Likely due to Huawei being part of the same PRC that want's to take Taiwan back. Doesn't make sense to provide the dual use technology to do that.
No.
TSMC hasn't 'decided' to stop manufacturing for Huawei (one of its biggest customers).
TSMC is prohibited from serving Huawei because of extraterritorial U.S orders.
Where have you been for the last month?
How could you have possibly missed this news?
I’m guessing, based on your pathological devotion to China and Huawei that you would support a PRC/CCP invasion of Taiwan to recover its lawful territory, right? Then Huawei and TMSC could live happily ever after together. Would that please you?
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
I highly doubt that.
The following sites say 87% for 7nm-> 5nm, 70% for 5nm-> 3nm:
10nm density = 52.5 million transistors per square mm, 7nm = 91.2m, 5nm = 171.3m, 3nm = 291.2m. Going from 10nm to 7nm was 73% increase (actual), 7nm to 5nm is 87% estimated. 5nm to 3nm is 70% estimated.
Given that they achieved 73% from 10nm to 7nm, the others are plausible estimates.
"Both silicon and MoS2 have a crystalline lattice structure, but electrons flowing through silicon have a smaller effective mass compared with MoS2. That is a boon when the gate is 5 nanometers or longer. But below that length, a quantum mechanical phenomenon called tunneling kicks in, and the gate barrier is no longer able to keep the electrons from barging through from the source to the drain terminals. This means we can’t turn off the transistors."
It has to scale cost-effectively to have billions of them though.
The roadmap all the manufacturers are planning for is 5nm, 3nm, 2nm, 1.4nm. There was an article from Samsung saying 1.4nm is as low as they can go.
I don't think commercially there will be much need for better than 1.4nm. This will be around 4.5x what we have at 7nm i.e 20TFLOP GPUs in laptops, that's close to half of a high-end Mac Pro in a laptop. When everybody is stuck at 1.4nm, it will come down to chip design and coprocessors to set performance apart.
Intel could possibly manufacture Apple's ARM chips once they can rival TSMC.
Likely due to Huawei being part of the same PRC that want's to take Taiwan back. Doesn't make sense to provide the dual use technology to do that.
No.
TSMC hasn't 'decided' to stop manufacturing for Huawei (one of its biggest customers).
TSMC is prohibited from serving Huawei because of extraterritorial U.S orders.
Where have you been for the last month?
How could you have possibly missed this news?
I’m guessing, based on your pathological devotion to China and Huawei that you would support a PRC/CCP invasion of Taiwan to recover its lawful territory, right? Then Huawei and TMSC could live happily ever after together. Would that please you?
I'm not a fan of invasions, independently of who is doing it, unless it is to recover something that was taken illegally by force in the first place.
Huawei and TSMC have a very good relationship. The 3nm node was going to be used by Huawei, too.
Surely you don't think TSMC is happy losing one of its best historical customers? Or seeing them potentially go to its competitors, or worse, China accelerating its already stated goals in becoming self sufficient in chip design AND fabrication and reaching those goals sooner than expected (even if they still be a few years off).
Likely due to Huawei being part of the same PRC that want's to take Taiwan back. Doesn't make sense to provide the dual use technology to do that.
No.
TSMC hasn't 'decided' to stop manufacturing for Huawei (one of its biggest customers).
TSMC is prohibited from serving Huawei because of extraterritorial U.S orders.
Where have you been for the last month?
How could you have possibly missed this news?
I’m guessing, based on your pathological devotion to China and Huawei that you would support a PRC/CCP invasion of Taiwan to recover its lawful territory, right? Then Huawei and TMSC could live happily ever after together. Would that please you?
I'm not a fan of invasions, independently of who is doing it, unless it is to recover something that was taken illegally by force in the first place.
Huawei and TSMC have a very good relationship. The 3nm node was going to be used by Huawei, too.
Surely you don't think TSMC is happy losing one of its best historical customers? Or seeing them potentially go to its competitors, or worse, China accelerating its already stated goals in becoming self sufficient in chip design AND fabrication and reaching those goals sooner than expected (even if they still be a few years off).
Taiwan was never taken illegally, though Maoist China wanted the world to believe that. It never belonged to the People's Republic of China. It was a divorce between too factions after WWII. For the record, it was the U.S. that helped start TMSC, and the U.S. that supported Taiwan, not Mainland China. More to the point, Tiawan is a vibrant democracy, that deserves a future that is not authoritarian. It's obvious that you don't give a shit about democracy, nor Chinese aggression in South Asia, and certainly not of human rights. All you care about is your precious technology, delivered soonest and cheapest. Hong Kong and Human Rights are certainly why the UK banned Huawei 5G installs and will remove all existing Huawei telecom equipment by 2017. The fact that the U.S. and many other Democracies have suspended their extradition treaties with the PRC is damning.
Apple, Qualcomm, Nvidia, and AMD are more important than Huawei, hence why I posted that link indicating Huawei's share at 14%. More to the point, which you ignored, is that Taiwan does not want to support dual use, civilian and military, technology in the PRC, something that would be used by the PRC during an invasion. As Huawei is tightly linked to and controlled by the PRC and the CCP, there isn't any reason that TMSC should have maintained its relationship.
"“According to your own local government statistics, the population growth in Uighur jurisdictions in that area has fallen by 84 percent between 2015 and 2018,” Marr pressed. “84 percent!”
That's a human rights violation. Uighur prisoners is a human rights violation, just as taking Tibet was.
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
I highly doubt that.
The following sites say 87% for 7nm-> 5nm, 70% for 5nm-> 3nm:
10nm density = 52.5 million transistors per square mm, 7nm = 91.2m, 5nm = 171.3m, 3nm = 291.2m. Going from 10nm to 7nm was 73% increase (actual), 7nm to 5nm is 87% estimated. 5nm to 3nm is 70% estimated.
Given that they achieved 73% from 10nm to 7nm, the others are plausible estimates.
"Both silicon and MoS2 have a crystalline lattice structure, but electrons flowing through silicon have a smaller effective mass compared with MoS2. That is a boon when the gate is 5 nanometers or longer. But below that length, a quantum mechanical phenomenon called tunneling kicks in, and the gate barrier is no longer able to keep the electrons from barging through from the source to the drain terminals. This means we can’t turn off the transistors."
It has to scale cost-effectively to have billions of them though.
The roadmap all the manufacturers are planning for is 5nm, 3nm, 2nm, 1.4nm. There was an article from Samsung saying 1.4nm is as low as they can go.
I don't think commercially there will be much need for better than 1.4nm. This will be around 4.5x what we have at 7nm i.e 20TFLOP GPUs in laptops, that's close to half of a high-end Mac Pro in a laptop. When everybody is stuck at 1.4nm, it will come down to chip design and coprocessors to set performance apart.
Intel could possibly manufacture Apple's ARM chips once they can rival TSMC.
Nonsense. None of the other process shrinks lately have l8ved up to the initial numbers, and this won’t either.
Likely due to Huawei being part of the same PRC that want's to take Taiwan back. Doesn't make sense to provide the dual use technology to do that.
No.
TSMC hasn't 'decided' to stop manufacturing for Huawei (one of its biggest customers).
TSMC is prohibited from serving Huawei because of extraterritorial U.S orders.
Where have you been for the last month?
How could you have possibly missed this news?
I’m guessing, based on your pathological devotion to China and Huawei that you would support a PRC/CCP invasion of Taiwan to recover its lawful territory, right? Then Huawei and TMSC could live happily ever after together. Would that please you?
I'm not a fan of invasions, independently of who is doing it, unless it is to recover something that was taken illegally by force in the first place.
Huawei and TSMC have a very good relationship. The 3nm node was going to be used by Huawei, too.
Surely you don't think TSMC is happy losing one of its best historical customers? Or seeing them potentially go to its competitors, or worse, China accelerating its already stated goals in becoming self sufficient in chip design AND fabrication and reaching those goals sooner than expected (even if they still be a few years off).
Taiwan was never taken illegally, though Maoist China wanted the world to believe that. It never belonged to the People's Republic of China. It was a divorce between too factions after WWII. For the record, it was the U.S. that helped start TMSC, and the U.S. that supported Taiwan, not Mainland China. More to the point, Tiawan is a vibrant democracy, that deserves a future that is not authoritarian. It's obvious that you don't give a shit about democracy, nor Chinese aggression in South Asia, and certainly not of human rights. All you care about is your precious technology, delivered soonest and cheapest. Hong Kong and Human Rights are certainly why the UK banned Huawei 5G installs and will remove all existing Huawei telecom equipment by 2017. The fact that the U.S. and many other Democracies have suspended their extradition treaties with the PRC is damning.
Apple, Qualcomm, Nvidia, and AMD are more important than Huawei, hence why I posted that link indicating Huawei's share at 14%. More to the point, which you ignored, is that Taiwan does not want to support dual use, civilian and military, technology in the PRC, something that would be used by the PRC during an invasion. As Huawei is tightly linked to and controlled by the PRC and the CCP, there isn't any reason that TMSC should have maintained its relationship.
"“According to your own local government statistics, the population growth in Uighur jurisdictions in that area has fallen by 84 percent between 2015 and 2018,” Marr pressed. “84 percent!”
That's a human rights violation. Uighur prisoners is a human rights violation, just as taking Tibet was.
None of that has much to do with what I said and I made zero reference to the geopolitical situation between China and Taiwan.
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
I highly doubt that.
The following sites say 87% for 7nm-> 5nm, 70% for 5nm-> 3nm:
10nm density = 52.5 million transistors per square mm, 7nm = 91.2m, 5nm = 171.3m, 3nm = 291.2m. Going from 10nm to 7nm was 73% increase (actual), 7nm to 5nm is 87% estimated. 5nm to 3nm is 70% estimated.
Given that they achieved 73% from 10nm to 7nm, the others are plausible estimates.
"Both silicon and MoS2 have a crystalline lattice structure, but electrons flowing through silicon have a smaller effective mass compared with MoS2. That is a boon when the gate is 5 nanometers or longer. But below that length, a quantum mechanical phenomenon called tunneling kicks in, and the gate barrier is no longer able to keep the electrons from barging through from the source to the drain terminals. This means we can’t turn off the transistors."
It has to scale cost-effectively to have billions of them though.
The roadmap all the manufacturers are planning for is 5nm, 3nm, 2nm, 1.4nm. There was an article from Samsung saying 1.4nm is as low as they can go.
I don't think commercially there will be much need for better than 1.4nm. This will be around 4.5x what we have at 7nm i.e 20TFLOP GPUs in laptops, that's close to half of a high-end Mac Pro in a laptop. When everybody is stuck at 1.4nm, it will come down to chip design and coprocessors to set performance apart.
Intel could possibly manufacture Apple's ARM chips once they can rival TSMC.
Nonsense. None of the other process shrinks lately have l8ved up to the initial numbers, and this won’t either.
A13 Bionic 7nm has 8.5 billion transistors in 98mm^2 = 86m transistors per square mm vs 91.2m listed. A12 Bionic 7nm has 6.9 billion in 83mm^2 = 83m/mm^2. A11 10nm has 4.3 billion in 87mm^2 = 49m/mm^2 vs 52.5m listed.
It depends on the chip design how tightly they fit the transistors in the products but from one generation to the next 83 vs 49 = 70% improvement in density. Here are some GPUs:
If they want to have better cooling then they can build a bigger chip so the density is lower. Mobile chips have to be small and Apple's chip density matches TSMC's numbers very closely.
This doesn't usually translate into the same amount of performance increase i.e 70% density doesn't mean 70% performance, sometimes they lower the power usage so 30% drop in power plus 30% increase in performance and they tend to split the improvements over 2 years.
These manufacturers invest billions of dollars into their plants, they have to come close to their estimates or their customers would use another manufacturer.
Likely due to Huawei being part of the same PRC that want's to take Taiwan back. Doesn't make sense to provide the dual use technology to do that.
No.
TSMC hasn't 'decided' to stop manufacturing for Huawei (one of its biggest customers).
TSMC is prohibited from serving Huawei because of extraterritorial U.S orders.
Where have you been for the last month?
How could you have possibly missed this news?
I’m guessing, based on your pathological devotion to China and Huawei that you would support a PRC/CCP invasion of Taiwan to recover its lawful territory, right? Then Huawei and TMSC could live happily ever after together. Would that please you?
I'm not a fan of invasions, independently of who is doing it, unless it is to recover something that was taken illegally by force in the first place.
Huawei and TSMC have a very good relationship. The 3nm node was going to be used by Huawei, too.
Surely you don't think TSMC is happy losing one of its best historical customers? Or seeing them potentially go to its competitors, or worse, China accelerating its already stated goals in becoming self sufficient in chip design AND fabrication and reaching those goals sooner than expected (even if they still be a few years off).
Taiwan was never taken illegally, though Maoist China wanted the world to believe that. It never belonged to the People's Republic of China. It was a divorce between too factions after WWII. For the record, it was the U.S. that helped start TMSC, and the U.S. that supported Taiwan, not Mainland China. More to the point, Tiawan is a vibrant democracy, that deserves a future that is not authoritarian. It's obvious that you don't give a shit about democracy, nor Chinese aggression in South Asia, and certainly not of human rights. All you care about is your precious technology, delivered soonest and cheapest. Hong Kong and Human Rights are certainly why the UK banned Huawei 5G installs and will remove all existing Huawei telecom equipment by 2017. The fact that the U.S. and many other Democracies have suspended their extradition treaties with the PRC is damning.
Apple, Qualcomm, Nvidia, and AMD are more important than Huawei, hence why I posted that link indicating Huawei's share at 14%. More to the point, which you ignored, is that Taiwan does not want to support dual use, civilian and military, technology in the PRC, something that would be used by the PRC during an invasion. As Huawei is tightly linked to and controlled by the PRC and the CCP, there isn't any reason that TMSC should have maintained its relationship.
"“According to your own local government statistics, the population growth in Uighur jurisdictions in that area has fallen by 84 percent between 2015 and 2018,” Marr pressed. “84 percent!”
That's a human rights violation. Uighur prisoners is a human rights violation, just as taking Tibet was.
None of that has much to do with what I said and I made zero reference to the geopolitical situation between China and Taiwan.
According to the company, the 3-nanometer process will provide quite a few benefits over earlier processes, even the relatively recent 5-nanometer. Compared to 5nm, 3nm will have a 15% greater transistor density, boast a performance increase of between 10% and 15%, and offer energy efficiency savings of between 20% and 25%.
N3 will have a 70% greater transistor density than N5.
I highly doubt that.
The following sites say 87% for 7nm-> 5nm, 70% for 5nm-> 3nm:
10nm density = 52.5 million transistors per square mm, 7nm = 91.2m, 5nm = 171.3m, 3nm = 291.2m. Going from 10nm to 7nm was 73% increase (actual), 7nm to 5nm is 87% estimated. 5nm to 3nm is 70% estimated.
Given that they achieved 73% from 10nm to 7nm, the others are plausible estimates.
"Both silicon and MoS2 have a crystalline lattice structure, but electrons flowing through silicon have a smaller effective mass compared with MoS2. That is a boon when the gate is 5 nanometers or longer. But below that length, a quantum mechanical phenomenon called tunneling kicks in, and the gate barrier is no longer able to keep the electrons from barging through from the source to the drain terminals. This means we can’t turn off the transistors."
It has to scale cost-effectively to have billions of them though.
The roadmap all the manufacturers are planning for is 5nm, 3nm, 2nm, 1.4nm. There was an article from Samsung saying 1.4nm is as low as they can go.
I don't think commercially there will be much need for better than 1.4nm. This will be around 4.5x what we have at 7nm i.e 20TFLOP GPUs in laptops, that's close to half of a high-end Mac Pro in a laptop. When everybody is stuck at 1.4nm, it will come down to chip design and coprocessors to set performance apart.
Intel could possibly manufacture Apple's ARM chips once they can rival TSMC.
Nonsense. None of the other process shrinks lately have l8ved up to the initial numbers, and this won’t either.
A13 Bionic 7nm has 8.5 billion transistors in 98mm^2 = 86m transistors per square mm vs 91.2m listed. A12 Bionic 7nm has 6.9 billion in 83mm^2 = 83m/mm^2. A11 10nm has 4.3 billion in 87mm^2 = 49m/mm^2 vs 52.5m listed.
It depends on the chip design how tightly they fit the transistors in the products but from one generation to the next 83 vs 49 = 70% improvement in density. Here are some GPUs:
If they want to have better cooling then they can build a bigger chip so the density is lower. Mobile chips have to be small and Apple's chip density matches TSMC's numbers very closely.
This doesn't usually translate into the same amount of performance increase i.e 70% density doesn't mean 70% performance, sometimes they lower the power usage so 30% drop in power plus 30% increase in performance and they tend to split the improvements over 2 years.
These manufacturers invest billions of dollars into their plants, they have to come close to their estimates or their customers would use another manufacturer.
Your figures make my point. All TSMC needs to do is to be better than their competitors.
Although I’m rooting for Apple silicon to debut with 12 Cores, eight cores and 16 threads on the Intel MacBook Pro will be hard to beat.
In the MacBook Pro 16 at best we have an i9-9980HK with 45W TDP (PL2 is capped at 70W). To be able to maintain 45W TDP the CPU has to clock down to a whopping 2,4GHz all-core. In a thermally restricted environment, aka notebook, Intel doesn't stand a chance as they can't drive clock rates high enough to match the performance Apple's Cores can achieve at much lower clock rates and with that at a much lower power consumption.
I'm pretty sure the Mac Book Pro 13 with Apple Silicon SoC will be able to match the performance of the fastest MacBook Pro 16 with an Intel CPU.
Seems like 3nm might be a starting prerequisite if Apple plans on a massive Pro chip like some are suggesting.
I don't understand, why would Apple have to design one huge monolithic die? Two dies, 16+4 cores each, N5P node, uses TSMC's LIPinCon, and put together with CoWoS. There you have it – sales start june, 21st 2022....
The A13: 6 cores. The A14 is widely rumored to have 8. The MacBook chips won't be mobile chips like the A12/A13/A14. They are going to be desktop and server class chips with 12 cores.
We don't know the core breakdown ... but I think that having 8 efficiency cores would be ridiculous for a device that is primarily going to be plugged in (not a smartphone or tablet that is going to rely primarily on a charged battery). I think that it will be 6 performance/6 efficiency at minimum. Possibly even - my own personal hope - 8 performance and 4 efficiency, although that configuration would not be possible in a fanless device. (You COULD use the A14 in a fanless MacBook Air however).
OK, I was wrong. How wrong? Who knows. Ampere ARM server chips have 64-128 cores. But Apple is just getting started. Who knows how many cores their 3nm chips are going to have in 2022.
That being said ... the last time Qualcomm - and MediaTek - went past 8 cores (5 years ago) things went badly and they haven't tried since. (Samsung hasn't even tried to surpass 8 cores.) So Lenovo, Dell and the rest are going to be stuck with Intel and AMD no matter how badly Apple is going to boatrace them.
And anyone who buys an Intel-based Mac instead of a 12 core Apple Silicon Mac is out of their mind. If they want to run with clearly inferior tech they should just go ahead and buy a Lenovo or Dell Windows machine. Sorry but facts are facts.
1) Why are you only NOW onboard with the both the industry changes to ARM and Apple’s progressions over the past decade? None of this was out of left field. We don’t have specific details but Apple has left us with more than enough clues to know that this is going to be a major boon for Apple’s Mac business. What I don’t get us how anyone could ever think Apple was just going to put an iPhone SoC in a Mac and call it a day.
2) WinPC makers will also follow suit as MS ramps up support for Windows on ARM. While you can claim (in many other of your posts) how ineffectual Apple is in the market, the facts show that they light fires in the industry. Apple’s announcement will speed up the transition and adoption for Windows support and their vendor adoption… and this is a good thing.
3) This won’t be good for Intel, but it’s also not the end for Intel. There are many, many more years before anything Intel is doing now becomes truly obsolete. Even if they had no significant advantage for another decade they’d still be here. Transitions aren’t going to be as smooth as it will be with Apple. Just look at Windows poor shift to 64-bit. I worry more about AMD’s CPU market in the long term over Intel as AMD simply doesn’t have the revenue compared to Intel to stay competitive once the shift is in full effect.
AMD is eating Intel's lunch in sales overseas and OEMs are now all lining up to match their laptops for AMD against Intel line for line, which AMD bests. I know what Intel has and they're 18 months behind their targets for 10nm, never mind they're going to get sandbagged with Zen 3 and even more with Zen 4.
The A13: 6 cores. The A14 is widely rumored to have 8. The MacBook chips won't be mobile chips like the A12/A13/A14. They are going to be desktop and server class chips with 12 cores.
We don't know the core breakdown ... but I think that having 8 efficiency cores would be ridiculous for a device that is primarily going to be plugged in (not a smartphone or tablet that is going to rely primarily on a charged battery). I think that it will be 6 performance/6 efficiency at minimum. Possibly even - my own personal hope - 8 performance and 4 efficiency, although that configuration would not be possible in a fanless device. (You COULD use the A14 in a fanless MacBook Air however).
OK, I was wrong. How wrong? Who knows. Ampere ARM server chips have 64-128 cores. But Apple is just getting started. Who knows how many cores their 3nm chips are going to have in 2022.
That being said ... the last time Qualcomm - and MediaTek - went past 8 cores (5 years ago) things went badly and they haven't tried since. (Samsung hasn't even tried to surpass 8 cores.) So Lenovo, Dell and the rest are going to be stuck with Intel and AMD no matter how badly Apple is going to boatrace them.
And anyone who buys an Intel-based Mac instead of a 12 core Apple Silicon Mac is out of their mind. If they want to run with clearly inferior tech they should just go ahead and buy a Lenovo or Dell Windows machine. Sorry but facts are facts.
1) Why are you only NOW onboard with the both the industry changes to ARM and Apple’s progressions over the past decade? None of this was out of left field. We don’t have specific details but Apple has left us with more than enough clues to know that this is going to be a major boon for Apple’s Mac business. What I don’t get us how anyone could ever think Apple was just going to put an iPhone SoC in a Mac and call it a day.
2) WinPC makers will also follow suit as MS ramps up support for Windows on ARM. While you can claim (in many other of your posts) how ineffectual Apple is in the market, the facts show that they light fires in the industry. Apple’s announcement will speed up the transition and adoption for Windows support and their vendor adoption… and this is a good thing.
3) This won’t be good for Intel, but it’s also not the end for Intel. There are many, many more years before anything Intel is doing now becomes truly obsolete. Even if they had no significant advantage for another decade they’d still be here. Transitions aren’t going to be as smooth as it will be with Apple. Just look at Windows poor shift to 64-bit. I worry more about AMD’s CPU market in the long term over Intel as AMD simply doesn’t have the revenue compared to Intel to stay competitive once the shift is in full effect.
AMD is eating Intel's lunch in sales overseas and OEMs are now all lining up to match their laptops for AMD against Intel line for line, which AMD bests. I know what Intel has and they're 18 months behind their targets for 10nm, never mind they're going to get sandbagged with Zen 3 and even more with Zen 4.
AMD has a 1/4 of the profits of Intel for 2019 so I don’t understand how you can say Intel’s “lunch” is being eaten by AMD. By comparison, AMD is still the heavily malnourished one in the market which means they have the least amount of protection from a major market shift.
I hope (and expect) that both AMD and Intel will carve out reasonable niches as the market changes, but AMD isn’t in the ideal place to capitalize on the market that you want it to be. Despite your very long push of saying Apple should drop Intel for AMD that never happened and now they are dropping Intel for ASi. That should tell you something.
The A13: 6 cores. The A14 is widely rumored to have 8. The MacBook chips won't be mobile chips like the A12/A13/A14. They are going to be desktop and server class chips with 12 cores.
We don't know the core breakdown ...
A 12 core SOC, 8 high performance and 4 efficiency, has been rumored since April:
People seem to forget that AMD has been writing drivers for Metal and Metal 2 since their inception. They know Apple's accelerated graphics stack, Metal Frameworks, Core Image, CoreGraphics, etc. and know OpenCL better than Apple and has carried that torch to OpenCL 3.0.
Apple knows it isn't going to best AMD laptops but since they never ``used'' AMD laptop APUs they'll only compare themselves to Intel. They'll accept 90% performance of Intel and talk about the future, but ignore the fact they'll be 30% behind AMD before Zen 3 and will never leap frog them because they chose ARM as their future.
Seeing as they had to pay Samsung $950 Million for missing their contractual agreements with them I think they'd have been better off investing that into a growing relationship with AMD for High End and ARM for general Consumer.
Apple will never surpass AMD's roadmap and CPU/GPU performance designs, but they have never intended to do so. They just want to be competitive for Intel until some years later when ARM runs out of steam and they'll switch over again to x86 down the road.
What I learned at NeXT and what we brought to Apple was that the CPU architecture is hot swap capable if you design the OS correctly. We had HP PA-RISC, Sun SPARC, Motorola 68k and x86. We would have had DEC Alpha as well. We then added PowerPC and the latest is ARM.
PA-RISC died in 2005. SPARC is dead except for Fujitsu Server designs. Moto 68k is long since dead. Alpha is long since dead. PowerPC (Power ISA) is server and moving forward.
ARM and x86 are both moving forward. OS X works on both. Eventually, a successor must arise from one or both.
AMD Zen 4 design is nearly complete and will be by this year with samples in the Spring of 2021. AMD's next roadmap will also be revealed next Spring. Typically that means four or five more generations. RNDA 2.0 GPGPUs are about to arrive along with CDNA 1.0 for the Server Machine Learning/AI world.
Zen 4 is DDR5 and beyond, along with PCIe 5.0/6.0 [we know 5.0 minimally].
The A13: 6 cores. The A14 is widely rumored to have 8. The MacBook chips won't be mobile chips like the A12/A13/A14. They are going to be desktop and server class chips with 12 cores.
We don't know the core breakdown ...
A 12 core SOC, 8 high performance and 4 efficiency, has been rumored since April:
Apple knows it isn't going to best AMD laptops but since they never ``used'' AMD laptop APUs they'll only compare themselves to Intel. They'll accept 90% performance of Intel and talk about the future, but ignore the fact they'll be 30% behind AMD before Zen 3 and will never leap frog them because they chose ARM as their future.
Apple will never surpass AMD's roadmap and CPU/GPU performance designs, but they have never intended to do so. They just want to be competitive for Intel until some years later when ARM runs out of steam and they'll switch over again to x86 down the road.
Zen2 and RDNA2 will be 7nm. AMD won't be on 5nm until Zen4:
Apple's SoC will be 5nm this year. This gives them 15 billion transistors in an iPhone sized chip, which is close to the amount in a 12nm 2080ti (18b).
The entry Macbook Air can easily get a 4TFLOP GPU in 15W. AMD can't do that with RDNA2.
AMD's 5300M can run Shadow of the Tomb Raider at 1080p 60FPS, the Vega 20 can't, these are 16" Macbook Pro GPUs:
Apple ran that on an A12z with binary translation.
About switching back to x86 once their processes catch up, I don't think this is just about the process but the chip design and price. Apple is already building their own chips and shipping hundreds of millions per year. It's cheaper for them to do this than buy products designed and built by 3rd parties and they are in much more control of pricing, design and availability. Apple is currently paying a premium for worse chips than they make themselves. Switching to ARM for the Macs didn't need an excuse to do it, it needed an excuse not to and it looks like they ran out of excuses not to.
If they buy from 3rd parties, they have to patch together multiple chips from different suppliers and manage their graphics framework with the drivers for the hardware. If Intel decides all their low-end chips will have a 1TFLOP GPU, Apple is stuck with shipping inferior products to 70% of their Mac customers. AMD has better GPU performance across the board but they still have to deal with driver issues for hardware they don't control.
Apple can build better hardware than anyone, they've proved it for years now. It's cheaper, faster, makes it easier to develop and deploy across their entire ecosystem. There would have to be a really compelling reason to go back to x86 and being able to run Windows isn't it.
Apple will never surpass AMD's roadmap and CPU/GPU performance designs, but they have never intended to do so. They just want to be competitive for Intel until some years later when ARM runs out of steam and they'll switch over again to x86 down the road.
The A13: 6 cores. The A14 is widely rumored to have 8. The MacBook chips won't be mobile chips like the A12/A13/A14. They are going to be desktop and server class chips with 12 cores.
We don't know the core breakdown ...
A 12 core SOC, 8 high performance and 4 efficiency, has been rumored since April:
People seem to forget that AMD has been writing drivers for Metal and Metal 2 since their inception. They know Apple's accelerated graphics stack, Metal Frameworks, Core Image, CoreGraphics, etc. and know OpenCL better than Apple and has carried that torch to OpenCL 3.0.
Apple knows it isn't going to best AMD laptops but since they never ``used'' AMD laptop APUs they'll only compare themselves to Intel. They'll accept 90% performance of Intel and talk about the future, but ignore the fact they'll be 30% behind AMD before Zen 3 and will never leap frog them because they chose ARM as their future.
Seeing as they had to pay Samsung $950 Million for missing their contractual agreements with them I think they'd have been better off investing that into a growing relationship with AMD for High End and ARM for general Consumer.
Apple will never surpass AMD's roadmap and CPU/GPU performance designs, but they have never intended to do so. They just want to be competitive for Intel until some years later when ARM runs out of steam and they'll switch over again to x86 down the road.
What I learned at NeXT and what we brought to Apple was that the CPU architecture is hot swap capable if you design the OS correctly. We had HP PA-RISC, Sun SPARC, Motorola 68k and x86. We would have had DEC Alpha as well. We then added PowerPC and the latest is ARM.
PA-RISC died in 2005. SPARC is dead except for Fujitsu Server designs. Moto 68k is long since dead. Alpha is long since dead. PowerPC (Power ISA) is server and moving forward.
ARM and x86 are both moving forward. OS X works on both. Eventually, a successor must arise from one or both.
AMD Zen 4 design is nearly complete and will be by this year with samples in the Spring of 2021. AMD's next roadmap will also be revealed next Spring. Typically that means four or five more generations. RNDA 2.0 GPGPUs are about to arrive along with CDNA 1.0 for the Server Machine Learning/AI world.
Zen 4 is DDR5 and beyond, along with PCIe 5.0/6.0 [we know 5.0 minimally].
AMD’s Metal support has never been terribly good, and word was that Apple wasn’t particularly happy about it, but since Nvidia refused to support it in favor of their own technologies, Apple had no choice about what to use. Their own Metal support is much better. So now they have that choice.
Apple will ride ASI until something better proves itself. That may be a resurgent x86, it may not. Personally, I’m not sure how much further process reductions will take them. It might be incremental from here on out. Is that’s the case, those who don’t brew their own will end up on the bottom of the heap performance wise.
Comments
https://fuse.wikichip.org/news/3398/tsmc-details-5-nm/
https://www.phonearena.com/news/tsmc-3nm-chips-will-contain-nearly-300-million-transistors-per-square-mm_id123963
10nm density = 52.5 million transistors per square mm, 7nm = 91.2m, 5nm = 171.3m, 3nm = 291.2m.
Going from 10nm to 7nm was 73% increase (actual), 7nm to 5nm is 87% estimated. 5nm to 3nm is 70% estimated.
Given that they achieved 73% from 10nm to 7nm, the others are plausible estimates.
There's 1nm in a lab using a different material than Silicon:
https://newscenter.lbl.gov/2016/10/06/smallest-transistor-1-nm-gate/
"Both silicon and MoS2 have a crystalline lattice structure, but electrons flowing through silicon have a smaller effective mass compared with MoS2. That is a boon when the gate is 5 nanometers or longer. But below that length, a quantum mechanical phenomenon called tunneling kicks in, and the gate barrier is no longer able to keep the electrons from barging through from the source to the drain terminals. This means we can’t turn off the transistors."
It has to scale cost-effectively to have billions of them though.
The roadmap all the manufacturers are planning for is 5nm, 3nm, 2nm, 1.4nm. There was an article from Samsung saying 1.4nm is as low as they can go.
https://wccftech.com/intel-2021-2029-process-roadmap-10nm-7nm-5nm-3nm-2nm-1nm-back-porting/
I don't think commercially there will be much need for better than 1.4nm. This will be around 4.5x what we have at 7nm i.e 20TFLOP GPUs in laptops, that's close to half of a high-end Mac Pro in a laptop. When everybody is stuck at 1.4nm, it will come down to chip design and coprocessors to set performance apart.
Intel could possibly manufacture Apple's ARM chips once they can rival TSMC.
Huawei and TSMC have a very good relationship. The 3nm node was going to be used by Huawei, too.
Surely you don't think TSMC is happy losing one of its best historical customers? Or seeing them potentially go to its competitors, or worse, China accelerating its already stated goals in becoming self sufficient in chip design AND fabrication and reaching those goals sooner than expected (even if they still be a few years off).
Apple, Qualcomm, Nvidia, and AMD are more important than Huawei, hence why I posted that link indicating Huawei's share at 14%. More to the point, which you ignored, is that Taiwan does not want to support dual use, civilian and military, technology in the PRC, something that would be used by the PRC during an invasion. As Huawei is tightly linked to and controlled by the PRC and the CCP, there isn't any reason that TMSC should have maintained its relationship.
https://dailycaller.com/2020/07/19/bbc-andrew-marr-liu-xiaoming-uighur-prisoners-trains/
"“According to your own local government statistics, the population growth in Uighur jurisdictions in that area has fallen by 84 percent between 2015 and 2018,” Marr pressed. “84 percent!”
That's a human rights violation. Uighur prisoners is a human rights violation, just as taking Tibet was.
A12 Bionic 7nm has 6.9 billion in 83mm^2 = 83m/mm^2.
A11 10nm has 4.3 billion in 87mm^2 = 49m/mm^2 vs 52.5m listed.
https://en.wikipedia.org/wiki/Apple_A13
https://en.wikipedia.org/wiki/Apple_A12
https://en.wikipedia.org/wiki/Apple_A11
It depends on the chip design how tightly they fit the transistors in the products but from one generation to the next 83 vs 49 = 70% improvement in density. Here are some GPUs:
https://www.techpowerup.com/gpu-specs/radeon-rx-5600m.c3492 (7nm, 10.3b, 251mm^2 = 41m/mm^2)
https://www.techpowerup.com/gpu-specs/geforce-rtx-2080-ti.c3305 (12nm, 18.6b, 754mm^2 = 24m/mm^2)
https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3581 (7nm, 40b, 627mm^2 = 63m/mm^2)
If they want to have better cooling then they can build a bigger chip so the density is lower. Mobile chips have to be small and Apple's chip density matches TSMC's numbers very closely.
This doesn't usually translate into the same amount of performance increase i.e 70% density doesn't mean 70% performance, sometimes they lower the power usage so 30% drop in power plus 30% increase in performance and they tend to split the improvements over 2 years.
These manufacturers invest billions of dollars into their plants, they have to come close to their estimates or their customers would use another manufacturer.
I'm pretty sure the Mac Book Pro 13 with Apple Silicon SoC will be able to match the performance of the fastest MacBook Pro 16 with an Intel CPU. I don't understand, why would Apple have to design one huge monolithic die?
Two dies, 16+4 cores each, N5P node, uses TSMC's LIPinCon, and put together with CoWoS.
There you have it – sales start june, 21st 2022....
AMD is eating Intel's lunch in sales overseas and OEMs are now all lining up to match their laptops for AMD against Intel line for line, which AMD bests. I know what Intel has and they're 18 months behind their targets for 10nm, never mind they're going to get sandbagged with Zen 3 and even more with Zen 4.
I hope (and expect) that both AMD and Intel will carve out reasonable niches as the market changes, but AMD isn’t in the ideal place to capitalize on the market that you want it to be. Despite your very long push of saying Apple should drop Intel for AMD that never happened and now they are dropping Intel for ASi. That should tell you something.
What I learned at NeXT and what we brought to Apple was that the CPU architecture is hot swap capable if you design the OS correctly. We had HP PA-RISC, Sun SPARC, Motorola 68k and x86. We would have had DEC Alpha as well. We then added PowerPC and the latest is ARM.
Apple's SoC will be 5nm this year. This gives them 15 billion transistors in an iPhone sized chip, which is close to the amount in a 12nm 2080ti (18b).
The entry Macbook Air can easily get a 4TFLOP GPU in 15W. AMD can't do that with RDNA2.
AMD's 5300M can run Shadow of the Tomb Raider at 1080p 60FPS, the Vega 20 can't, these are 16" Macbook Pro GPUs:
https://www.notebookcheck.net/AMD-Radeon-RX-5300M-Laptop-GPU.436565.0.html
https://www.notebookcheck.net/AMD-Radeon-Pro-Vega-20-GPU-Graphics-Card.361941.0.html
Apple ran that on an A12z with binary translation.
About switching back to x86 once their processes catch up, I don't think this is just about the process but the chip design and price. Apple is already building their own chips and shipping hundreds of millions per year. It's cheaper for them to do this than buy products designed and built by 3rd parties and they are in much more control of pricing, design and availability. Apple is currently paying a premium for worse chips than they make themselves. Switching to ARM for the Macs didn't need an excuse to do it, it needed an excuse not to and it looks like they ran out of excuses not to.
If they buy from 3rd parties, they have to patch together multiple chips from different suppliers and manage their graphics framework with the drivers for the hardware. If Intel decides all their low-end chips will have a 1TFLOP GPU, Apple is stuck with shipping inferior products to 70% of their Mac customers. AMD has better GPU performance across the board but they still have to deal with driver issues for hardware they don't control.
Apple can build better hardware than anyone, they've proved it for years now. It's cheaper, faster, makes it easier to develop and deploy across their entire ecosystem. There would have to be a really compelling reason to go back to x86 and being able to run Windows isn't it.
of course, I’ve been wrong before...