Intel's deal with TSMC is only short-term: 2022-2024. In 2025, Intel will start producing chips on their 5nm process and will no longer need TSMC to compete with AMD. (Despite what Apple fans believe, Intel's competition is AMD, not Apple.) Yes, the primary benefit of Intel buying capacity from TSMC is that it will further exacerbate AMD's supply problems. Which leaves poor AMD between a rock and a hard place: capacity problems at TSMC and yield problems at the only viable alternative Samsung.
If Intel hold to past behavior they will need TSMC well past 2025: Intel's 7nm is Broken, Company Announces Delay Until 2022, 2023; Intel's first 10nm desktop CPUs are still a year away…oh and 7nm is delayed (again). Intel is great at making road maps but really lousy at following them schedule wide
Intel has never kept up on a schedule they were just the only game in town after Motorola and IBM gave up because they couldn’t see a future world of portable powerful cpu’s which were also energy efficient at the same time. And now Intel (barn burners) has no in house OS to go with a in house cpu, they are disrupted at every level by Apple (SOC) and are dead men walking.
Intel's deal with TSMC is only short-term: 2022-2024. In 2025, Intel will start producing chips on their 5nm process and will no longer need TSMC to compete with AMD. (Despite what Apple fans believe, Intel's competition is AMD, not Apple.) Yes, the primary benefit of Intel buying capacity from TSMC is that it will further exacerbate AMD's supply problems. Which leaves poor AMD between a rock and a hard place: capacity problems at TSMC and yield problems at the only viable alternative Samsung.
If Intel hold to past behavior they will need TSMC well past 2025: Intel's 7nm is Broken, Company Announces Delay Until 2022, 2023; Intel's first 10nm desktop CPUs are still a year away…oh and 7nm is delayed (again). Intel is great at making road maps but really lousy at following them schedule wide
Intel has never kept up on a schedule they were just the only game in town after Motorola and IBM gave up because they couldn’t see a future world of portable powerful cpu’s which were also energy efficient at the same time. And now Intel (barn burners) has no in house OS to go with a in house cpu, they are disrupted at every level by Apple (SOC) and are dead men walking.
Huh?
I suspect that IBM gave up because they couldn't see a lot of profit in it. And, they were right: PCs transitioned from high end devices to commodities: My first PC cost over $9,000. My latest one a little over $200 (i7, 250gb SSD & 16Gb RAM) -- even though it has many times the power of the first one. The last few years we've seen a resurgence of high end laptops, but that will fade.
And, except for Apple, hardware and software have always been separate. Combining & integrating the two has been a source of strength for Apple. But it isn't the norm.
Intel is not "in peril" in any sense. Quite the contrary, Intel is #1 by a mile. AMD is #2 even if only because being forced to share TSMC with Apple, Qualcomm, MediaTek and Nvidia means that they can't come close to matching Intel's volume. (AMD is FINALLY shifting manufacture of their low end chips to Samsung later this year.) Apple is a distant #3. ARM CPU makers like Qualcomm, MediaTek and Samsung are going to be held back by the limitations of Windows on ARM (though there is some potential with Chrome OS ARM, which works a lot better). Indeed, now that Intel is re-entering the discrete GPU market, they are going to sell more chips in 2022 without Apple than they ever did with them.
Hello Intel employee.
I worked for Nokia around 2008/9. At the time, they were selling hundreds of millions of smartphones a year, outselling Apple by an order of magnitude. Comfortably the number #1. But the writing was on the wall. Everyone internally knew it.
I doubt Intel will ever crash and burn like Nokia but it's only a matter of time until Microsoft crack ARM on Windows. Once that happens, no-one is going to pay Intel's inflated prices for CPUs.
Then Intel will simply get a license and start making ARM based processors. Easy-Peasy. No problem.
Big problem - the majority of code for PC desktops is X86. Without a good translator (rather than emulator) Intel will have the same problem Microsoft had with Windows on ARM did - little adoption. PC user like to hold on to their machines until way past their use by date (many times due ironically to cost). Why do you think there are still a good number of ATMs running Windows XP? Apple can force its users on to ARM - Intel cannot without some sort of software help (a hardware option would defeat the whole idea of going to ARM).
I think you missed to read the context of George's response - @xbit mentioned that Microsoft will take care of getting windows fully ready for ARM processors eventually (say in 2-3 years, not explicitly mentioned, but implied). George's response should be read in that context. The question of translator/emulator does not arise in that scenario.
Not that easy for Microsoft. There are hundreds or thousands accessories that were designed to work with Microsoft Windows. They will become obsolete with ARM based Windows OS.
As Apple shows if you come up with a good enough translator the conversion from x86 can be relatively smooth...provided the programs are reasonably well written and there in is the problem. Windows users are notorious for using old code (people wanting to run 16-bit code on Windows 10) while Apple having a closed architecture could just cut off old antiquated code off at the kneecaps (poorly written 24/32 code and later 32-bit code). That need to back support into insanity is why Windows for ARM has been languishing since 2017 - it was limited to just 32-bit x86 code and did it in emulation not translation. The cherry on this little FUBAR cake was most games (the main reason people get windows) are written in 64-bit and many professional programs were moving to 64-bit as well.
Intel is not "in peril" in any sense. Quite the contrary, Intel is #1 by a mile. AMD is #2 even if only because being forced to share TSMC with Apple, Qualcomm, MediaTek and Nvidia means that they can't come close to matching Intel's volume. (AMD is FINALLY shifting manufacture of their low end chips to Samsung later this year.) Apple is a distant #3. ARM CPU makers like Qualcomm, MediaTek and Samsung are going to be held back by the limitations of Windows on ARM (though there is some potential with Chrome OS ARM, which works a lot better). Indeed, now that Intel is re-entering the discrete GPU market, they are going to sell more chips in 2022 without Apple than they ever did with them.
Hello Intel employee.
I worked for Nokia around 2008/9. At the time, they were selling hundreds of millions of smartphones a year, outselling Apple by an order of magnitude. Comfortably the number #1. But the writing was on the wall. Everyone internally knew it.
I doubt Intel will ever crash and burn like Nokia but it's only a matter of time until Microsoft crack ARM on Windows. Once that happens, no-one is going to pay Intel's inflated prices for CPUs.
Then Intel will simply get a license and start making ARM based processors. Easy-Peasy. No problem.
Big problem - the majority of code for PC desktops is X86. Without a good translator (rather than emulator) Intel will have the same problem Microsoft had with Windows on ARM did - little adoption. PC user like to hold on to their machines until way past their use by date (many times due ironically to cost). Why do you think there are still a good number of ATMs running Windows XP? Apple can force its users on to ARM - Intel cannot without some sort of software help (a hardware option would defeat the whole idea of going to ARM).
I think you missed to read the context of George's response - @xbit mentioned that Microsoft will take care of getting windows fully ready for ARM processors eventually (say in 2-3 years, not explicitly mentioned, but implied). George's response should be read in that context. The question of translator/emulator does not arise in that scenario.
Not that easy for Microsoft. There are hundreds or thousands accessories that were designed to work with Microsoft Windows. They will become obsolete with ARM based Windows OS.
As Apple shows if you come up with a good enough translator the conversion from x86 can be relatively smooth...provided the programs are reasonably well written and there in is the problem. Windows users are notorious for using old code (people wanting to run 16-bit code on Windows 10) while Apple having a closed architecture could just cut off old antiquated code off at the kneecaps (poorly written 24/32 code and later 32-bit code). That need to back support into insanity is why Windows for ARM has been languishing since 2017 - it was limited to just 32-bit x86 code and did it in emulation not translation. The cherry on this little FUBAR cake was most games (the main reason people get windows) are written in 64-bit and many professional programs were moving to 64-bit as well.
Good point!
I inherited a mission critical system (running the transactions from ATMs) that had been converted using an automated translator shortly after it was put into production. To say it was a nightmare would be a gross understatement. And that was relatively clean code. I'm sure much of the crap running on Windows makes normal crap look pretty good.
I give Microsoft a LOT of credit for supporting legacy software. Companies have a lot invested in that old software and often it is difficult to upgrade -- so it just chugs along.
But, they might have to make a clean break adding ARM support. And, that would be OK because the only ones needing it would be those running state of the art equipment -- and they probably have the resources to upgrade their software.
Intel is not "in peril" in any sense. Quite the contrary, Intel is #1 by a mile. AMD is #2 even if only because being forced to share TSMC with Apple, Qualcomm, MediaTek and Nvidia means that they can't come close to matching Intel's volume. (AMD is FINALLY shifting manufacture of their low end chips to Samsung later this year.) Apple is a distant #3. ARM CPU makers like Qualcomm, MediaTek and Samsung are going to be held back by the limitations of Windows on ARM (though there is some potential with Chrome OS ARM, which works a lot better). Indeed, now that Intel is re-entering the discrete GPU market, they are going to sell more chips in 2022 without Apple than they ever did with them.
Hello Intel employee.
I worked for Nokia around 2008/9. At the time, they were selling hundreds of millions of smartphones a year, outselling Apple by an order of magnitude. Comfortably the number #1. But the writing was on the wall. Everyone internally knew it.
I doubt Intel will ever crash and burn like Nokia but it's only a matter of time until Microsoft crack ARM on Windows. Once that happens, no-one is going to pay Intel's inflated prices for CPUs.
Then Intel will simply get a license and start making ARM based processors. Easy-Peasy. No problem.
Big problem - the majority of code for PC desktops is X86. Without a good translator (rather than emulator) Intel will have the same problem Microsoft had with Windows on ARM did - little adoption. PC user like to hold on to their machines until way past their use by date (many times due ironically to cost). Why do you think there are still a good number of ATMs running Windows XP? Apple can force its users on to ARM - Intel cannot without some sort of software help (a hardware option would defeat the whole idea of going to ARM).
I think you missed to read the context of George's response - @xbit mentioned that Microsoft will take care of getting windows fully ready for ARM processors eventually (say in 2-3 years, not explicitly mentioned, but implied). George's response should be read in that context. The question of translator/emulator does not arise in that scenario.
Not that easy for Microsoft. There are hundreds or thousands accessories that were designed to work with Microsoft Windows. They will become obsolete with ARM based Windows OS.
As Apple shows if you come up with a good enough translator the conversion from x86 can be relatively smooth...provided the programs are reasonably well written and there in is the problem. Windows users are notorious for using old code (people wanting to run 16-bit code on Windows 10) while Apple having a closed architecture could just cut off old antiquated code off at the kneecaps (poorly written 24/32 code and later 32-bit code). That need to back support into insanity is why Windows for ARM has been languishing since 2017 - it was limited to just 32-bit x86 code and did it in emulation not translation. The cherry on this little FUBAR cake was most games (the main reason people get windows) are written in 64-bit and many professional programs were moving to 64-bit as well.
Good point!
I inherited a mission critical system (running the transactions from ATMs) that had been converted using an automated translator shortly after it was put into production. To say it was a nightmare would be a gross understatement. And that was relatively clean code. I'm sure much of the crap running on Windows makes normal crap look pretty good.
I give Microsoft a LOT of credit for supporting legacy software. Companies have a lot invested in that old software and often it is difficult to upgrade -- so it just chugs along.
But, they might have to make a clean break adding ARM support. And, that would be OK because the only ones needing it would be those running state of the art equipment -- and they probably have the resources to upgrade their software.
Why would they switch? If it is just a software upgrade, they can afford to make another investment on same software. But if its an accessory, they have to wait for the accessory company. Why would the accessory company put resources to redevelop? Most accessories don't sell as much as Microsoft sells its bloated software. Only a minute fraction.
Intel is not "in peril" in any sense. Quite the contrary, Intel is #1 by a mile. AMD is #2 even if only because being forced to share TSMC with Apple, Qualcomm, MediaTek and Nvidia means that they can't come close to matching Intel's volume. (AMD is FINALLY shifting manufacture of their low end chips to Samsung later this year.) Apple is a distant #3. ARM CPU makers like Qualcomm, MediaTek and Samsung are going to be held back by the limitations of Windows on ARM (though there is some potential with Chrome OS ARM, which works a lot better). Indeed, now that Intel is re-entering the discrete GPU market, they are going to sell more chips in 2022 without Apple than they ever did with them.
Hello Intel employee.
I worked for Nokia around 2008/9. At the time, they were selling hundreds of millions of smartphones a year, outselling Apple by an order of magnitude. Comfortably the number #1. But the writing was on the wall. Everyone internally knew it.
I doubt Intel will ever crash and burn like Nokia but it's only a matter of time until Microsoft crack ARM on Windows. Once that happens, no-one is going to pay Intel's inflated prices for CPUs.
Then Intel will simply get a license and start making ARM based processors. Easy-Peasy. No problem.
Big problem - the majority of code for PC desktops is X86. Without a good translator (rather than emulator) Intel will have the same problem Microsoft had with Windows on ARM did - little adoption. PC user like to hold on to their machines until way past their use by date (many times due ironically to cost). Why do you think there are still a good number of ATMs running Windows XP? Apple can force its users on to ARM - Intel cannot without some sort of software help (a hardware option would defeat the whole idea of going to ARM).
I think you missed to read the context of George's response - @xbit mentioned that Microsoft will take care of getting windows fully ready for ARM processors eventually (say in 2-3 years, not explicitly mentioned, but implied). George's response should be read in that context. The question of translator/emulator does not arise in that scenario.
Not that easy for Microsoft. There are hundreds or thousands accessories that were designed to work with Microsoft Windows. They will become obsolete with ARM based Windows OS.
As Apple shows if you come up with a good enough translator the conversion from x86 can be relatively smooth...provided the programs are reasonably well written and there in is the problem. Windows users are notorious for using old code (people wanting to run 16-bit code on Windows 10) while Apple having a closed architecture could just cut off old antiquated code off at the kneecaps (poorly written 24/32 code and later 32-bit code). That need to back support into insanity is why Windows for ARM has been languishing since 2017 - it was limited to just 32-bit x86 code and did it in emulation not translation. The cherry on this little FUBAR cake was most games (the main reason people get windows) are written in 64-bit and many professional programs were moving to 64-bit as well.
Good point!
I inherited a mission critical system (running the transactions from ATMs) that had been converted using an automated translator shortly after it was put into production. To say it was a nightmare would be a gross understatement. And that was relatively clean code. I'm sure much of the crap running on Windows makes normal crap look pretty good.
I give Microsoft a LOT of credit for supporting legacy software. Companies have a lot invested in that old software and often it is difficult to upgrade -- so it just chugs along.
But, they might have to make a clean break adding ARM support. And, that would be OK because the only ones needing it would be those running state of the art equipment -- and they probably have the resources to upgrade their software.
Why would they switch? If it is just a software upgrade, they can afford to make another investment on same software. But if its an accessory, they have to wait for the accessory company. Why would the accessory company put resources to redevelop? Most accessories don't sell as much as Microsoft sells its bloated software. Only a minute fraction.
Enterprise software can be very complex and very expensive. And even without the expense, to get to new software, they need the staff to design, develop, implement and support it. Software that's been chugging along for a decade or more may only have a support programmer keeping it running.
Upgrading the hardware could easily be the cheapest and easiest part.
Comments