 watch bands even. Tablets and TV boxes that cost $30. Those devices are so cheap they need to report what you are doing & advertise and sign you up for services or they are losing money.
    The point isn't that it is bad or undignified to make cheap devices. The point is that companies that are doing this are not going to be delivering the future, the same way that 16-bit PC makers quickly became irrelevant when tech shifted, or basic phone makers evaporated after iPhone, etc.
    Google and Samsung can alternate between suggesting that they are selling affordable phones because they're trying to save you money, and jacking up their phone prices to try to actually stay in business, but if you focus on volume you end up selling an uncompetitive product short term. If you focus on product, you make enough money to stick around.
    How many affordable electric car makers are giving Tesla any real competition?
    Enough companies do chip design, even expensive chip design. All kinds of chips for all kinds of needs. And from bleeding edge manufacturing processes to old manufacturing processes at other companies. Sometimes the chip design doesn't call for the best of the best. Apple actually doesn't design that many.

    If you go back to the PowerPC days, we, in this space, focus on the laptop/desktop versions but even way back then, PowerPC processors were being put into space probes and there was a huge market for embedded  low power PowerPC variants. IIRC, those chips were in around 50% of all cars at one point. Far more PowerPC chips were going into cars than Apple products. Then consoles, networking gear and highly specialised areas like the oil industry. That technology lives on (what might be called a branch of it anyway) in the Power Architecture. Famed for extreme reliability and a huge and hardened software toolkit to play off.

    It's an expensive business but other companies can afford to be in it, even if margins on final products aren't the highest in the business. Samsung, Qualcomm and Huawei all have fantastic custom silicon. Silicon that actually goes further than Apple's which is basically limited to CE coverage.

    It's expensive but the money is there.

    To answer the OP, other companies can (and do) achieve the same or better than Apple in chip design.

    Google, Nvidia, Samsung, Huawei, NXP are all in there with some excellent designs and covering a very broad range of use cases.

    Just one example, the Kirin A1:

    - world’s first BT 5.1 & BLE 5.1 wearable chip
    - in-house designed audio DSP
    - BT-UHD support
    - audio processing frequency at rates of up to 356 MHZ
    - new generation of Bluetooth signal selection algorithm brought about by intelligent adaptive frequency-hopping technology
    - Active and adaptive, real time signal interference reduction
    - in-house isochronous dual channel Bluetooth chip
    - true stereo sound to wireless earbuds, connecting to both buds simultaneously, yet independently.
    - low latency (190ms)
    - low power
    - Ear canal simulation, the output frequency and intensity of the noise reduction signal are adjusted for different ear canal shapes, achieving active noise reduction effect, with open-fit design.

    All in a footprint of 4.3 mm x 4.4 mm.







    Really, Huawei’s SoCs aren’t great, and you know it. The Kirin use standard ARM designs. That’s been noted on many different occasions. Like others, they=y are allowed to do two things when they have design licenses. One is to adapt the cores to a requires process technology. Two is they’re allowed to add proprietary modules to the chip. Bu5t they’re not allowed to modify the cores otherwise, change the logic, add instructions, etc.

    their latest 990 has added modems, which  is allowed. That’s not unusual for a company that also makes modems, it’s obviously what Apple will be doing once they have their own.

    i also don’t see all that Mitch in your list that hasn’t already been done by others, including Apple.
    Pretty much everything on the list is not done by Apple. That's why I gave it as an example.
    Except that you’re wrong. And we don’t know everything Apple does in house, just what they let out. Sometimes we only find out years later.
    What is that supposed to mean? That's just wacky. The exact same logic is applicable to Huawei. I'm talking about announced products!

    "Except your wrong"

    Great reply - not - with nothing to back
    it up (as usual). Please run through the list and point me to the equivalent Apple product.

    If I'm wrong, tell me where.
  • Reply 34 of 34
    avon b7avon b7 Posts: 6,500member
    melgross said:

    Really, Huawei’s SoCs aren’t great, and you know it. The Kirin use standard ARM designs. That’s been noted on many different occasions. Like others, they=y are allowed to do two things when they have design licenses. One is to adapt the cores to a requires process technology. Two is they’re allowed to add proprietary modules to the chip. Bu5t they’re not allowed to modify the cores otherwise, change the logic, add instructions, etc.

    their latest 990 has added modems, which  is allowed. That’s not unusual for a company that also makes modems, it’s obviously what Apple will be doing once they have their own.

    i also don’t see all that Mitch in your list that hasn’t already been done by others, including Apple.
    You totally missed the point there.

    We are talking about processors and
    SoCs, not just cores. There is far more involved than the cores!

    So much so, that HiSilicon didn't even bother to run with the latest ARM cores on the Kirin 990. And it isn't the first time they have done it. The 970 was in the same situation.

    Let that sink in for a minute.

    The SoCs have a whole lot of other components and where many of the differences are made. The inclusion of a 5G NSA/SA modem on the SoC is a major achievement. Trying to imply otherwise or suggesting Apple will 'obviously' do the same, also misses the point that Apple is way behind and is playing catch-up.

    I gave the Kirin A1 as just one example. Simply because there was a similar chipset from Apple.

    However, there are lots more examples where there is simply NO Apple equivalent and that was the point. The manufacturers I listed are also doing great work in the chipset field. Obviously I focus on Huawei because I know a bit about the situation. There are enough companies out there doing amazing work without billions in the bank.

    Apple has limited itself to the CE space. Really little more than that. That is the reality.

    Take a look at some of these systems:

    Ascend
    Kunpeng
    Tiangang
    Balong
    HiSilicon Kirin

    You will instantly see there is more involved than simple cores!

    Apple literally isn't on the same playing field here.

    Is there an Apple equivalent to the Atlas 900?



    The Ascend 910?



    MindSpore?



    Then of course there are the lesser known aspects which are actually possible today but not out of the pipe yet.



    Yes, Huawei is actively involved in quantum computing research too


    edited October 2019
Sign In or Register to comment.